The present invention relates to a data input system for inputting
data to a database.
It is now becoming increasingly common for an operating system, such
as an industrial plant, to be operated on the basis of an "expert system" which
operates on the basis of data stored in a database. It is generally accepted that
a database used in an expert system is part of a "growing system", because the
data is constantly being increased or changed, by actual operation of the system
by the user, so that no final completion is ever achieved.
In such a system, the data represents relationships, and in this
specification the term data (or information) refers to an operating relationship
between operating parameters of the operating system. Those parameters may be
operating conditions, process variables, or fixed values.
Some of that data may be initially input to the system (stored in
the database) by the manufacturer of the system and therefore the validity of the
data may be determined before the system is in operation. However, when the system
is in operation, further data must be manually entered into the database by an
operator and it is therefore necessary to ensure the integrety and validity of
such data. Furthermore, differences may develop between the data initially stored
in the database and the data that is actually used after, perahps, years of operation
of the system. These factors can result in invalid data within the system, which
is to be avoided.
When considering these two factors, various arrangements have been
proposed to try to overcome them. In Japanese Patent Application No. 60-41128,
the data input by several operators is first passed to a database administrator,
who is trained to determine which can data can be stored in the database. This
system, of course, requires that the database administrator be fully aware of
the correct operating relationships, and the level of training needed for this
is a limit to the practicality of this proposal.
The problem of the development of invalid data over a period of time
was considered by Japanese Patent Application No. 58-192161, in which "age" information
is associated with the data being input, and the data is automatically deleted
after expiry of a predetermined age limit. This is also not a satisfactory solution,
since valid data, which may subsequently be useful, will be deleted if it has
reached its age limit.
Thus, neither of these two proposals adequately face the problem
of input of invalid data, and this provides a limit to the practicality of the
use of databases in such expert systems.
Another relevant prior art is FUTURE GENERATION SYSTEMS. vol. 3,
no. 3, September 1987, AMSTERDAM, NL, pages 201-215; P.S. SHEU AND W.S. LEE 'Efficient
Processing of Integrity Constraints in Deductive Databases'. In this document it
is described how the process of validating integrity constraints can be made efficient
when a database is changed.
The present invention proposes that the data input system stores
therein "fundamental rules" which are compared with the data being input. The invention
is defined by the appended claims. Each of those fundamental rules is a predetermined
relationship between the operating parameters of the operating system, and thus
it becomes possible to check, for any given datum, which of the fundamental rules
corresponds to those operating parameters, and which of such fundamental rules
are satisfied by the data being input. In this way, it becomes possible to determine
if the data is valid and the datum can be transmitted to the database when every
corresponding fundamental rule is satisfied, or transmission can be prevented
if any corresponding fundamental rule is not satisfied.
Thus, many fundamental rules between the operating parameters and
the operating system may be stored in a suitable memory, and when a new datum is
input, the first step is to determine which operating parameters are used in the
operating relationship of the datum, and to extract all the fundamental rules defining
relationships between those operating parameters. Then, the datum can be compared
with those fundamental rules to see if it is valid.
If those fundamental rules are satisfied, the datum is valid and
similarly if any is not satisfied, the datum is invalid.
It can occur, however, that for given datum, there are no corresponding
fundamental rules (because the relationship between the operating parameters required
by the datum have not been predefined). In this case, the system cannot directly
determine the validity or invalidity of the datum, and then it may be necessary
for a skilled operator to determine whether that datum should be input or not
(such a datum being hereinafter referred to as a "reserved" datum). To ensure that
only a sufficiently trained operator can cause a reserved datum to be input to
the database, the input of the system may associate a plurality of priorities
to the users of the system, and permit transmission of a reserve datum only when
the user has a predetermined priority (e.g. corresponding to a trained operator).
Furthermore, the system may have a management history table which
records the history of input of the data, so that if subsequently a given datum
or data is found to be invalid, it becomes possible for that data to be extracted
by determining when the datum or data was input in terms of the management history
of the data. Thus, any incorrect data can be easily updated or deleted on the basis
of the history of the database.
The present invention is applicable to any database in which the
data represents operating relationships. It is particularly concerned with a data
input system for operating an industrial plant, but may also be applicable to
other situations such as medical databases.
An embodiment of the invention will now be described in detail, by
way of example, with reference to the accompanying drawings in which :
- Figure 1 shows a block diagram of a data input system being an embodiment of
the present invention;
- Figure 2 shows a detail of the hierarchy of the database;
- Figure 3 shows various flow processes;
- Figure 4 shows the relationship between two processed items;
- Figure 5 shows the influence of process data;
- Figure 6 shows a table illustrating operation states;
- Figure 7 shows a block diagram of a thermal power plant which may be operated
on the basis of a system according to the present invention;
- Figure 8 shows examples of fundamental rules;
- Figure 9 shows the organisation of fundamental rules;
- Figure 10 shows a management history table;
- Figure 11 shows the matching-enable conditions for main and sub data bases;
- Figure 12 shows how history from the history management table is displayed
on a screen;
- Figure 13 shows schematically how access to the database is controlled; and
- Figure 14 shows a form for registration of data.
Figure 1 shows the schematic configuration of a data input/control
system of an informational database according to an embodiment of the present invention.
In Figure 1, a data input/output (I/O) device 21 is connected via
a user interface 22 to a database 26. The connection between database I/O device
21 and the user interface is illustrated by buses 6, 8, and the connection between
the user interface 22 and the database 26 is illustrated by buses 1, 7. The user
interface 22 is conected via buses 9, 10 to a management history table 27 which
records history of data input through the system by I/O device 21. Indeed, as will
be described later, it is also possible to display history information on a display
of the I/O device 21 from the management history table 27 when desired.
