JPH0535990A - Alarm processing supporting device in plant - Google Patents

Abstract

PURPOSE:To suitably support the condition judgement of an operator in the case of generating abnormality at a plant generating a lot of alarms. CONSTITUTION:A knowledge data base 18 is constructed and preserved for the attributes of alarms, causality classified at every condition of generating the alarm, knowledge concerning processings in the case of plant abnormality, the condition of activating the constitutive equipments of the plant, causality to relate the operations of the equipments and the state of the plant and knowledge concerning the levels of effects exerted upon the state of the objective plant by the operations of the equipments. Then, whether the abnormality is generated at the plant or not is judged based on alarm signals, equipment state signals, process signals and preserved knowledge, and the judged result is preserved as plant signal data 19 together with the respective signals. When the plant abnormality is judged, an alarm developing route is identified at every. abnormal event, the alarm of cause is judged and a sequence operation is monitored. Then, the state of the plant is decided against the abnormality, and a control command to the equipment to be operated and a process guide to be presented to the operator are decided.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an alarm processing support device for a nuclear power plant, a thermal power plant, a chemical plant or the like, which is suitable for abnormal processing of a large-scale plant, in particular.

[0002]

2. Description of the Related Art In large-scale plants, many alarms are set as important indexes to be monitored in various operating modes such as a starting state and a rated output state. Especially, since the central control room controls the center of the operating function, the control panel displays alarms indicating the state of equipment such as "low pressure" and "high temperature", and the entire plant such as "isolation" and "scrum". There are various alarms that indicate the status.

When an abnormality occurs in a plant, its influence is often observed as a transient change of a plurality of signals,
If the change is large, multiple alarms are issued. Plant operators are required to promptly judge the state of the plant from these multiple alarms and take appropriate measures.However, when the scale of the abnormality is large or the propagation of transient phenomena is fast, , The burden on the operator becomes excessive,
It may lead to misjudgment. Therefore, an alarm processing support device has been developed for the purpose of supporting the operator's situation judgment in the event of an abnormality.

[0004]

As an alarm processing support device of this type, a device which determines an abnormality cause alarm and a latest alarm, and predicts a future alarm has been developed. However, they did not have the function of displaying the processing manual for the important warning. Also, for important devices, from a fail-safe standpoint, there is a function to automatically start a backup system with the same function to prevent the propagation of abnormalities, but if the backup system fails to start, The operator had to select and implement the operation including the following measures. Further, even if the cause alarm is determined by the alarm processing support device, the corresponding action to be performed later may be different depending on the cause event. Therefore, the operator determines the cause of the abnormality after confirming whether or not the result obtained from the alarm processing support device is correct from the subsequent event progress based on the knowledge about the structure and design of the plant and his empirical knowledge. However, an appropriate treatment must be selected based on the result of the determination, leaving a possibility of misjudgment.

Furthermore, in the conventional alarm processing support device,
It has been developed to save the expected sequence of alarm occurrences in advance as a database for each abnormal event, and to determine the cause of the abnormality by comparing the alarm sequence with the database. However, there is a problem that it is difficult to deal with an unexpected event, and a great deal of manpower is required for database creation and management.

The present invention has been made in consideration of the above circumstances, and provides an alarm processing support device for appropriately supporting the operator's situation judgment when an abnormality occurs in a plant where alarms frequently occur. The purpose is to

[0007]

In order to achieve the above object, an alarm processing support apparatus of the present invention is provided with an alarm signal for notifying a plant abnormality, a status signal indicating the status of equipment constituting the plant, and a process signal from the plant. , Etc. online, a signal input means, the attributes of the alarm that the alarm signal means, the causal relationship divided by each alarm occurrence condition, and knowledge about processing when a plant abnormality occurs, and the starting conditions of the equipment that constitutes the plant A knowledge database constructing means for constructing and storing as a knowledge database a causal relation associating a plant condition with a control command, equipment operation and a plant state, and a signal input means Whether there is an abnormality in the plant based on the signals captured by the system and the knowledge stored in the knowledge database A state monitoring means for judging, and stores the determination result as the plant signal data with the signal,
When the status monitoring means determines that the plant is abnormal,
Using the knowledge stored in the knowledge database, the diagnostic means for identifying the alarm progress path for each abnormal event that occurred, determining the cause alarm, and performing sequence operation monitoring, and determining the state of the plant that deals with the abnormality, Using the plant state obtained by the diagnostic means and the knowledge stored in the knowledge database, the corresponding treatment determining means for determining the equipment to be operated, the control command for the equipment and the treatment guide to be provided to the operator, and the corresponding treatment determination. The device control means for sending and controlling a control command to the equipment determined to be operated by the means, the diagnosis result obtained by the diagnosis means, the determination result obtained by the state monitoring means, and the corresponding action determination means. Process for providing the determined treatment guide to the operator automatically or upon request, and knowledge database constructing means Characterized by comprising by input and output processing means for registering and editing process knowledge.

The knowledge database constructing means develops knowledge about alarms in the knowledge database in a three-dimensional hierarchical space based on the degree of influence of the generation / non-generation of alarms on the plant and the functional meaning of the alarms. -Can be saved.

Further, it is preferable that the construction of the knowledge database by the knowledge database construction means is performed off-line.

The knowledge database construction means can execute knowledge registration and editing processing by drawing a causal relationship between alarms as an alarm sequence diagram on a CRT of a computer by using an input means such as a mouse or a keyboard. it can.

The diagnostic result display means displays the result by using an alarm sequence diagram showing the progress route of the generated alarm.

The equipment control means judges whether or not the equipment selected can be operated by comparing the current plant state with the operating conditions of the equipment. If the equipment cannot be operated, the target is determined based on the knowledge of the equipment. Automatically create alternatives to achieve different plant conditions.

[0013]

In the alarm signal used in the present invention, not only the alarm signal set on the control panel of the plant but also the status signal of the equipment and the process signal necessary for identifying the alarm progress path, or the system A signal that does not exist on the control panel, which newly creates a state in the apparatus of the present invention by monitoring the overall state and determining normality / abnormality, is also included. Therefore, in the knowledge database, signals indicating all states in the plant such as devices, alarms, and grids are registered.

The diagnostic means of the present invention operates by projecting a newly generated alarm state on these, based on the data relating to the network-like alarm causal relationship for each hierarchy registered in the knowledge database. The alarm progress path 3
Dimensionally identify. As a result, of the generated alarms, the most upstream alarm in the route identified on the highest hierarchical level is determined to be the cause alarm, and the downstream alarms are checked by checking the alarms generated within the same level for downstream alarms. Forecasts are provided, and the operator is provided with a guide regarding the countermeasures for these forecast results and the progress route of the generated alarm. The knowledge about the treatment is registered in advance for the route indicating the causal relation of the alarm, and dynamically created for the identified progress route.

The above sequence operation monitoring (sequence
Monitoring) is based on the causal relationship between the alarms generated by the safety protection system and plant interlocks registered in the knowledge database and the time delay, and uses the data on the next alarms predicted by the diagnostic means. ,
It is to monitor whether or not it has occurred correctly as expected. For the basic operation of the sequence monitoring, refer to "Plant abnormal time action determination method" disclosed in Japanese Patent Laid-Open No. 61-228501.

The alarm signals registered in the knowledge database include, in addition to the actual alarm state as described above, a signal indicating the plant state. Generally, the larger the range in which the abnormal state propagates, the greater the number of alarms to be issued.
Even if it is possible to simply reduce the number of alarms generated here, appropriate measures cannot be taken unless the current plant state can be accurately grasped. Therefore, for example, when the alarms a, b, and c are generated and the alarm d is not generated or is not generated according to the alarm sequence, the plant state e that has a larger influence on the plant than before is given. From plant status e to alarms a, b,
It is positioned as an upper alarm of c and d, and that fact is registered in the knowledge database (the plant state e is generated when alarms a, b and c are generated and the alarm d is not generated). For this reason, the signal indicating the plant state is mainly registered in the higher rank of the knowledge database.

As described above, when the propagation range of the abnormality is local, the diagnosis is mainly performed by using the lower level knowledge, and the influence is wide, and when a large number of alarms are generated, the upper level knowledge is used. It is possible to make a diagnosis by using knowledge and grasp the condition from a global perspective. In other words, the diagnostic result at the highest hierarchical level among the dynamically identified evolutionary paths is
It means that the plant state such as "what kind of state is it now?" Further, similarly to the process of predicting the next-generation alarm, it is possible to predict the condition to be taken next from the plant condition predicted from the diagnosis within the above level.

As much as possible, a plurality of equipment candidates to be operated from the predicted plant state are selected based on the knowledge about the causal relationship associating the operation of the equipment with the plant state. Of the selected equipment, the equipment most effective for the target plant state is determined as the equipment to be operated. Next, it is determined whether the selected equipment is inoperable by comparing the current plant state and the equipment start-up conditions, and if it is operable, a control command is sent to those equipments. If it is inoperable here, by selecting / controlling the next effective device from the above candidates,
The target plant condition can be achieved. If there is no candidate for a device to be operated, or if all devices are inoperable, a message to that effect is displayed to the operator.

The registration / editing of knowledge by graphically representing an alarm sequence (alarm sequence diagram) enables the operator, who is the user, to determine the degree of influence (number of layers) of the occurrence of an alarm on the plant and the level of one layer. The number of alarms to be registered, the alarms to be positioned in the upper hierarchy level and the number thereof can be arbitrarily set.

The display of the diagnostic result by the alarm sequence diagram is performed by displaying the alarm generated in the created sequence diagram and its development route in different colors, blinking, etc.
Dynamically display whether or not there is a sequence error. The display of diagnostic results provides the operator with what was obtained at the highest hierarchical level. It should be noted that, at the request of the operator, it is possible to provide results (information regarding detailed cause alarms) at other hierarchical levels.

[0021]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in more detail below with reference to the drawings showing the embodiments.

FIG. 1 is a functional block diagram of an alarm processing support device in a plant constructed according to an embodiment of the present invention. The illustrated alarm processing support device 10 includes a signal input unit 11, a state monitoring unit 12, a diagnostic unit 13, a corresponding action determination unit 14, a device control unit 15, and an input / output processing unit 1.
6, knowledge database construction means 17, knowledge database 18, plant signal data storage means 19, and prediction warning data storage means 20.

The alarm processing support device 10 is activated at regular time intervals, and outputs an alarm signal for notifying an abnormality of the plant 2, a status signal indicating the status of each component, the plant 2
Process signals such as pressure and temperature are captured online via the signal input means 11.

The state monitoring means 12 is based on the inputted alarm signal, equipment state signal, process signal, etc. on the basis of the knowledge about the condition for judging the presence / absence of abnormality of each signal registered in the knowledge database 18. And plant 2
It is determined whether or not an abnormality has occurred in the plant signal data, and the determination result is stored in the plant signal data storage means 19 together with each signal as plant signal data. Also, before the previous diagnosis time, the delay time of the alarm predicted by the diagnosis means 13 is counted down, and it is checked whether or not the alarm has been generated after the lapse of a predetermined time. If the warning is not issued as expected, it is determined that the sequence is abnormal, and the input / output processing means 16 described later sets the data to be displayed.

The diagnostic means 13 is activated when the status monitoring means 12 detects an abnormality. Here, based on the knowledge of the cause and the result of each alarm registered in the knowledge database 18, by projecting the evolution route of the alarm generated in the plant signal data onto a three-dimensional hierarchical space, for example, as shown in FIG. Dynamically creating a network-like alarm progress route as illustrated in Fig. 2, and determining and identifying the uppermost alarm of the progress route at the highest hierarchical level, which was created only by the occurrence alert, as the cause alarm. Note that, in FIG. 2, the circles filled in a mesh indicate an alarm generated, the hollow circles indicate an alarm not generated, and the arrows indicate causal relationships between alarms. In addition, in the figure, the upper part has a relatively large effect on the plant, and the lower part has a relatively small effect.

The causal relationship between the two alarms (a1, a2) can be roughly classified into two types. The first type is, for example, that if a "coolant pump trip" (a1) alarm is issued, a "reactor scrum" (a2) alarm is issued. This is a case where the alarm a2 is generated with a 100% probability due to the normal sequence operation of the lock mechanism. The second type is "coolant flow rate low".
(A1) Even if an alarm is issued, "low coolant flow rate" (a
2) This is a case in which the occurrence of the alarm a1 is not necessarily linked to the occurrence of the alarm a2 with a probability of 100%, such as that an alarm is not always generated.

In the case of the first type of causal relationship, an alarm a1
If the alarm a2 does not occur after the occurrence of the predetermined delay time t, it is determined that the sequence is abnormal. Therefore, in the knowledge database 18, the causal relationship between the preceding alarm and the subsequent alarm is registered, and when the sequence operation is performed, the delay time thereof is also registered, and the alarm newly generated is generated by the sequence operation. The presence / absence of the alarm a2 to be checked is checked, and if there is, it is determined as a predictive alarm and registered together with the delay time. Then, as described above, after the next diagnostic cycle, the state monitoring means 12 checks whether or not a predictive warning has occurred after a predetermined time has elapsed. Also, the action information (plant state to be taken next) for the alarm progress path identified based on the knowledge database 18 is dynamically created.
For example, in the above examples of "coolant pump trip" and "reactor scrum", the target plant conditions such as the need to shut down the reactor and to secure the coolant necessary for decay heat removal. Is obtained at the same time as the prediction of “Reactor Scrum”.

The knowledge database 18 includes attributes such as the name of an input signal, a signal representing an alarm or a device state, a type of process signal or not, a condition for determining whether an abnormality has occurred, a cause of occurrence and a result of occurrence of each alarm occurrence condition, Knowledge of the rank of the hierarchy to which it belongs, actions related to operation or non-operation of a series of alarm sequences determined at the time of plant design (see "Knowledge about alarm" in knowledge database 18 of FIG. 3), start conditions and control of equipment A causal relationship associating the command and the operation of the device with the plant state and knowledge about the magnitude of the effect of the operation of the device on the plant state (see “Knowledge about the device” in the knowledge database 18 of FIG. 4) are registered.

Incidentally, the "knowledge about alarm" in the knowledge database 18 of FIG.
For n, knowledge An1 regarding attributes, knowledge An2 regarding generation conditions (knowledge regarding rank of the layer to which the attribute belongs, etc.) are included. Further, "knowledge about equipment" in the knowledge database 18 of FIG. 4 includes knowledge n1 about attributes of the operation n of the equipment, knowledge n2 about the relationship with the plant state, and the like. Such knowledge registration is performed according to a procedure shown in FIG. 6 described later by drawing an alarm sequence diagram on the CRT using an input device such as a mouse or a keyboard offline before starting diagnosis.

These processes for the knowledge database 18 are performed by the knowledge database constructing means 17 via the input / output processing means 16. The registered knowledge is created so as to have a series of connections in a three-dimensional space based on the degree of influence of the occurrence / non-generation of an alarm on a plant and the functional meaning of the alarm. Therefore, the above knowledge may be created so that all the alarms in the level have a series of connections on the same two-dimensional hierarchical level, and it is possible to create a partial connection such as a device, a unit, or a system. You may divide and create it.

The alarm registered in the knowledge database 18 represents the state when the plant is viewed from a global perspective as the influence on the plant increases. For example, for a nuclear plant, the top level of the hierarchy is
Only 2 alarms such as "Reactor Scrum" alarm → ← "Turbine Trip" alarm are registered (These alarms are generated in the lower level of the above "Reactor Scrum" alarm or "Turbine Trip" alarm. Cause alarm is registered).

Therefore, these simply represent the alarms hierarchically from the relation of the plant configuration (parts level → device level → unit level → system level, etc.) or represent the device status as one alarm. Is a little different from. With the configuration of the knowledge database 18 as described above, the plant state can be grasped at a glance even if many alarms are issued.

In the case of a coolant circulation pump or a lubricating oil supply pump of a nuclear power plant, for example, the hierarchical expression of the above-mentioned alarm causal relationship means that the function of the pump is to prevent the bearing of the circulating pump from being damaged. If the standby unit of the supply pump automatically starts when the supply pump trips, it is possible to prevent the expansion of abnormalities, and further to continue operating the plant by using the standby unit as a service unit afterwards. , "Coolant circulation pump bearing abnormal" alarm, "coolant circulation pump lubricant oil supply pump trip" alarm as a starting point "supply pump standby machine start" signal, and "coolant circulation pump lubricant oil pressure low" It is equivalent to position it as a high-level collective alarm. Note that the "coolant circulation pump bearing section abnormality" alarm is an alarm that does not exist on the control panel, and its state is created inside this device. Therefore, a "coolant circulation pump lubricating oil supply pump trip" alarm is generated, and the supply pump standby machine fails to start ("supply pump standby machine start" signal is not generated), and "coolant circulation pump lubricating oil" When the "low pressure" alarm is generated, the "coolant circulation pump bearing abnormality" alarm is said to have occurred.

