JPH0512724B2 - - Google Patents

Description

[Detailed description of the invention]

[Technical Field of the Invention] The present invention relates to a plant abnormality diagnosis device for diagnosing abnormality in a plant. [Technical background of the invention and its problems] A conventional example of a plant abnormality diagnosis device will be described with reference to the block diagram shown in FIG. In the figure, reference numeral 1 denotes a signal input device for inputting plant process signals, and the value of this input signal is compared with a preset threshold value in a signal condition processing device 2 to determine the state of the signal. The result of the state processing of the signal processed by the signal state processing device 2 is stored in the signal state storage device 3. Further, 4 is a storage device for a cause-and-effect tree created in advance. The abnormality diagnosis processing device 5 identifies the cause of the abnormality and predicts event propagation based on the state of the input signal output from the signal state storage device 3 and the cause-and-effect tree in the cause-and-effect tree storage device 4. The diagnosis results are displayed on the display device 6. Next, the operation of the abnormality diagnosis apparatus shown in FIG. 3 will be explained with reference to a cause-and-effect tree, an example of which is shown in FIG. In the cause-and-effect tree in Figure 4, M
7, M13 is a diagnostic message, G ₁ , G ₂ , G ₄ , ₅ are AND gates, G ₃ , G ₆ are OR gates, N ₁ ~
_N13 is a node representing the state of the signal, and this state is stored in the signal state storage device 3. The cause-and-effect tree execution processing is performed by the abnormality diagnosis processing device 5 based on information from the signal state storage device 3 and the cause-and-effect tree storage device 4, and the following processing is performed to identify the cause of the abnormality. Ru. First, for a node in which a signal whose state has changed is used, it is checked whether a condition for that node to be true holds true. For example, in Figure 4, if node N ₇ becomes true, the condition for node N ₇ to become true is that one of N ₅ , N ₁ , N ₂ N ₆ , N ₃ , and N ₄ holds true. That's true. The set that is true in this way will be called the minimum cassette set. Also, for node N ₁₁ to be true, N ₈ · N ₇ · (N ₅ · N ₁ · N ₂ + N ₆ · N ₃ · N ₄ ) =
N ₈ , N ₇ , N ₅ , N ₁ , N ₂ + N ₈ , N ₇ , N ₆ , N ₃ , N ₄ must hold. At this time, there are two minimum cut sets: N ₈ , N ₇ , N ₅ , N ₁ , N ₂ N ₈ , N ₇ , N ₆ , N ₃ , and N ₄ . Therefore, when the state of a signal used by a certain node changes, the upstream side of that node (the fourth
In the figure, the state of all nodes located at the bottom is upstream and the top is downstream) is investigated, and it is determined whether any minimum cut set is established.
If a certain minimum cut set is established, it is assumed that the cause identification has been successful, and if there is a diagnostic message for that node, that diagnostic message is displayed. For a node for which the minimum cut set has been established, the downstream logic of that node is traced to predict the progress of the event. That is, in FIG. 4, node N ₇ is true, and N ₇
If a minimum cut set is established with N7 as the topmost node, the state of node _N11 located one downstream of _N7 is checked. At this time, since gate G ₄ is an AND gate, both N ₇ and N ₈ must be true in order for N ₁₁ to hold true. Now, assuming that both N ₇ and N ₈ are true, if N ₁₁ is false, N ₁₁ becomes the predicted event, and if N ₁₁ is true, we examine the further downstream node N ₁₃ state, and N ₁₃ is false, N ₁₃ becomes N ₇
This will be a predicted event when it is established, and the corresponding diagnostic message will be displayed. Since conventional diagnosis using a cause-effect tree involves execution processing as described above, it is essential to establish a minimum cut-set only in identifying the cause of an abnormality. And since the minimum cut set is composed of all the nodes upstream of the node of interest, as the event progresses, the amount of minimum cut sets with the node of interest at the top becomes enormous. Therefore, there is a problem that the analysis time increases. Furthermore, if the order of changes in events becomes different due to time leakage of events or the way thresholds are given, it is assumed that the minimum cut/set will not be established and the identification of the cause of the abnormality will fail. [Object of the invention] The present invention has been made in view of the above circumstances, and