Figure 1 also illustrates that the user interface 22 is connected
to a fundamental rule memory 24, which stores the relationships (fundamental rules)
which are predetermined between the operating parameters of the operating system.
When data is input from the I/O device 21, the fundamental rules relating to the
corresponding operating parameters are extracted from the fundamental rule memory
24 via the data edit processing device 23, and transmitted via a conflict/error
selecting device 25 and hence to the input/output device. The data edit processing
device 23 compares the data with the corresponding fundamental rules to determine
whether the data conforms to the relationship established within the fundamental
rules, and conflicting or erronious data can be extracted by the selecting device
25. Then, the user only stores the data whose validity has been confirmed in the
database 26 by suitable operation of the I/O device 21.
The operation of the system of Figure 1 will be described in more
detail later. Before doing that, however, the detailed structure and effects of
the fundamental rule memory 24, the database 26, and the management history table
27 will be discussed.
1. Fundamental Rule Memory 24
The structure and concept of the fundamental rule section 24 will
be described using a thermal power plant as an example of an object system operated
by an expert system utilizing data base 26.
Figure 7 shows the typical configuration of a thermal power plant.
As shown in Fig. 7 steam generated in a boiler 201 is sent to a high-pressure turbine
202 through a main steam pipe 218. Some of the thermal energy of the steam is
converted into mechanical energy to drive a generator 204. The steam that operates
high-pressure turbine 202 passes through a low-temperature reheated steam pipe
219, is heated again in reheater 216 then is sent to reheated turbine 203 through
a high-temperature reheated steam pipe 220 for subsequent operation. The steam
that operates reheated turbine 203 is sent (in steam form) to a condenser 205,
is cooled by cooling water (such as sea water), then condenses into water. The
condensate is pumped up by condensate pump 206, which performs thermal recovery
through heat exchangers such as condensate heat exchanger 207, air extracting device
208, and gland condenser 209. The temperature of the condensate is raised by a
low-pressure feedwater-heating device 210 and deareating device 211. The temperature
of the feedwater (whose pressure is increased in boiler feedwater pump 212) is
further raised in a high pressure feedwater-heating device 213. This feedwater
is then supplied to the boiler 201 through a main feedwater pipe 221. The feedwater
in the high-pressure feedwater-heating device 213, deaerator 211, and low-pressure
feedwater-heating device 210 is heated by the bled air of the turbine. In the boiler,
the fuel is controlled by a fuel adjusting valve 217 as it passes through a fuel
burner 214, then the necessary amount of air is added to the fuel for combustion
in a furnace.
The feedwater receives the heat radiated by this combustion, converts
it into steam, which is superheated in superheating device 215 and sent to the
turbine. The fundamental rules previously prepared and stored in fundamental rule
section 24 will be described in respect of the causal relations established between
the various process data on the above-described thermal power plant (object system).
Section 1-1 below discusses the extraction of the causal relations established
in the object system; Section 1-2 describes the concept for extracting all causal
relationships. Note that the process data used in extracting the causal relationships
may be in analog or digital form.
1-1. Extracting causal relationships.
A) Note the directions in which the liquid and heat flow in this process
(see Fig. 3).
This directional flow is necessary in such processes. The flow of
such liquids as water, condensed steam, oil, and gas and the flow of heat and electricity
are considered normal in plant operation, especially when the relation accompanying
the flow is specified. The rules relating to flow rate, pressure, temperature,
and heat are described below. Figure 3 (a) shows a piping system of the object.
The high-temperature, high-pressure fluid source is shown on the left of the figure
and a branched fluid consumption part is shown on the right. When considering such
a simplified piping system, the following relationship exists for the flow rate,
pressure, and temperature at each point I - IX.
Flow Rate: The total inflow discharge and outflow discharge at a system branch
point is always equal. As shown in Fig. 3 (b), the following relation always exists
in each flow rate (Qa to Qe :Qa = Qb
+ Qc, Qc = Qd + Qe), and if the total
outflow discharge is smaller than the inflow discharge, an abnormality that causes
a leak exists on the upstream side of the discharge flow rate detecting point
(behind the branched point) is detected. If the inflow flow rate is smaller, use
of abnormal mixtures of other liquids from an external system can be detected.
Pressure: The pressure detection value on the upstream side due to the pressure
loss attirbuted to the fluid flow speed (restricted to systems that do not use
pumping action) is always larger than the pressure detection value on the downstream
side. As shown in Fig. 3(c), the following relationships exist:
P1 ≥ P2 ≥ P3 (P'3), P3
≥ P4 ≥ P6.
If these relationships did not exist, the direction in which liquids flow would
be abnormal. Thus, the flow rate must be checked, and information on the anomaly
must be prepared.
Temperature: Liquid is needed for heat exchange in the flow process with an external
system. In case of low-temperature liquid, heat flows in from an external system.
For high-temperature liquid, heat is discharged to the external system. Figure
3(d) shows the latter case. The following relationships exist:
T1 ≥ T2 ≥ T3 (T'3), T3
≥ T4 ≥ T6.
Heat: Even if the liquid used as a thermal medium did not exist, any temperature
difference causes heat to be transferred thereby increasing the entropy.
Monitoring the thermal stress caused by temperature difference between
the interior and exterior walls is important in plant equipment that uses thick
metal materials. As shown in Fig. 3(e), the transfer of heat can be estimated
under individual operation conditions, and under certain conditions, the relationships
Ta>Tb, Tc>Td, or Ta
- Tb < set value, Tc - Td < set value exist.
