JPS6377537A - Plant operation supporting device - Google Patents

Abstract

PURPOSE:To prevent troubles of a plant in advance by providing a process calculating mechanism calculating the material balance and the heat balance of the plant, a presuming mechanism of causes and places of abnormalities, and data transmitting mechanism connected with such mechanisms. CONSTITUTION:A process input and output control mechanism 2 consists of apparatus, instrumentations, process computers, etc., of the plant 1 to be operated. An interface mechanism 3 is to connect the process calculating mechanism 4 with the process input and output control mechanism 2. The process calculating mechanism 4 carries out process data disposal, namely, disposes measured data in the apparatus to be operated, and carries out the calculations based on the inclinations for changes, widths of fluctuations, material balances, heat balances, etc. The data transmitting mechanisms 6, 7 connect the process calculating mechanism 4 to the presuming mechanism 8, and carry out data communication by DMA method.

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention detects a sign of an abnormality in plant operation, estimates the cause of the abnormality and the location of the abnormality, and notifies the plant operator of the results by voice and image. The present invention relates to a plant operation support device.

[Prior Art] A general plant is usually equipped with a control mechanism and a monitoring mechanism, and is normally controlled to return to a normal state even if an abnormality occurs. In addition, if the abnormality progresses beyond a certain level despite the control operation, a lamp display or an alarm sound will notify you of the abnormality. For example, in JP-A-61-125611, this setting is automatically performed in order to save labor in setting alarm limit values used to determine whether input signal values notified from each part of the plant are normal. , a plant monitoring device is disclosed that issues an alarm when this limit value is exceeded.

When an alarm is issued, a plant operator checks the process values displayed on the instrumentation panel or records on a recorder to understand the abnormal state. Then, the cause of the abnormality and the location of the abnormality are estimated, and countermeasures are considered. These operations require - in-depth knowledge of the entire plant and experiential knowledge of the equipment in question, and it is only on the basis of this knowledge and insight that appropriate decisions can be made.

[Problems to be Solved by the Invention] However, in current plants, it is becoming increasingly difficult to make calm and accurate decisions when an abnormality occurs. The reasons are: 1) The plants themselves are becoming more complex.

2) As the automation of plant operation progresses and the number of abnormal situations is decreasing, there are fewer young operators who experience abnormal situations. etc. are possible.

Therefore, after an alarm is issued, it is no longer easy to grasp the abnormal situation, investigate the cause, and consider countermeasures.

If the abnormality is serious, an even more urgent response is required, but it is often difficult to respond calmly and appropriately. Furthermore, it is impossible for operators to keep an eye on the plant and not miss any abnormalities.

The present invention has been made in view of the problems in the conventional example described above, and its purpose is to constantly monitor processes using a computer and to infer whether or not there is an abnormality and the cause of the abnormality when it occurs. And, by notifying the operator of the results by voice and image, the operator's load is reduced, product loss due to abnormalities is reduced, and plant accidents are prevented.

[Means and effects for solving the problem] The plant operation support device according to the present invention includes a process calculation mechanism that constantly monitors the brand condition and calculates the material balance and heat balance of the process, and calculates steady values from the results. Knowledge that stores rules for detecting deviations and abnormal changes, recognizing plant abnormalities based on the experiential knowledge of experienced operators and scientific knowledge about processes, and inferring the cause and location of abnormalities. The invention is characterized by comprising a base, an inference mechanism that searches and processes the knowledge base and performs inference, an image display mechanism connected to the inference mechanism, and an audio output mechanism.

Preferably, the process calculation mechanism and the inference mechanism are connected by a method of direct memory access, so that the process calculation and inference processing are not interrupted for the transmission of calculation results.

In the present invention, the process calculation mechanism calculates the rate of change and fluctuation range, and calculates the material balance and heat balance based on process data input from the process input control device. These results are sent to the reasoning system 1, which determines whether or not there is an abnormality, infers the cause and location of the abnormality, and decides on countermeasures. Furthermore, since these inference results are notified to the operator in the form of images and sounds, the operator can immediately know the cause and countermeasures without having to make detailed examinations and inferences about the abnormality.