In the corresponding action determining means 14, the diagnosing means 13
Using the target plant state obtained by the above and the knowledge stored in the knowledge database 18, the operator 5
Determine the equipment to be operated and the contents of control commands for the equipment. First, from the obtained plant state, the operation of the equipment required to achieve it is determined. In the case of the above-mentioned “reactor scrum”, two states are obtained: the reactor needs to be shut down and the coolant necessary for the decay heat removal is secured. Therefore, from the knowledge related to the operation of the equipment and the plant state registered in the knowledge database 18, the control rod should be urgently inserted in the former case, and the pony motor or the auxiliary cooling system should be started in the latter case. It is possible to obtain the equipment necessary to achieve the above state such as, and its operating state. Compared with the case of using the auxiliary cooling system, the target condition of the plant can be achieved more reliably and quickly with the pony motor, so the cause of the pony motor startup and decay heat removal (device operation and plant condition) It is registered in the knowledge about the relationship that the effect of the equipment operation on the plant state is great. As a result, even if a plurality of candidate device operations that achieve the target plant state are selected, as in the latter example described above, the situation can be improved by selecting the device operation that gives the greatest effect. Appropriate equipment can be selected.

Next, the starting condition of the selected device is checked to determine whether it is operable. In the latter case, it is determined whether the pony motor start condition, that is, the pump speed T seconds after the pump trip is X rpm or more, and whether the pony motor start switch is in the automatic start position. Corresponds to this. If the pony motor cannot be started by the pump stick, the procedure for securing the coolant by starting the pony motor is canceled, and the auxiliary cooling system is selected. These processes are checked until the corresponding device is selected, and the device to finally operate is determined. If there is no device that achieves the target Punrat state, or if all the selected devices are inoperable, a message to that effect is displayed to the operator 5.

The device control means 15 sends a control command to the device to be operated, using the knowledge about the control command of the device, in accordance with the content determined by the corresponding action determination means 14.
Automatically control in response to abnormal events. These control commands may be given directly to the plant 2 or may be given via another control device 3.

The input / output processing means 16 provides the operator 5 with the diagnosis result obtained by the diagnosis means 13, the sequence operation status obtained by the state monitoring means 12, and an appropriate treatment guide automatically or in response to a request. The process of providing and the process of off-line registration / editing of knowledge in the knowledge database construction means 17 are performed. The alarm sequence diagram that can be understood at a glance is used to display the diagnosis result. This is to display the generated alarm and its progress route in different colors in combination, blinking, etc.Alarm sequence diagram or tree-like sequence diagram starting from the newly created cause alarm To use.
In the present invention, the information on the progress route at the highest hierarchical level is automatically provided by the occurrence alarm, but detailed lower level information can also be displayed at the request of the operator.

In FIG. 1, a solid arrow indicates a data flow when online, a dashed arrow indicates a reference data flow in each means, and a diagonal double short dashed line arrow indicates a data flow when offline. .

Next, a series of procedures from online signal input to device control will be described with reference to the flowchart of FIG.

First, various alarm signals relating to the plant 2 are taken into the alarm processing support device 10 via the signal input means 11 (step 41). The state monitoring means 12 checks whether or not a predictive alarm has been issued (step 42). If the predictive alarm has not occurred, the sequence abnormality data is set (step 43), and if the predictive alarm has occurred, it is immediately checked whether a new alarm has occurred (step 44). If no new alarm is generated, this flow ends as it is. If a new alarm is generated, the diagnostic means 13 identifies the alarm progress route (step 45) and determines the cause alarm (step 46). After that, the warning to be generated next is predicted (step 47), and the target treatment determining means 14 sets the plant target state (step 48). Step 4
The predicted warning data obtained in step 7 is stored in the predicted warning data storage means 20. The target state set in this way is displayed on the input / output processing means 16 as a diagnosis result (step 49), and the device control means 1 is operated accordingly.
5, the corresponding equipment of the plant 2 is controlled directly or via another control device 3 (step 50). In each of the above steps, the knowledge stored in the knowledge database 18 is referred to as necessary, and the plant signal data stored in the plant signal data storage unit 19 is referred to.

Registration of knowledge in the knowledge database 18 is performed by the knowledge database constructing means 17 in an off-line process according to the procedure shown in FIG.

First, it is checked whether the target registration is a new registration (step 61). If it is a new registration,
The knowledge is edited / created in accordance with the above-mentioned point (step 62), registered in the knowledge database 18 (step 63), and the process ends. If it is not new registration, that is, if the already registered knowledge is changed, the knowledge to be changed is read from the knowledge database 18 (step 6).
4) Change the read knowledge content (step 65), re-register it in the knowledge database 18 (step 63), and end the registration procedure.

A more specific description will be given below by taking the case of the "coolant pump trip" alarm in a nuclear power plant as an example.

Here, in order to prevent the coolant pump from being damaged due to a decrease in the flow rate, a "coolant pump rotational speed low" alarm is generated when some abnormality occurs in the flow rate control system of the coolant pump.・ Assume that a “trip” alarm has also occurred.

In the knowledge database 18, the "coolant flow rate low" alarm is registered as a higher level alarm than the "coolant pump trip" alarm, and the "coolant pump rotational speed low" alarm to the "coolant pump The causal relationship due to the plant interlock is registered in the "trip" alarm on the same level.

Since the coolant pump has tripped, a pony motor is selected as a device to be operated and activated to remove decay heat. Further, the operator 5 is provided with a treatment guide such as confirmation of decay heat removal and confirmation of plant safety stop.

It is also assumed that the "coolant pump rotational speed low" alarm is also issued after the "pump trip" alarm is issued.

In this case, since the "low rotation speed" alarm is generated and there is a possibility of a pump stick, the operation of removing the decay heat by the pony motor is canceled to protect the pump bearing. , Start the auxiliary cooling system. The operator 5 is provided with a treatment guide such as confirmation of the stop of the pony motor and investigation of the pump bearing portion.

[0049]

As described above, according to the alarm processing support device of the present invention, the equipment can be automatically controlled according to the abnormal event while confirming the plant state at the time of abnormal condition. The necessary treatment guides can be obtained at any time depending on the situation, and the knowledge can be easily constructed and managed, which makes the operator's situation judgment more reliable, reduces the risk of operator's erroneous operation, and improves plant safety. Can be expected to improve.

[Brief description of drawings]

FIG. 1 is a functional configuration diagram according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a network-like alarm progress route.

FIG. 3 is a diagram showing a configuration example of a knowledge database regarding alarms.

FIG. 4 is a diagram showing a configuration example of a knowledge database regarding equipment.

FIG. 5 is a flowchart showing a process when online.

FIG. 6 is a flowchart showing processing when offline.

[Explanation of symbols]

2 plants 3 Other control devices 5 operators 10 Alarm processing support device 11 Signal input means 12 Status monitoring means 13 Diagnostic means 14 Corresponding treatment determining means 15 Equipment control means 16 Input / output processing means 17 Knowledge database construction means 18 Knowledge Database 19 Plant signal data storage means 20 Predictive warning data storage means

Claims (2)

[Claims] 1. A signal input means for taking in an online alarm signal for notifying a plant abnormality, a status signal indicating the state of equipment constituting the plant, a process signal from the plant, and an attribute of the alarm meaning of the alarm signal. , The causal relationship divided for each alarm generation condition, knowledge about processing at the time of plant abnormality, and the start condition of equipment that constitutes the plant,
A knowledge database constructing means for constructing / saving a control result, a causal relationship associating the operation of the equipment with the plant state, and knowledge about the magnitude of the effect of the operation of the equipment on the target plant state as a knowledge database; and the signal input means. State monitoring means for determining the presence or absence of an abnormality in the plant based on the signal captured by the knowledge and the knowledge stored in the knowledge database, and the determination result is stored as plant signal data together with each signal, by the state monitoring means When it is determined that the plant abnormality, using the knowledge stored in the knowledge database,
Identification of the alarm progress path for each abnormal event that occurred, determination of the cause alarm, and performing sequence operation monitoring, diagnostic means for determining the state of the plant to cope with the abnormality, and the plant state obtained by the diagnostic means, and Using the knowledge stored in the knowledge database, a corresponding action determining means for determining a device to be operated, a control command therefor, and a treatment guide to be provided to an operator; and a corresponding action determining means to be operated. A device control means for sending and controlling a control command to the controlled device, a diagnosis result obtained by the diagnosis means, a determination result obtained by the state monitoring means, and a treatment guide determined by the corresponding treatment determination means. To automatically provide the operator with a request, and the knowledge database construction means Characterized by comprising input and output processing means for performing recording and editing process, warning processing support device in the plant. 2. The knowledge database construction means is a three-dimensional hierarchical space based on the degree of influence of the knowledge about the alarm in the knowledge database on the plant by the occurrence / non-occurrence of the alarm and the functional meaning indicated by the alarm. The alarm processing support device according to claim 1, wherein the alarm processing support device is expanded and stored on the upper side.