If this relationship did not exist, abnormal information would be generated.
B) Consider the correlation between process data (See Fig. 4).
Process data is essential for managing and controlling the final
process quantity of plant production, and attention must be given to the important
quantities in such correlations as the primary production process quantity for
the amount of materials supplied, secondary production process quantity, specific
parameters, and final process quantity. As shown in Fig. 4(a) - (c), because two
process items have such correlations as linear, disproportional, and saturation
characteristics, dependent variable side process data Y (corresponding to the measurement
value of independent variable side process data X) is indicated in the table by
(d) with the correlation equation (when possible).
C) Consider the influence exertion route of the process data (Fig. 5).
When considering the process data, we must assume that there is other
data related to the first process data. The assumed data should consist of the
data group that forms the source of effecting the process data, and the data group
that forms the destination of the effects as shown in Fig. 5.
D) Consider the operation terminal state (Fig. 6).
A close relationship exists between the operation state of a plant
and each operation terminal, particularly when the plant is in an anomaly state.
Therefore, attention must be paid to the operation state of each operation terminal
in each normal/anomaly state, Each operation state is specified in a specific fundamental
rule or in group. Figure 6 shows the case wherein an individual operation terminal
(valve A only) is checked, and the case where two operation terminals (valves A
and B) are checked.
1-2 Concept for extracting all causal relationships.
Though several concepts for extracting the fundamental rule of an
object systen are shown, it is difficult to extract all fundamental rules efficiently,
particularly for a large system. The following describes the concept for efficiently
extracting all fundamental rules by using the above-described thermal power plant
as an example.
When the above-described thermal power plant is divided in the production
process as an energy carrier, the overall operation can be expressed as four main
processes. In other words, the main processes consist of a air-gas-fuel system
41, a water-steam system 42, a turbine-electricity system 44, and a cooling water
system 443, When such divided processes are allowed to correspond to the equipment
composition shown in Fig. 7, the range enclosed by the respective dotted lines
in Fig. 7 can be obtained. From the standpoint of the equipment composition, the
water-steam system 42 includes a heat-exchange function achieved by the boiler
201 and a heat-consumption function achieved by the turbine 202. The main process
in which the flow of substances (as an energy carrier) can be classified into several
subprocesses which determine the flow. For the water-steam system 42, the equipment
is divided into a low-pressure heater group, high-pressure heater group, and a
boiler turbine. Further, the subprocess includes a group of equipment used to
perform a charged work. This group is considered a uni-process. In the high-pressure
heater group, an individual high-pressure heater is also considered a uni-process.
Therefore, the entire power generation plant can be understood from one side surface
by dividing the process into equipment operation-terminal levels at the plant.
Moreover, the equipment for the high-pressure heater includes a high-pressure
heater, outlet valve, inlet valve, bypass valve, and bleed valve.
Therefore, all fundamental rules can be efficiently extracted by
precisely classifying the entire thermal power plant into a main process, subprocesses,
uni-processes, and equipment, and by successively extracting the causal relationships
between a precisely classified level or levels in terms of major and subordinate
Figure 8 shows an example of a fundamental rule extracted in respecct
of the high-pressure heater group of the water-steam system, the correspondence
between the divided system processes, an example of the extracted fundamental
rule, and the concepts (A-D) for extracting the rules described in Figs. 3 to 6.
Example 1: Fundamental rule of equipment level (intrinsic for equipment)
This is related to the state of operation terminal in fundamental
rule extraction group D, and the specified condition that a heater bleed valve
is always in the correctly opened state during normal plant operation.
Example 2: Fundamental rule (between equipment)
This is related to the state of operation terminal in fundamental
rule extraction group D, and the specified condition that the opening/closing states
of the heater's outlet valve and inlet valve are the same to establish the correlation
between the equipment used in one high-pressure heater set. When a time difference
(that can be ignored in the opening/closing states of both valves) exists, processing
to eliminate such time difference is executed.
Example 3: Fundamental rule of a single process level (in a single process)
This is related to the state of operation terminal in fundamental
rule extraction group D, further related to the flow direction of process A, with
the condition that the inlet and outlet valves are correctly opened and a bypass
valve is correctly closed during normal plant operation, or the inlet and outlet
valves are correctly closed, and the bypass valve is correctly opened for the anticiapted
abnormal plant operation, i.e. a heater bypass operation is specified. In the
latter case, the specified condition is that the feedwater flow rate at the heater
outlet is equal to the feedwater flow rate at the heater inlet (at a constant
water level in the heater).
Example 4: Fundamental rule of single process level (between single processes)
This is related to the flow direction of the process in fundamental
rule extraction group A, and the specified condition that the feed water temperature
at the heater's outlet is lower than the feedwater tamperature at the heater's
outlet in the next stage (slip stream side) and with the specified condition that
the feedwater pressure at the heater's outlet is higher than the feedwater pressure
at the heater's outlet in the next stage. These rules are based on the heat exchange
and flow passage pressure loss in the feedwater line system near the high-pressure
Example 5: Fundamental rule of subprocess level (in subprocess)
This is related to the flow directin of the process in fundamental
rule extraction group A, and the specified condition that the sum of the bleed
flow rate of a high-pressure heater group is equal to the drain flow rate of the
final stage heater with a constant water level in the heater. This is also related
to the correlation between the process data in fundamental rule extraction group
B, and the specified condition that the total rising value of feedwater temperature
in each high-pressure heater exceeds a certain set value, and the fact that total
heat exchange in the high-pressure heater group is correct when considering the
heat balance from one side surface.
Example 6: Fundamental rule of subprocess level (between subprocesses)
Nothing is specifically specified.
Example 7: Fundamental rule of main process level (in main process)
This is related to the correlation between the process data on fundamental
rule extraction group B, and the specified condition that a linear relation exists
between the feedwater flow rate at the outlet of the high-pressure heater and
the main steam flow rate, and that the relation between water and steam in the
water/steam process is monitored in terms of plant performance.
Example 8: Fundamental rule of main process level (between main processes)
This is related to the correlation between the process data on fundamental
rule extraction group C, and the specified condition that power generator output
in a plant that permits heater cut operation may increase by 5% during operation.
This means that in a different process, the correlation between the main data on
the plant is changed under certain operation conditions.
Though several kinds of examples related to high-pressure heaters
are listed, it is not necessary to specify the rule in which the difference between
different divided processes and that between similar processes is not distinct
Each fundamental rule extracted in the above-described procedure
is organized in a summary table shown in Fig. 9. These rules are stored in the
fundamental rule section shown in Fig. 1. Figure 9 shows part of the fundamental
rule extracted as shown in Fig. 8, and the correlation between the process data
and related process data is recorded (in the form of a definition table) as a set
of information. For example, rule R1 in the figure is the described in '1' in
Fig. 8. This content describes high-pressure heater bleed flow rate I2A master
trip delay set signal I4D alarm reset signal I5D, and generator output signal
I3A (which are confirmed as process data related to heater bleed valve opening
degree signal I, and are abnormal process data to be monitored when I2A or I3A
is larger than each prescribed value. When I4D and I5D are in the reset state,
normal operation is determined. Rules R2 and R3 are the contents of steps 7 and
3-3 in Fig. 8.
Both are two independent rules determined in relation to the high-pressure
heater outlet feedwater flow rate I4'A. In rule R2, this rate is described
as main steam flow rate I5A, while main steam pressure I6A and main steam temperature
I7A are the related process data to be monitored. The following relationship exists
in this case: |I4A - I5A x I6A x I7A| ≤ C4A x I3A. Here, C4A is a constant.
In R3, it is described as high-pressure heater inlet feedwater flow rate I8A,
while high-pressure heater water level I9A is the related process data to be monitored.
In this case, the following relation exists:
|I4A - I8A| ≤ C9A x I9A (C9A is a constant). Though not described in Fig. 9,
other fundamental rules in Fig. 8 are described in relation to the process data.
According to such fundamental rules, the table shown in Fig. 9 is
searched according to the data name sent through signal line 11 shown in Fig. 1.
The related process data (written in relation to the process data and correlation
established between the process data) is then read and output to signal line 3.
II. Data Base 26
Figure 2 shows the structure of the data base 26. The data base is
basically constructed to enable the effective use of the data. Thus, it is important
to have a data arrangement from which data can be easily extracted. The structure
of this device is determiend as being hierarchial, corresponding the the system
configuration of the object system and control device configuration (in relation
to the preparation procedures for fundamental rule section 24).
The data base may use a double structure of a main data base 26M
and a sub data base 26S. This is based on the fact that data base 26 is used as
the data source of an expert system, etc., and an inference engine based on the
supplied data contents. The data base is also used online to control the object
system during operation according to the results of inference processing. In other
words, the data base used the double structure of the main data base 26M for the
inference engine and sub data base 26S for checking the data content offline. The
main and sub data bases 26M and 26S have the same contents when operation is started,
if the operator determines that the data base contents require correction based
on experience acquired through actual plant operation, the operator must access
the contents of the sub-data base 26S to correct the data. Consequently, the contents
of both data bases will gradually become different. When the number of different
data items exceeds a certain number or the data containing important parameters
is corrected, the contents of the main data base 26M are corrected to match the
contents of the sub data base 26S, while the inference accuracy of the inference
engine is improved during online operation.
Generalizing part 26I is positioned between both data bases 26M,
26S to count the number of different data items or to set the matching timing between
the data bases.
Although Fig. 2 shows the hierarchial structure of the sub data base
26S as an example, the main data base 26M also has the same structure.
The object system shown in Fig. 7 is based on a hierarchial configuration
consisting of a main process, subprcess, uniprocess, equipment, and (in fundamental
rule section 24) in relation to the rule extraction to prepare the fundamental
rules, the information data belonging to the main process is stored in main process
data base which is divided into parts 26Sm1, 26Sm2.
The informational data belonging to the subprocess is stored in a
subprocess data base divided into parts 26SS1, 26SS2, 26SS3.
Further, a uniprocess data base and equipment data base divided into parts 26SK1,
26Sk2, 26SSK3 have similar configurations. The reason why a hierarchial configuration
was adopted is that some fundamental rules used in data edit processing must consider
the flow directions of liquids in the process and the influence exertion route.
Consideration must also be given to a process quantity because liquids determine
the boundary conditions in the continuous lateral direction for each division
in the hierarchial configuration, as well as to the effects exerted in the longitudinal
and lateral directions of the hierarchial configuration. Therefore, the information
can easily be incorporated when stored as individual items of data due to the advantage
in understanding the overall concept.
III. Management History Table
Various data input accesses of informational data bases are inevitable
during long-term operation. Therefore, the informational data base is known as
a "growing system" because final completion is never ahcieved. Therefore if the
operator and time for addition and correction work are not known after the initial
preparation, the information in the data base will become confising and disordered,
which will inhibit data base use. For example, two or more items of informational
data (for process data) may be stored in a data base, and conflicting conditions
may occur in the informational data. Otherwise, numerical discordance will occur.