The purpose is to provide a plant abnormality diagnosis device that quickly analyzes plant abnormalities through more flexible diagnosis and provides the diagnostic information to operators to prevent the spread of the abnormality. [Summary of the Invention] In order to achieve the above object, a plant abnormality diagnosis device of the present invention includes a signal state processing device that processes the state of an input signal input from the plant, and a signal state processing device that stores the state processing result of the input signal. A signal state storage device, a cause-and-effect tree storage device that pre-describes the propagation sequence of abnormal events that occur in a plant in the form of a logical tree, and the scores of nodes on the cause-and-effect tree, if the upstream gate is an AND gate, The score S _i of that node is Here, S _j is the score of the node upstream of the AND gate, T _j is the certainty of the signal used at the node upstream of the AND gate, and C _j is the
The score is determined by whether the node on the upstream side of the AND gate is true or false, and j is j = 1 to n.
The number of nodes on the upstream side of the AND gate. If the gate on the upstream side is an OR gate, the score S _i of that node is S _i = MAX (S _j1 *T _j1 *C _j1 , S _j2 *T _j2 *C _j2 ,
...) Here, let S _i be the maximum score among the scores (S _j *T _j *C _j ) for nodes on the upstream side of the OR gate, and calculate the maximum score for the node at the most upstream side of the tree (primary node). The score shall be uniformly given a certain reference point S ₀ , and it is calculated by , and is performed for all nodes in the initial setting, and thereafter, the score is stored only for nodes downstream of the node whose state has changed. Node state memory is used. and an abnormality diagnosis processing device that determines the truth or falsity of the node based on whether the score of the node is equal to or higher than the reference point when the state of the input signal corresponding to the node on the cause-effect tree changes. It is characterized by: [Embodiment of the Invention] An embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of an embodiment of the present invention. In the figure, a process signal is input from a plant (not shown) to a signal input device 7 at regular intervals. When the input signal is an analog signal, the value of the input signal is compared with a predetermined threshold value, and a change in state is recognized depending on whether the value of the input signal is outside the allowable range. Furthermore, if the input signal is a digital signal, it is checked whether there is a change in the on/off state. These processes are executed in the signal state processing device 8. This signal state processing device 8
The result of the state processing of the signal processed in is stored in the signal state storage device 9. The cause-and-effect tree storage device 10 also stores a cause-and-effect tree created in advance. Therefore, in the abnormality diagnosis processing device 11,
Using a change in the state of the input signal as a trigger, abnormality diagnosis is performed using the cause and effect tree stored in the cause and effect tree storage device. At this time, the score of the node on the cause-effect tree changes in accordance with the change in the state of the input signal, and is stored in the node state storage device 12, and analysis is performed while referring to this score. The diagnostic results obtained by the analysis are displayed on the diagnostic result display device 13. Next, as an example of the present invention, analysis of the cause-and-effect tree shown in FIG. 2 will be described in detail. Figure 2 is part of a cause-and-effect tree showing an event in which the water supply flow rate to the reactor increases in a BWR plant, where N ₁ to N ₁₇ are nodes, G ₁ , G ₂ , G ₄ , G ₅
indicates an AND gate, and G ₃ and G ₆ indicate an OR gate. Here, N ₁ to _{N 17} correspond to the signal states shown in Table 1, and N ₈ , N ₉ , N ₁₀ , and N ₁₆ represent unobservable signals that appear in the tree structure. An abnormality in the plant is determined by whether or not the AND gate (G ₁ , G ₂ , G ₅ ) at the most upstream is satisfied. In this example, the AND gate
High side failure of water supply main controller when G ₁ is established,
When G ₂ is established, the output is increased manually by the main water supply controller, and when G ₅ is established, the output is increased manually by the TD-A controller.