[Embodiments of the invention] Hereinafter, the present invention will be explained in detail using the drawings.

FIG. 1 shows a schematic configuration of a plant operation support device according to an embodiment of the present invention. In the figure, 1 is a plant whose operation is supported by the apparatus of this embodiment, 2 is a process input/output control mechanism, and 3 is an interface (IF) R structure. In this embodiment, the process input/output control mechanism 2 is a process computer that controls the instrumentation equipment and equipment of the plant 1 to be operated, and the process of (.Furthermore, the !F mechanism 3 is a process input/output control mechanism. It connects the mechanism 2 and the process calculation 1 mechanism 4, and here the BSC
Approximately 80 procedures are used. Note that, as the data transmission method for connecting the FW structure 3, other standardized methods such as the TTY method, the basic procedure, the HD 10 procedure, or the GP-IB method can be used.

4 is a process calculation mechanism that exclusively processes process data, that is, processes measured values from the equipment to be operated and separate equipment, and calculates calculations based on scientific grounds such as change trends, fluctuation ranges, material balance, heat balance, etc. It is a calculator that performs. 5 is a display device that displays measured values and calculation results. The process controller 4 and the display device 5 do not need to be special devices; in this embodiment, a small general-purpose computer equipped with a CRT (cathode ray tube) display device is used.

6 and 7 are data transmission mechanisms, which connect the computers of both the process calculation mechanism 4 and the inference mechanism 8;
Data communication is performed using the (direct memory access) method. Note that the program transfer method may be used instead of the DMA method, but in that case, the calculation in the process calculation mechanism 4 and the inference in the inference machine v48 are interrupted during data transfer, so it is difficult to make a timely and accurate diagnosis. Sometimes I can't get it down. According to the DMA method, data can be transferred between the two storage devices regardless of calculation in the process calculation mechanism 4 and inference processing in the inference mechanism 8.

The inference i structure 8 is a computer that performs inference processing exclusively using the rules stored in the knowledge base 12, and 9 is a display device for CR.
It is a T display device. In this embodiment, an integrated small-sized general-purpose computer is used as the inference mechanism 8, display device 9, and knowledge base 12. The knowledge base 12 is one in which rules for inference are stored in a magnetic disk device which is an auxiliary storage device of the small-sized general-purpose computer.

10 is a speech synthesis device which receives the character string resulting from the inference in the inference mechanism 8 and synthesizes natural language speech. In this example, since the speech is obtained using the regular speech synthesis method, it is possible to synthesize speech using any character string using the inference system, which is versatile, and it is convenient because it synthesizes sounds close to human voices. be.

Reference numeral 11 denotes a speaker that outputs the sound synthesized by the speech synthesizer. The speech synthesizer 10 is connected to the reasoning mechanism 8 via an R8232C interface.

The inference mechanism 8 outputs the inference result to the speech synthesis device 10 as a character string. The following is an example of such a string.

"Feedstock oil storage tank level decrease" In addition to the above-mentioned inference results regarding abnormalities, the inference mechanism 8 also outputs countermeasures for the abnormalities. and,
This countermeasure is also output in the form of audio. The following is an example.

``Check G1 with the level gauge, warm up the raw material, and then replenish.'' Next, an overview of the oil refinery plant that is the target of operation in this example will be explained with reference to the flow shown in Figure 2.The process in the figure is as follows. , heavy oil is cracked using a catalyst to obtain light oil such as gasoline or naphtha.

Heavy oil, which is a raw material, is preheated to a predetermined temperature from a raw material oil storage tank (unit 1), and then sent at a constant flow rate by a pump to a riser pipe (unit 3) where it is atomized in a part of the mixer.

Then, it mixes with the high-temperature catalyst sent from the regeneration tower (unit 5) and rises while reacting inside the riser pipe.