In fundamental rule memory 24, data edit processing device 23 and selecting device
25 can only select input data that does not result in such confusion. Consequently,
correct selection is impossible. Furthermore, conflicting or erroneous data may
be created due to operator mistakes. When such erroenous data is registered in
the data base, the erroneous data cannot be extracted and displaying these results
outside cannot be achieved. If an error occurs in the results of inference engine
processing based on data in the informational data base, the presence of erroneous
data is clarified by assuming that an error is in the data that forms the basis
of the inference. In such cases, a data input management history table (shown in
Fig. 10) must be used to search for the time point when the error ocurred in the
above-described data input. All information related to such access controls as
renewal access and confirmation access of data in the adata base are recorded in
this table so that the history of the informational data can be determined. The
data input access history for the informational data base includes the name, qualification
and other items of individual information about an operator who has accessed the
system. It also includes the input knowledge-base information for each access operator,
and the type of information (addition, deletion, change, etc.) used in the operation.
Such important items of enviromental information as the date, day of week, load
and atmospheric conditions may also be used. When the data base structure has
a double data base structure of a main data base and a sub data base, access control
of data input is provided with individual records in both data bases, and information
related to the copying work between both data bases is also recorded in the history
IV. Data I/O Device
Data I/O device 21 is designed as a man-machine system that uses
a screen (CRT), keyboard, etc., to read data from and write data to the data base
26. It also displays and writes information to the management history table 27,
transfers data to the data edit processing device 23, displays the contents of
set fundamental rules, and updates the fundamental rules. One example of the display
function is the simultaneous display of the contents of data before and after
correction on the same screen. When the operator corrects the data, the corrected
part is easily confirmed by its changing its colour by underlining, or by flashing
indication. The guidance display during the operation-inhibited period prohibits
change, even if the management history data can be displayed and the alarm display
when the contents conflict with the displayed confirmation results, but not with
the fundamental rule when additional information can be placed on the right to
access the data base by the operator.
V. Data Edit Processing Device
When informational data to be input into the informational data base
26 is input using data I/O device 21, the input data is sent to the data edit processing
device 23 through the user interface section 22. The data is read into the device
23 according to the fundamental rule (including the ID number for identifying the
process data that makes up the data from fundamental rule memory 24). Moreover,
the contents are compared to determine whether they match. As a result, when the
contents match, they are determined to be normal. If the contents do not match,
but there is no conflicting data the contents are "reserved". If the contents are
conflicting, however, they are judged to be abnormal. For contents that are judged
to be abnormal, the process data that makes us the information is automatically
rearranged. For a serial information composition (A - B - C) consisting of three
process data items, the entire composition (BAC, BCA, CAB, CBA and ACB) is prepared
and checked according to the fundamental rule. In other words, like the same process
data, elements are rearranged, and only process data that does not conflict with
the fundamental rule is sent to the next stage of processing. If there is no such
conflicting data in the rearrangement, operation occurs and, the process data is
sent to the next stages of processing in the form of the original information.
In the next stage of transmission, the fundamental rule also serves as the basis
for a check to be made at the same time, together with the information for which
a normal determination has yet to be rendered.
VI. Selecting Device
The edit processing result of all data is input into conflict error
selecting device 25. When the data is normal the data is sent as is to data base
I/O device 21. Reserved data and abnormal data are appended with the fundamental
rule as the basis of determining reservation or abnormality, while abnormal information
is prepared as a whole. Furthermore, the selecting function for rewriting normal,
reserved and abnormal data into a form in which the operator of this device can
easily understand is provided. This is the basic function of conflict/error selecting
device 25. The information after selection is transmitted to data I/O device 21.
The question of each component of the system shown in Fig. 1 will
now be described, as will the effects that can be achieved when the operator executes
operation from data I/O device 21.
The operator of this device confirms the manually or automatically
prepared knowledge-base information on a display device (such as screen (CRT) in
data I/O device 21, then sends the knowledge-base information to data edit processing
device 23 through signal line 6, interface section 22, and signal line 2. The function
of user interface section 22 to achieve a conversion function for information
processing signals and external peripheral equipment between the memory in data
base 26. Bus 1, bus 7, etc., illustrate the flow of information, while a single-line
bus or multiple cable lines may be adopted in a practical embodiment. Data I/O
device 21 is equipped with a display device (such as a screen (CRT), keyboard,
and operating tools), and the information read by data I/O device 21 is displayed
on the screen (CRT). The presence of the designated information input by an operator
is also confirmed. After this confirmation, the operator transmits the information
to data edit processing device 23 through bus 2. The flow of information described
above is the same as the single data processing of each data processing of each
data item or the batch processing of multiple data items.
After receiving the input information from data I/O device 21, data
edit processing device 23 automatically reads the necessary fundamental rule through
bus 3 from fundamental rule memory 24 or based on a start instruction issued from
data I/O device 21. The required fundamental rule may be any of the fundamental
rules that are transmitted through user interface section 22 and signal line 11,
and prepared using an identification code attached to the process data that makes
up the input information. For example, for input information a1 >
b1 (a1, b1: process data), the necessary data
is obtained according to the fundamental rule using al and all fundamental rules
using b1. When the fundamental rules corresponding to the above are
three fundamental rules (R1: a1 = K+b1 (K >
O), R2: a1
< C1 + C3, R3:
b1 - c1 < O), all three fundamental rules are extracted,
then the data information a1 > b1 is applied to these
fundamental rules. The rules that conform, do not conform, and those for which
application is impossible are classified. The conforming rule is considered normal;
the nonconforming rule is considered abnormal; and the rule for which application
is impossible is considered to be reserved. In this example, fundamental rule R1:
a1 = aO + b1 (aO > O) is satisfied
and is determined to be normal. The other two fundamental rules (R2
and R3) are reserved. Further, in this example, if the first input data
information is a1 < b1, the judgement is as follows: R1
R2 (reserved), and R3 (reserved). In this case, elements
a1 and b1 are rearranged and compared with the fundamental
rule again. Although the contents subsequently match the above-described conclusion,
the rearranged part is attached to the index information.
The following describes the editing of specific information by using
bearing vibrration as an example in the previously described thermal power plant.
Based on the assumption input information Do (described below) exists, Ro is considered
the corresponding fundamental rule.
Fundamental rule Ro: Fundamental rule for MV of input information Do and rule considering
the influence source related to "vibration generation (major abnormality) 1; high
bearing lubricating oil temperature, exertion source 2; dangerous turbine speed,
exertion source 3; excessive rate of generator coil temperature variation.
- Information not applicable to edit processing, and which is currently in data
I/O device 21.
- Preparation date (19 April 1988, 23.00)
- Person responsible for preparation (A) (third-class qualified)
- TMS 550 °C, MV > LA
- Bearing vibration amplitude value
- Bearing vibration alarm set value
- Main steam temperature
In this case, operator B (first-class qualified) checks whether information
data Do (yet to be subjected to the above-described edit processing) is true or
false by using this device. Also,
(Considered part of the input information)
- Preparation date (20 April 1988, 23:00)
- Operator B (First-class qualified)
Knowledge-base information items Do1, Do2, and
Do3 are prepared by operator A, then are extracted from data I/O device
21 by operator B for display on a CRT not shown in the figure. This display form
conforms to the settings of individual devices. For example, when using the IF
- THEN form:
IF (TMS > 550°C), THEN (MV > LA), Input by (A -3) 1988-4-19-23.00.
Operator B confirms the contents displayed on the screen (CRT), then
transmits the contents to data edit processing device 23. Then, a search is made
for the corresponding fundamental rule (including process date) for each item
of process data (TMS, MV included in information Do3).
A check is also made for conflicting data in the above-described information. Because
Ro is the applicable fundamental rule in the example, the following expression
- IF (TBRG > LTBRG OR LSC1 < S < LSC2
OR RTGC > LRTGC)
THEN (MV > LA)
- Here, TBRG:
- Bearing lubricating oil temperature
- Bearing lubricating oil maximum temperaturs set value
- Dangerous turbine speed minimum set value
- Dangerous turbine speed maximum set value
- Turbine speed
- Generator coil temperature variation rate
- Generator coil temeprature variation rate maximum set value
Because there is no fundamental rule associated with TMS,
knowledge-base information Do3 (prepared by operator A) only refers
to fundamental rule Ro. A check is then made on rule conformity. The matching results
of fundamental rule Do and input information Do show that Do3
conflict with fundamental rule Ro, but does not conform to it.
Information on normal, abnormal, and reserved conditions after the
above edit processing work is transmitted to conflict/error selecting device 25
through bus 4, with the fundamental rule that forms the basis of judgement. The
function of the selecting part is not affected even if the operation is executed
in edit processing device 23. The function relieves the function's workload to
divide the equipment in each processing mask.
Although the information transmitted as normal data to selecting
device 25 may be received in singular form and transmitted to the next stage of
processing, the information applied to the rearrangement is further applied to
the comment addition processing to help the operator understand the index information.
For this work, preparations are made in advance in th user interface section then
operation is started by the index information signal. The information that conflicts
with the contents of the fundamental rule is attached to the fundamental rule
that forms the basis. The information to which information is attached is then
prepared and transmitted to the data I/O device. Although the reserved information
may be deleted (ignored) by the selecting function here, the information is attached
to the fundamental rules as reference data for the operator before being transmitted
to the next stage and displayed on a screen (CRT), where it is finally judged
by the operator. For this reserved information, Do3 (which is transmitted
to conflict/error selecting device 25 through bus 4) and fundamental rule Do with
the fundamental rule are set into pairs to form one unit of reserved information.
The form of the reserved information is expressed as follows:
Do3 - PENDING
By rule (IF TBRG > LTBRG OR
LSC1 < S < LSC2 OR
RTGC > LRTGC)
Note that when the results (abnormal or normal) are determined, the
data attached with the reason (applicable fundamental rule) is transmitted to data
I/O devide 21 with a form of expression form that is considered the same in these
As previously described, the information arranged in the conflict/error
selecting device is transmitted to the data base I/O device through bus 5. The
I/O device is set into the wait state so that the necessary information can be
immediately displayed on the screen (CRT) immediately when requested by the operator
(according to operator instructions). Therefore, sufficient memory capacity is
important when considering the overall information of processing capacity.
Although the evaluated data is collected in data I/O device 21, some
data stored in data base 26 cannot be used as valid data. Thus, measures must be
taken to prevent valid data from being deleted during this deletion operation.
For this purpose, a deletion instruction signal for all the information is sent
according to a method that gives consideration to preventing erroneous operation
by the operator. For example, the information to be deleted must be confirmed
through conversational processing using the screen (CRT) to prevent the operator
for accidently deleting any edited information. Another measure taken to ensure
correct deletion is that the data read into the I/O device 21 is stored in a temporary
memory area of the interface section for subsequent deletion after a series of
operations. The information waiting in the data base I/O device 21 (as instructed
by the operator) is displayed on the display device (i.e. the screen (CRT) and
matched with the information read and temporarily stored at device initialization.
For normal data, the same data is displayed in the display area is different from
the original data, this part and the fundamental rule are simultaneously displayed
in the display area before and after correction. If part of the data is different
from the original data, this part and the fundamental rule are simultaneously
displayed to help the operator determined the actual differnces. The operator checking
the displayed contents executes an operation on the device side to confirm that
confirmation has been completed in principle, determines the correctness of storing
the data in the data base, the stores the data in the data base. To reduce the
operator's workload, the data can be automatically stored in the data base when
the operator determines that the data is information related to a simple subject.
To enable the operator to complete the above opearations, such basic
and required operation tools as those for addition, deletion, change, storage,
and reading must be provided in the I/O device of the data base.
The information after data editing and selecting is storedf in data
I/O device 21, and is displayed for confirmation as being "stored" or "deleted"
by the operator. For reserved information, operator B determines the correctness
of storing data in the data base when the check results obtained through this I/O
device 21 does not conflict with the fundamental rule, but which were not determined
as being normal. When operator B is qualified to determined whether the information
is to be stored in the data base, he/she can also make this determination for
reserved information. Note that there are many cases where the check results are
reserved among the set of fundamental rules, while the degree for operator qualification
is based on the operator's ability to make the proper decision in such cases.
For example, if operator B is informed that no causal relationship
exists betwween the rise in main steam temperature and the bearing vibration amplitude
value, the reserved information can be deleted.
When operator B determines that information Do is to be stored in
the data base and this storage operation is executed, data Do3 is stored
in the data base, while items of information Do3- Do5 related
to the operation are recorded in the management history table 27 (Fig. 10) through
bus 9.. The operator's name, qualification (Do5), content of information
before and after correction (information, date Do4), load of preparing
information, and any important items of enviromental data (if prepared) are recorded
in the management history table 27. This table can also be used to prepare a summary
table by using the data base access order and operator sorting function so that
the history information can be read later as required. In this example, after operator
B stores the reserved information Do in the data base, then is informed a week
later that a judgement error was made, the management history table 27 can be
opened to delete the contents of processing stored in the data base. Note that
one term is added to the history term in the management history table when the
data for actual information stored in the data base is deleted.
Therefore, only the history of the changed part (from the original
data base) is batch-controlled so that when information in the data base is corrected,
the management history table 27 is opened to compare each item of information
before and after data correction. At this time, the operator confirms the information
to be deleted, then the specified deletion is executed. In this way, the operator
can access the data base. Although the management history table 27 enables the
contents to be displayed for confirmation, the contents cannot be changed. This
is because management history is necessary for data correction. If the history
could be changed, for example, it would become difficult to determine the established
guidelines for correcting data in the informational data base. As a result, all
edited data in the data base would have to be rechecked. Thus, when the data base
contains a large volume of data, such management is essential for maintaining
the reliability and integrity of the data contents.
The following method is used to display information in the career
management table onto the screen (CRT) of the data I/O device 21. According to
theis method, whether the plant operation state matches the related knowledge-base
information on the data base can be easily and visually determined. In other words,
the processing system diagram designated by the operator is displayed on the screen
(CRT) as shown in Fig. 12. After the type of process quantity to be observed is
designated and the position in the designated system is specified, the process
quantity can be specified. Searching for recorded information in parameter oriented
form can be done by real-time display on the screen (CRT) of the knowledge-base
information that contains the specified process quantity (as compositio elements)
and which is stored in the management history table 27. This method can also be
commonly used with a selection system that uses a mouse to enter characters. Moreover,
the operator can easily understand the information through the color-display of
the high or low usage frequency of data in a specific process level system diagram,
based on the number of times that a process quantity (of knowledge-basic information)
has been used. In this case, such usage can be better understood by the operator
when a graphics display is set according to an arbitrarily specified number of
times of use).
The following desribes the portable performance of information in
the management history table 27. Although the need for information recorded in
the management history table 27 is considered essential, this requirement is basically
satisfied by displays enabled through the man-machine section of the data input
control system. This recorded information is very important in terms of helping
the operator use and operate the informational data base over long-term plant
operation, and to evaluate successful plant hardware operation. Management is provided
so that when the principles of plant operation or technical level of the operator
are examined, this information can be easily extracted so that the contents can
Therefore, such equipment as a floppy disc drive mechanism, memory
read and write mechanism, a screen hard copy mechanism, and a mechanism for copying
onto IC cards or ID cards must be provided.
To extract the knowledge-base information in real time and to read
out the above recorded information to an external device, each item of information
has an external device, each item of information has an individual identification
code. Information is extracted according to the index of information recorded according
to the corresponding assortment of periods, operator qualifications, types of
operations, and operating terminals. In this operation, an input sequence of entering
the required date, name qualification, terminal used, and type of operation for
each operation is provided. Thus, operation can be executed by simply monitoring
the establishment conditions of each type of assortment.
In data I/O device 21, data Do3 (to be stored in data
base 26) is stored in data base 26 and management history table 27. The old data
is deleted. In this case, the operator adds and deletes the information in data
base 26 (in Fig. 2) in an off-line mode for sub-data base 26S because the data
is not necessarily to be used in inference processing, etc., immediately after
being stored in the data base. Data Do3 (in the sub-data base) is stored
in the bearing system uniprocess data base section of the data base. This is because
the data base structure is that of the bearing system uniprocess as one process
in the hierarchial division at each plant process level. The usage frequency of
the knowledge-base information is also considered when bearing vibration is generated.