【table】

[Table] By the way, the following analysis method is used to analyze the cause-and-effect tree shown in FIG. 1 If the upstream gate is an AND gate, the score S ₁ of that node is calculated from the following equation (1). Here, S _j : Score of the node upstream of the AND gate T _j : Confidence of the signal used at the node upstream of the AND gate C _j : Whether the node upstream of the AND gate is true or false The score determined by j: j = 1 to n, n is the number of nodes on the upstream side of the AND gate 2 If the gate on the upstream side is an OR gate, the score S _i of that node is calculated from the following equation (2) . S _i =MAX(S _j1 *T _j1 *C _j1 , S _j2 *T _j2 *C _j2 ,
…) …(2) In other words, the maximum score S _j *T _j *C _j for the node on the upstream side of the OR gate is
Let S _i be 3. Assume that the score of the node at the most upstream side of the tree (primary node) is uniformly given a certain reference point S ₀ . Thus, the score calculation is performed for all nodes in the initial setting, and thereafter only for the nodes downstream of the node whose state has changed. Next, regarding the cause-and-effect tree shown in FIG. 2, the following assumptions are made as an example. That is, 1 S ₀ = 100 2 T _i = 1 (i: for all nodes) 3 C _i = 1 (when node N ₁ is true) C _i = 0.5 (when node N ₁ is false) 4 Not possible An observation node considers the state of the node to be true when its score is greater than or equal to the reference point.Now, in the initial settings, nodes N ₁ , N ₂ , N ₁₁
Assuming that is true, the score of each node will be as follows using equations (1) and (2) above. S ₈ = 75, S ₁₅ = 59.4 S ₉ = S ₁₆ = 50, S ₁₀ = 37.5, S ₁₇ = 29.7 Scores of other nodes S _i = 100 Here, assume that node N ₄ becomes true. At this time, the scores of nodes N ₈ , N ₁₀ , N ₁₅ , and N ₁₇ on the downstream side are calculated. The scores of these four nodes are respectively S ₈ = 87.5, S ₁₀ = 43.8, S ₁₅ = 71.9, and S ₁₇ =
It becomes 35.9. Similarly, if N ₃ is true, the score of node N ₈ , which represents the water main controller high-side failure, is 100.
becomes. Since this node _N8 is an unobservable node, if this score is higher than the reference point, _N8 is considered to be true. The above explanation is about when the primary node becomes true, but this also applies to nodes other than the primary node. For example, when node N ₁₅ becomes true, node N ₁₇ becomes true.
is recalculated. [Effects of the Invention] As explained above, according to the plant abnormality diagnosis device of the present invention, when analyzing a cause-effect tree, it is only necessary to refer to the information of the node that is one upstream of the node of interest. Compared to the method of examining the states of all nodes upstream, this method is expected to reduce analysis time, and is particularly effective when an event progresses downstream in the cause-and-effect tree.
Furthermore, even though the signal state of the node of interest is true due to a shift in the order of event progression, if the upstream node is not considered to be true because it is false, the upstream node will later become true. It is also possible to calculate the score of the downstream node, and if it exceeds the reference point, then consider that node to be true and display a diagnostic message. Therefore, the cause-and-effect tree has the function of correcting deviations in the order of progression of events, allowing for flexible analysis.
Furthermore, when a node becomes true, the score can indicate the certainty of the diagnosis, and it is also possible to change the content of the diagnostic message accordingly, which is an excellent effect not found in conventional technology. play.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an embodiment of the present invention.
Figure 2 is a diagram showing part of a cause-and-effect tree showing an event in which the water supply flow rate to the reactor increases in a BWR plant, Figure 3 is a block diagram of a conventional plant abnormality diagnosis device, and Figure 4 is a diagram showing the cause. FIG. 3 is a diagram showing a portion of a result tree. 7... Signal input device, 8... Signal state processing device, 9... Signal state storage device, 10... Cause-effect tree storage device, 11... Abnormality diagnosis processing device, 1
2...Node status storage device, 13...Diagnosis result display device.

Claims (1)

[Scope of Claims] 1. A signal state processing device that processes the state of an input signal input from a plant, a signal state storage device that stores the state processing result of the input signal, and a signal state storage device that stores the state processing result of the input signal, and a signal state processing device that processes the state of an input signal input from a plant. The cause-and-effect tree storage device described in advance in the form of a logical tree and the scores of nodes on the cause-and-effect tree are as follows: When the upstream gate is an AND gate, the score S _i of that node is Here, S _j is the score of the node upstream of the AND gate, T _j is the certainty of the signal used at the node upstream of the AND gate, and C _j is the
The score is determined by whether the node upstream of the AND gate is true or false, and j is j = 1 to n.
The number of nodes on the upstream side of the AND gate. If the gate on the upstream side is an OR gate, the score S _i of that node is S _i = MAX (S _j1 *T _j1 *C _j1 , S _j2 *T _j2 *C _j2 ,
...) Here, let S _i be the maximum score among the scores (S _j *T _j *C _j ) for nodes on the upstream side of the OR gate, and calculate the maximum score for the node at the most upstream side of the tree (primary node). The score shall be uniformly given a certain reference point S ₀ , and it is calculated by , and is performed for all nodes in the initial setting, and thereafter, the score is stored only for nodes downstream of the node whose state has changed. Node state memory is used. and an abnormality diagnosis processing device that determines the truth or falsity of the node based on whether the score of the node is equal to or higher than the reference point when the state of the input signal corresponding to the node on the cause-effect tree changes. A plant abnormality diagnosis device characterized by: