JPS58137098A - Apparatus for avoiding plant abnormal state - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention is characterized by being capable of detecting abnormal conditions in a conventional plant before the normal warning level, and making it possible to predict the progress of abnormalities in a plant while correlating these abnormal conditions. The main focus is not only on fixing the causes of plant abnormalities in real time while the plant abnormality is progressing, but also on cases where the cause cannot be identified, it is possible to estimate possible causes and avoid abnormal conditions. This relates to wrinkles.

Conventionally, as a device for this embroidery, there was a proper plant abnormality diagnosis system shown in Figure @1, but in Figure 1, (1
) is a plant, (2) is a device in the plant whose MF will reach the plant (1) if it becomes abnormal,
(3) is a detector that detects some kind of abnormality (e.g. high temperature, low pressure, etc.) in the device shown in (2), and (4) is a detector (
3) is a process signal coming out from , (5) is a signal processing converter that converts the Brosse signal, and (6) is a signal processing converter (5).
Detector (7) is located in the apparatus to which the detector (3) belongs and detects process wax that has some causal relationship with the process wrinkles detected in (3); (8)
is the plant diagnosis logic (9) and its display system (
(9) is the plant diagnosis logic in the computer application software, and σQ is the signal processing converter (5).
) is a limit check software circuit using computer application software that outputs the process signal from the plant at an earlier level (safety side level of the plant) than the plant normal alarm level, and α9 is the output signal (binary value) from the limit check software circuit QO. signal), (2) is a start command for the plant diagnosis logic (9), and (to) is a point detected as a warning in the plant diagnosis logic when the output signal Qυ from the limit check software circuit turns ON. (hereinafter referred to as a node), α◆ is a node detected in the plant diagnosis logic (9) like the node (to), but is a node caused by the detector (7) of the plant (1), (to) ) is a logic gate that indicates ND, a ring is a logic gate that indicates OR, and the target is one of the various possible cases in plant (1) from a lower node to an upper node where the shadow extends. The minimum time delay O゛), (to) is after node (to) is detected as a warning and it is determined that the cause is node α→, after avoiding the abnormal state indicated to the operator, or after the other end of T , the node that will be detected later, and (e) are the nodes in the final state where the abnormality spreads.

Next, the operation with respect to FIG. 1 will be explained.

When the plant (1) becomes abnormal, the process quantities detected from the equipment (2) of the abnormal part @ (3) and (7) pass through the signal processing converter (5) and are sent to the plant diagnosis system. Inside the computer (8), it is converted into a process signal (6) convenient for the operation of the limit check software circuit (10) which is set to a level below the alarm setting level. When the ON signal αυ appears, the plant diagnostic logic (9) is activated by the @Nyoru activation command, and then the ON signal (B) will be detected in the node σ field. It has been investigated in advance that the abnormality in the detected detector (3) is due to the influence from the detector (7) or other sources in the plant, and the causal relationship is determined by node α◆ and node () in the plant diagnosis logic (9). On the other hand, if node Q4 is also detected before the alarm level is detected at the time when it is detected at node Q4 and the plant diagnosis logic (9) is activated, node The cause of the situation in which node α was detected is that the situation in which node α was detected is the cause of the abnormality in detector (3). It turns out that there is an abnormality.At that point (
When the cause of the abnormality is identified, an abnormal situation avoidance operation message (
) will be announced.

Also, node 4 is detected and the plant diagnosis logic (9)
is activated, the next possible state of the plant is predicted by inferring the upper level of the logic, and the operator selects the node α of the final state of the plant.
It is now possible to take precautionary measures to prevent this from happening.

In addition, due to the time delay (T,) when reaching node (2) to node 09, the time delay (T3) sets the minimum possible time in plant (1), so node (to) If node α is detected after a time shorter than (T,) has passed since it was detected, it can be determined that there is a failure in the main body of the detector on the plant (1) side that detects node α. sell.

Moreover, if this plant diagnosis logic (9) is displayed as is on the CRT as a means of conveying these cause identification and future predictions to the operators, the detectors t3), By adding the abnormality II amount in (7) and adding the operation message (g), the operator can perform operations to avoid the abnormality line while grasping the plant status.

The conventional plant abnormality diagnosis logic device has the above configuration, and in this device, the plant diagnosis logic (
9) is started, node α→ is detected and the cause of the abnormality is found to be an abnormality in the detector (7). In order for an abnormal situation avoidance operation message to be issued, the node corresponding to the cause of the abnormality (in the above case, The condition is that node 04) is detectable from within the plant (1), that is, it is an observable node. However, in the current plant, it is rare that it is possible to measure up to the node corresponding to the cause (although the node that detects the abnormality (node (2) in the above) is an observable node). In other words, there are many cases in which there are no detectors in the plant (1) circle or those that cannot be set, so even if an abnormality is detected by the node that detects the abnormality, the cause of the abnormality cannot be determined. When searching for the corresponding node (observable node), there was a drawback that in most cases, the cause was not found and the cause was considered to be a failure.

This invention was made in order to eliminate the above-mentioned drawbacks of the conventional system, and it packs sub-messages in addition to abnormal state avoidance operation messages attached to each node in the plant diagnosis logic (9) in the conventional system. This is a system that analyzes the progress of abnormalities in a plant and performs so-called abnormal event analysis.
By creating Fault 'Free in advance and outputting its content corresponding to 8 non-1@i measurement nodes as the content of this sub-message, any cause can be identified and the abnormality class must be determined. The purpose of the present invention is to provide a plant abnormal state avoidance device that can avoid such problems.

An embodiment of this invention will be described below with reference to the drawings. @2
In the figure, (1) to (A) represent the same functions as in Figure 1, among which (3), (7) and 0. It is assumed that α-curvature is observable information. (c)
is an abnormality in a part of the device (2) that would affect the plant (1) if it becomes abnormal, but it cannot be detected (in other words, it cannot be observed because a detector cannot be installed or a detection binding has been developed, etc.) What! I! Point ii) Specifically, this is the point where the abnormality extends to the abnormality of the detector (3). (2) cannot be formally expressed as a logic in plant diagnosis logic (9) because (c) is an unobservable point, but it is a virtual logic that can be considered by assuming the path that the form will take. . (2) is the assumed point (
(goods) corresponding to virtual empty logic) is a virtual empty node on ). (Incorporated) is a similar node (towards ) is a message to avoid foreign state logic.

Next, to explain the operation, the operations from (1) to Kan are the same as those of the conventional plant abnormality N diagnostic logic device, so they will be omitted here.After an abnormality is detected at node (2), Suppose that when a node below the node (to) of the plant [Fr logic (9)] was searched for as the cause of the abnormality, no abnormality was detected at the node next to it (the cause of the failure). What is the cause of the abnormality detected at the node (to)? Lio chic (9) In terms of circles, it can be said to be an abnormality from a virtual empty node in a virtual empty logic. However, such virtual empty logic (2) cannot actually be expressed as logic in the plant diagnosis logic (9) because the node @ is unobservable. In order to cover this, in node (2), an abnormal state IJJ expressing the content of this virtual empty logic [il operation avoidance message (c) is provided separately from the abnormal state avoidance operation message O1, and the cause This was an attempt to avoid plant abnormalities by notifying the distribution authorities when identification failed.

Furthermore, the expression content of this abnormal state logic avoidance message (c) is when the cause of the abnormality is thought to be an unobserved lead (for example, to a node). Based on experience, the assumed state of the IA boar cause is the unobservable thought of the plant (1).
), it is made into intersection lines, and appropriate operations are also included.

Although this invention was realized with 8/W, it would also be possible to assemble it with a hardwired one.

As described above, according to the present invention, the plant diagnosis logic (9) of the cause node (for example, node α rumor) does not have observable features that undermine the conventional plant abnormality diagnosis logic device.
C also has an unobserved causal node (e.g. node C4)
Even when it is thought that the cause is caused by the problem, the conventional plant reference logic (e.g. (9) in Figure 1) should have an abnormal condition 1 logic avoidance message (e.g. (goods)) as a sub-message. Since the configuration is such that the possible causes of an abnormality can always be estimated by using only the above information, it is possible to effectively identify all the causes of abnormalities by 115M with just a small change of adding submessages to the conventional one, which is a great effect.

[Brief explanation of the drawing]

FIG. 1 is a configuration diagram showing a conventional plant abnormality diagnosis logic device, and FIG. 162 is a configuration diagram showing an embodiment of the plant abnormality avoidance device of the present invention. In the figure, (1)...plant, (2)...equipment in the plant, (3)...detector, (4)...process signal output from detection 5 (3), (5)... ... Signal processing converter, (6) ... Process signal from signal processing converter (5), (7) ... Detector, (8) ... Plant diagnostic logic computer, (9) ...Plant diagnosis logic, (IQ...Limit check sost circuit,
(b)...binary signal of ON or OFF, sub...startup command for plant diagnosis logic (9), cL1 wealth detector (
3), Q4... Node caused by detector (7), (G)... AND is a St logic gate, So... logic gate indicating OR, α force... Minimum time Delay n(T), (goods)... Abnormal state avoidance operation message, σt... node, ■... Final state 7-do, 2
9...Unobserved assumed point, Iwa...Temporary empty a-sick, (2)...Temporary empty node, (to)...Abnormal state logic avoidance message. Note that the same reference numerals in the figures refer to the same or corresponding parts. Agent Shin Kuzuno − □

Claims (1)

[Claims] When a plant is about to fall into an abnormal state, when even the slightest sign of this appears, it can be detected at a state before the plant alarm level and compared with the logical tree created by surveying the plant in advance. In addition to identifying the cause of an abnormal state and predicting its usual effects, even if the cause is an abnormality that is not expressed in the diagnostic logic, the abnormality can be diagnosed in a wider range than the diagnostic logic based on the above logic tree. Using a fault propagation configuration tree based on time line analysis,
This is a plant abnormal situation avoidance device that outputs this information to fix any causes and ensure that abnormal situations can be avoided.