Thereafter, it is further cracked in a reactor (unit 4), and the oil is sent to a distillation column (unit 6) where it is fractionated into cracked gas, gasoline/light gas oil, and heavy gas oil. On the other hand, the catalyst with coke attached is separated from oil in a reactor and sent to a regeneration tower via a catalyst level adjustment valve (valve 1), where it is combusted with air and regenerated. The regenerated catalyst is sent again to the mixer part of the riser tube at a predetermined flow rate and used for the reaction. The generated cracked gases, combustion gases, and produced liquids are automatically added and their compositions are analyzed using an attached analyzer. The analysis data J3 and operation data are sequentially input to the process combicoater, so that the material balance, etc. of the entire plant can be determined. A control system based on digital control (DDC) is used for plant operation. Furthermore, the reaction temperature, regeneration temperature, feedstock oil flow 9, reaction pressure, and coke suspension on the regenerated catalyst can all be controlled arbitrarily, and the setting conditions can be selected.

Next, the operation of the apparatus of this embodiment will be explained.

3 to 6 show the display device 9 (the first
An example of the display screen in Figure) is shown. FIG. 3 is a display screen when the plant shown in FIG. 2 is operating normally. "Normal" is displayed in the lower left corner, and a diagram corresponding to the flow of the plant is also displayed on the screen.

In FIG. 1, while the plant is in operation, the process input/output control mechanism 2 always inputs each measured value from the plant and sends the data to the process calculation mechanism 4 via the IFII structure 3. After performing predetermined calculations, the process calculation mechanism 4 sends the data to the inference mechanism 8 via the data transmission mechanisms 6 and 7.

The inference mechanism 8 constantly performs inference processing based on these data, and the results are displayed on the display device 9 and the speech synthesis device 1.
Output to 0. Therefore, during normal operation, the screen display as shown in FIG. 3 above continues (there is no audio output during normal operation), and when an abnormality is detected, a display and audio output indicating the abnormality are immediately made. Even after an abnormality is detected, data acquisition from the process and inference processing continue, so the display device 9 and the speech synthesizer 10 always show the current state of the plant.

The reasoning logic is expressed by a program built into the reasoning mechanism 8. This program is called a production system, and consists of production rules that express causal relationships and an inference system that processes them and makes inferences.

A production rule is a causal relationship between events expressed in the following format.

Conditional part -> Execution part Here, the conditional part consists of a statement expressing the state of the process and a logical value indicating whether the statement is correct or not. The value is T (true) if it is correct, and F (false) if it is incorrect. The execution part represents a hypothesis supported by the establishment of the condition part.

The inference mechanism 8 determines whether the hypothesis of the execution section holds true based on the truth of the condition section based on the data from the process calculation mechanism 4.

Next, the operation when an abnormality occurs somewhere in the plant will be explained.

Abnormality detection example 1 For example, in the plant shown in Figure 2, the catalyst moves from the reactor (unit 4) to the regeneration tower (unit 5) under its own weight, but the pressure in the regeneration tower is higher than that in the reactor, so pressure control is very delicate. is necessary. Therefore, an abnormality in the control valve 1 is detected by paying attention to the gas flow rate, which is closely related to the pressures of the reactor and regeneration tower. If we first focus on the gas flow rate, the following rules can be considered.

F (F7NOWV, 0.8: *) & F (F7M
OVE, -0,1: 0.1)->F (to F2O,
T> This rule is stored in the knowledge base 12.

What this means is that if F7NOWV is 0.8 or more and F7MOVE is between -0.1 and 0.1, F7NOWV is
70K is true. Here F7NOW
V is 1 of the data sent from the process calculation mechanism 4
and the gas flow m trained at position F7 in FIG.
It is. F7MOVE is the width of the gas flow rate fluctuation at F7, which is similarly calculated by the process calculation mechanism 4. In other words, it is the difference between the maximum value and minimum value over a certain period of time in the past. In other words, this rule states that if the gas flow rate at F7 is above a specified value and there is not much variation, then F7 is normal.

Next, suppose that an abnormality is found in the differential pressure gauge DP3 (Fig. 2) between the reactor (Unit 4) and the regeneration tower (Unit 5), F (D3H1, T) & F (F2O, T) -> )−
The rule 1(DPRVLV, 0.7> is applied, which means that D 3 HI is true and F70K
If is true, it means that the hypothesis DPRVLV holds true with a confidence level of 0.7. The degree of confidence indicates the probability of the hypothesis. In other words, if the hypothesis absolutely holds true, 1 (T); if it absolutely does not hold, -
1, and taking a value between them indicates the possibility of the hypothesis being established.

D3HI becomes true because the value of the differential pressure gauge DP3 is abnormally high and an alarm has been generated, and the status of the alarm is sent to the inference mechanism 8 through process calculation 1!4. F2O is supported by the previous rules. When the hypothesis DPRVLV is established in this way, the hypothesis is similarly associated with the next sentence in the knowledge base.

“'DPRV-3 abnormality” As a result, this character string is sent from the inference mechanism 8 to the speech synthesizer 1.
Sent to 0. Then, from the speech synthesizer 10, the speaker 11 allows you to listen to the music.
is pronounced. At the same time, the bulb 1 is displayed in red on the display screen 9.

Abnormality Detection Example 2 In FIG. 2, DPR-4 indicates "rFLUE gas filter differential pressure". By paying attention to this differential pressure DPR-4, the condition of the filter unit (unit 2) can be estimated. First, the following rules are stored in the knowledge base 12.

F (D4MOVE, 20:*) ->F (D4UP, T) This rule indicates that the 1-minute trend of differential pressure DPR-4 is +
20m1'n1-(2o or more indicates that the differential pressure DPR-4 is considered to be at an upper limit.Furthermore, the following rules regarding the blockage of each line are stored in the knowledge base 12.

F(D4UP, T)->H(C3FILT, 0.7
)F(D4H1,T)->H(CLFTLT, 0.
7>, H (C3F ILT, -1) (1: F (D4UP, T), F (D4H1, T)
) -> H (ZF I LCH, 0, 7) Here, C3FILT indicates that the filter unit is becoming clogged, CLF I LT indicates that the filter unit is blocked, and ZFILCH indicates that the filter needs to be replaced and regenerated. Each indicates something.

Therefore, the above rule is: If the differential pressure DPR-4 is rising (D4UP),
If the degree of confidence that the filter unit is clogged (C3F I LT) is 0.7 and the differential pressure DPR-4 high alarm (exceeds the predetermined upper limit) is issued (D4
If one of the following conditions holds true: "rDPR-4 is elevated" or "rDPR-4 is high alarm", the degree of confidence that the condition is occlusion (CLFI LT) is 0.7, rather than occlusion (HI), then filter switching and regeneration is performed. This indicates that the level of confidence that it is necessary to take the corresponding action (ZFILCH) is 0.7.

When these rules are applied, part of the filter in Figure 2 (
Unit 2) is displayed in red, and the above inference results are output together with audio.

Example of double-sided display FIGS. 4 and 5 show double-sided display when the inference mechanism 8 reports an abnormality. The display "Normal" in the lower left has changed to "Abnormal". Further, at the top of the screen, two sentences each indicating what kind of abnormality has occurred as a result of the inference and countermeasures are displayed. For example, in Figure 4, rLRC3
Control nitrogen is low. '' has a confidence level of 0.70, and the PID value of rLRC3 seems inappropriate. ” indicates that the anomaly occurs with a confidence level of 0.60. Correspondingly, the portion indicated by number 21 is displayed in red, and the portion indicated by number 22 is displayed in yellow. Furthermore, two countermeasures that should be taken against this abnormality are displayed along with the degree of confidence that the countermeasures should be taken.

The same is true in FIG. 5, and the part numbered 23 is displayed in red corresponding to the detected abnormality.

Note that as a result of the inference in the inference mechanism 8, a plurality of causes and countermeasures for the abnormality are listed, but in this embodiment, for convenience, two causes of the abnormality and two countermeasures are output from the one with the highest degree of certainty. Therefore, these outputs may be one or three or more.

Abnormality Detection Example 3 In FIG. 2, LP01 indicates the level of raw oil in the raw oil measuring pipe. The knowledge base 12 stores the following rules. It is assumed that L2M0VE represents a 1-minute trend of level LR8-2.

F (12MOVE, -0,05: 0,05
)->F (12NC, T) F (L2MOVE, *: 0) ->F (L2DN, T) The above rule indicates that 12M○VE is -0.05 to 0.05%
If so, LR8-2 is 12NC) without any change, L
2MOVE is over 0% T4-atl [LR8-2 is decreased (
L2DN).

Additionally, the following rules are stored.

*■F F　(P4.T)& (1:F (L2DN, T).

F (12NC, T)) *THEN F (L2.T) ->H(ARLUCK, -0,8).

H (ZARXCG, 0.8).

H (P4AOVE, 0.3).

H (ZAOVCK, 0.3) *END This means that if feedstock oil supply pump P-4 is activated and LR8-2 remains unchanged or decreases,
Normally valve LR3V-2 is closed, but if it is open (F (L2゜T)->), the raw material oil is decreasing <APLLICK), and the raw material [
△R1 must be warmed up before replenishment (ZARX
CG) and the degree of confidence is 0.8. Or, LR8V-2 valve is abnormal (P4AOVE) and A
This rule indicates that the degree of confidence that an OV check is necessary (ZAOVCK) is 0.3.

By applying this rule, the above-mentioned inference results are output together with audio, and abnormalities are displayed in red and yellow in order of suspicion.

FIG. 6 shows the display screen at this time. The raw oil storage tank (number 24 in the figure), which is thought to be the most likely cause of the abnormality, is displayed in red, and the valve of LR8V-2 (number 25 in the figure), which is the second suspect, is displayed in yellow.

In this way, the plant is constantly monitored by the device of this embodiment, and if an abnormality occurs, the cause and location of the abnormality are immediately estimated, and countermeasures are displayed with images and sounds, reducing the burden on plant operators. significantly reduced.

[Effects j] As explained above, according to the present invention, the plant status is constantly monitored, the material balance and heat balance of the process are calculated, abnormal changes are detected from the results, and knowledge-based rules are applied. Based on this, the cause of the abnormality, the location of the abnormality, and countermeasures are inferred, and the results are outputted externally in the form of images and sounds, which homogenizes the operation of plant equipment and has the following effects.

1) Product loss due to operational errors is prevented.

2) Operation can be entrusted to inexpensive operators (unskilled).

3) Since switching of raw materials, change of operation mode, etc. are performed smoothly, off-spec products can be prevented and irregular periods can be shortened.

4) Confusion of operators in the event of an abnormality is prevented.

5) Abnormalities are dealt with accurately and in a timely manner.

6) Even small abnormalities are not overlooked and are dealt with appropriately.

[Brief explanation of the drawing]

FIG. 1 is a schematic configuration diagram of a plant operation support system according to an embodiment of the present invention, and FIG. 3 to 6 are examples of display screens in the apparatus of the above embodiment. DESCRIPTION OF SYMBOLS 1... Plant, 2... Process input/output control mechanism, 3... IF mechanism, 4... Process calculation mechanism, 5.9... Display device, 6, 7... Data transmission structure , 8... Reasoning mechanism, 10... Speech synthesis device, 1
1...Speaker, 12...So-called base.

Claims (1)

[Scope of Claims] 1. Receives data from a process input control device through an interface mechanism for importing data from the plant, processes the data, and obtains temporal changes while at the same time a process calculation mechanism that calculates material balance and heat balance; a knowledge base that stores knowledge for estimating the cause of an abnormality and the location where the abnormality occurs;
an inference mechanism that performs inference using the knowledge; a data transmission mechanism that connects the process calculation mechanism and the inference mechanism; an image display mechanism that presents the inference results of the inference mechanism to humans; and/or an image display mechanism that displays the inference results in natural language. 1. A plant operation support device characterized by comprising a voice output mechanism for reporting. 2. The plant operation support device according to claim 1, wherein the data transmission mechanism is a communication control mechanism for connecting the process calculation mechanism and the inference mechanism to each other in a direct storage access method. 3. Claim 1, wherein the voice output mechanism is connected to the inference mechanism through a transmission line, receives a character string expressed in a natural language from the inference mechanism, and synthesizes voice using a regular voice synthesis method. Or the plant operation support device according to item 2.