In other words, when bearing vibration is generated, a similar vibration tends
to occur in the contiguous bearing system in many cases. When considering the maintenance
of the entire bearing system, the acquisition of information related to multiple
bearings is desired. Thus, when the bearing system uniprocess batch-extracts information
from the data base of the subprocess (including higher-level bearing systems),
the mutual correlation and casual relationship can be easily understood. Furthermore,
when the main data base has the same structure as that of the above sub-data base,
the main data base is used to provide internally used information, while the sub-data
base is used to complete stored information. In this case, the main data base
is used online during plant operation, while the sub-data base is used offline
during plant operation. Thus, the storage and utilization of information can be
clearly and rationally divided. In this type of arrangement, even if erroneous
information is input into the sub-data base, the erroneous information can be
corrected. For example, previously input contents can be deleted up to when the
sub-data base contents are copied into the main data base. More specifically, a
qualified operator should check the management history table after completing
the preceding copying work, immediately before copying the contents of the sub-data
base to the main data base. More effective data base processing is also afforded
by the following:
- a) Before accessing the informational dta base, qualification is checked by
using a magnetic card or an IC card, then data is entered by using a portable and
simplified input device. The manufacturer who prepared the informational data
base device as the supply source of the knowledge-base information transmits the
data over a transmission line.
- b) Even if a matching-enabled condition for both the main data base and sub-data
base is not satisfied, the operator must match the contents of both data bases
based on this accidental occurance, and decide not to execure inference calculation
using the knowledge-base information of the main data base. A function for interrupt
processing that ignores matching-enabled conditions may be provided. As shown
in Fig. 11, a contents-matching instruction is issued from each interrupted data
base, and the task of registering other knowledge-base information in the data
base is temporarily stopped. Only where the stop operation completion condition
and AND condition of the interruption instruction are satisfied, and both data
contents allowed to be matched. At the same time, if a delayed-start instruction
is issued to prevent other tasks from being started during the matching work, and
to obtain the necessary time for matching. The contents-matching instruction that
was transmitted once is held until the matching coincidence completion is satisfied.
When the degree of difference between the registered contents and
main data base contents used for operation increases due to addition, correction,
or other operations involving the knowledge-base information in the sub-data base,
the operator must confirm this degree of difference. Therefore,the contents of
both data bases are compared, and the different parts are displayed on the screen
(CRT) or output to an external output device, etc. The method of outputting the
order, range, and form of data may be in real-time sequence, at a specific process
level, or in a form divided into right and left parts on the screen (CRT). By
assuming the above-described difference between both data bases, plant operations
ranging from the inference engine to the operation terminal can be simulated using
the knowledge-base information in the sub-data base. Therefore, the response of
a plant processing can be confirmed in advance by using only the sub-data base
contents related to control system operations involving the operation terminal
according to the knowledge-base information in the main data base.
d) Qualified operators having the right to access the informational data base are
classified as A and B. "A" refers to a qualified operator on the manufacturer side
of this devide; "B" refers to a qualified operator on the user side. As shown
in Fig. 13, operator "A" can use any of the terminals 60 (1,2,3,4,) installed on
the manufacturer side during preparation or testing or to enter the basic required
knowledge-base information into the device. Operator "A" can also use any of the
terminals 60 (5,6,7,8,) on the user side (after delivery) for instruction-assisted
technical servicing. Conversely, a qualified operator B on the user side can only
use one of the terminals 60 (5,6,7,8,) on the user side. Restrictions are also
placed on the applicable ranges of data base access changed by "A". Because various
restrictions are placed on the terminals used and the range of data base access,
operators A and B must work on the respective manufacturer and user sides, and
must generally work with the divided sections inside. Since the operator's role
in management in each section is limited, and because a mutual relation is established,
the installation site of each terminal 60 must be the same in each section.
Figure 14 shows an example of the registration form for entry into
the informational data base by an ooperator using the data base input control system.
When a qualified operator having the righ to accesss the data base
observes vibration V due to a drop in plant equipment bearing oil pressure (P)
at time T=To, and the related contents are registered as knowledge-base information
in IF-THEN form, subject A(P) and subject C(V) are registered as A(P) → C(V)
in the data base (in hierarchial form at the subprocess level) with the sufficient
and required conditions. After item T1 elapses, a qualified operator
having the right to access the datq base observes vibration V due to a rise in
the plant equipment bearing discharged oil temperature (T). If the related contents
are registered as knowledge-base information in IF-THEN form, this information
is then registered in the data base by adding contents B(T) → C(V) to the
above-described knowledge-base information A(P) → C(V). In other words, registration
is done using A(P) or B(T) as the sufficient condition and C(V) as the required
condition. To register data in the data base, it must be recorded in a form that
considers the coordination between individual elements that make up the information.
Otherwise, the data remains in the operation form used in the management history
table. In the above-described example, at time T = To, the input knowledge-base
information (displayed on a (CRT) screen 0001) is converted into the data base
registration form (displayed on a screen (CRT) 0101), then is stored. At this time,
the form corresponding to screen (CRT) 0001 is recorded in the career management
table as is to prevent accidental deletion. Moreover, after time T1
elapses and the operator is to input screen (CRT) 0002 into the data base, the
same condition at that applying to the required condition part C(V) is searched
for in the data base. Only the supposition condition B(T) is stored in the data
base as an element of knowledge-base information displayed on the screen (CRT)
002. In addition, the necessary programs are run to process this element in parallel
to the other element A(P) stored with the registered-base knowledge information.
Then, the form displayed on a screen (CRT) 0102 is converted in the data base,
and the element is registered. At this time, the data base memory location corresponding
to screen 0101 is reset and replaced by screen 0102.
By adopting this invention, even unskilled operators can easily obtain
the correct informational data. Moreover, if incorrect informational data is registered,
the data can be easily corrected.