JP6815954B2 - Plant operating status estimation device and operating status estimation method - Google Patents

Description

The present invention relates to an operating state estimation device and an operating state estimation method for estimating the operating state of a plant using on-site know-how.

On-site know-how accumulated by skilled workers plays an important role in the stable operation of plants such as factories and equipment. Patent Document 1 stores operation data such as pH, turbidity, chemical injection amount for pH adjustment, and coagulant injection rate in water purification plant facilities, and operates data from multi-item form data and the like with a support vector machine or the like. A method of automatically extracting the law as know-how by grouping the above is disclosed. Further, Patent Document 2 discloses a method in which an expert inputs and registers the know-how of operation in a nuclear power plant into an apparatus.

Japanese Unexamined Patent Publication No. 2011-60135 Japanese Unexamined Patent Publication No. 2005-215314

In the method disclosed in Patent Document 1, if the judgment criteria are determined, the know-how that matches the judgment criteria can be extracted, but it is not possible to set the judgment criteria so that the know-how can be accurately extracted. Since it is difficult, it is conceivable that know-how that cannot be extracted may occur, or information that is not know-how may be extracted. Further, in the method disclosed in Patent Document 2, since the expert inputs the know-how into the device, only the know-how can be registered accurately, but the know-how that is difficult to express in words and the expert recognizes it as know-how. It is not possible to register unknown knowledge. Therefore, in the methods disclosed in Patent Documents 1 and 2, the extracted or registered information may be inappropriate, and the know-how cannot be sufficiently effectively utilized for the operation of the plant.

The present invention has been made in view of such circumstances, and a main object thereof is to provide an operating state estimation device and an operating state estimation method of a plant capable of solving the above problems.

In order to solve the above-mentioned problems, the plant operation state estimation device according to one aspect of the present invention inputs operation data for a predetermined target item of the plant, and is used for items different from the target items of the plant. An estimation means that inputs the operation data to an estimation model that outputs an estimated value of the operation state and acquires the estimated value output from the estimation model, and an evaluation of the estimated value acquired by the estimation means. Evaluation means received from the user, an operation record storage unit that stores the operation data of the plant, and an estimation result storage that stores an estimated value when the operation data stored in the operation record storage unit is input to the estimation model. The unit, the evaluation result storage unit that stores the evaluation of the estimated value stored in the estimation result storage unit, the operation data stored in the operation performance storage unit, and the estimation stored in the estimation result storage unit. It is provided with a model updating means for updating the estimated model based on the value and the evaluation stored in the evaluation result storage unit.

In this aspect, the estimate model includes adjustable parameters so that the model update means updates the estimate model by adjusting the parameters based on the operation data, the estimate and the evaluation. It may be configured.

Further, in the above aspect, the estimation model may include a statistical model obtained by a statistical analysis method.

Further, in the above aspect, the estimation model may include a physical model that mimics a physical phenomenon in the plant.

Further, in the above aspect, even if the model updating means is configured to set a weight on the operation data and update the estimation model based on the operation data weighted by the set weight. Good.

Further, in the above aspect, the plant is an electric furnace used for steelmaking, and the operation data includes the resistance value of the electrodes of the electric furnace, the raw material to be charged into the electric furnace, and the by-product gas generated by the electric furnace. The estimated value may be the length of the electrode, including information on each of the components and the temperature of the by-product gas.

Further, in the method of estimating the operating state of a plant according to another aspect of the present invention, a computer inputs operating data for a predetermined target item of the plant and estimates the operating state of an item different from the target item of the plant. A step of inputting the operation data into an estimation model having a value as an output and acquiring the estimation value output from the estimation model, and a step of receiving the evaluation of the acquired estimation value from the user by the computer . The computer has a step of updating the estimation model based on the past operation data of the plant, the estimated value when the operation data is input to the estimation model, and the evaluation of the estimation value.

According to the plant operating state estimation device and the operating state estimation method according to the present invention, the know-how can be effectively utilized as compared with the conventional case.

The schematic diagram which shows the structure of the operation state estimation system which concerns on embodiment. The schematic diagram which shows the outline of the structure of an electric furnace. The block diagram which shows the structure of the operation state estimation apparatus which concerns on embodiment. The flowchart which shows the procedure of the operation state estimation processing by the operation state estimation apparatus which concerns on embodiment. A flowchart showing the procedure of the model update process. A flowchart showing the procedure of the estimation process. A flowchart showing the procedure of estimation result output processing. A flowchart showing the procedure of the evaluation process. The figure which shows an example of the estimation result screen.

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings. It should be noted that each of the embodiments shown below exemplifies a method and an apparatus for embodying the technical idea of the present invention, and the technical idea of the present invention is not limited to the following. Absent. The technical idea of the present invention can be modified in various ways within the technical scope described in the claims. Further, in each of the following embodiments, an electric furnace for steelmaking will be described as an example, but the application target of the present invention is not limited to these, and a plant other than the electric furnace is to be applied. Is also possible.

<Configuration of operating status estimation system>
FIG. 1 is a schematic diagram showing a configuration of an operating state estimation system according to the present embodiment. The operating state estimation system 10 includes an electric furnace 20 and an operating state estimation device 30.

FIG. 2 is a schematic view showing an outline of the configuration of the electric furnace 20. The electric furnace 20 includes a container-shaped furnace main body 21 and a furnace lid 22 provided at an upper opening thereof. A plurality of (for example, three) electrodes 23 are supported on the furnace lid 22 so as to penetrate vertically, and the lower portion of the electrodes 23 is inserted into the furnace. Further, the furnace lid 22 is provided with a raw material charging inlet 24, and the raw material 25 is charged into the furnace from the raw material charging inlet 24. Electricity is flowing through the electrode 23, and the arc heat and resistance heat generated at the tip (lower end) portion of the electrode 23 melt the raw material 25 near the tip of the electrode 23. The molten metal 26 accumulates in the lower part of the furnace and is discharged from the discharge port 27 provided in the furnace main body 21 at regular intervals by batch operation.

The electrode 23 is an artificial graphite electrode composed of graphite, and has a rod shape extending vertically. The electrode 23 is droopingly supported by a support portion 28 provided on the furnace lid 22. The support portion 28 can move the electrode 23 in the vertical direction, whereby the tip position of the electrode 23 can be raised or lowered. Further, since the electrode 23 is consumed as it is used, the raw material of the electrode is charged from above the support portion 28 according to the amount of consumption, and the electrode 23 is sequentially fired to form the electrode 23.

See FIG. 1 again. A plurality of sensors 40 to 44 are installed in the electric furnace 20. The sensor 40 detects the exhaust gas component, and the sensor 41 detects the exhaust gas temperature. The sensor 42 detects the resistance value of the electrode 23, and the sensor 43 detects the amount of raising and lowering of the electrode 23. The sensor 44 detects the power consumption of the electrode 23. These sensors 40 to 44 are connected to the operating state estimation device 30.

Next, the configuration of the operating state estimation device 30 according to the present embodiment will be described. During the operation of the electric furnace 20, the tip of the electrode 23 is buried in the raw material 25 in the furnace, and the position of the tip cannot be confirmed or detected from the outside. Therefore, the operating state estimation device 30 estimates the electrode length, which is the distance from the reference position of the electrode 23 (for example, the lower end position of the support portion 28) to the tip, as the operating state from the operating data related to the operation of the electric furnace 20. To do.

FIG. 3 is a block diagram showing a configuration of an operating state estimation device according to the present embodiment. The operating state estimation device 30 is realized by a computer. As shown in FIG. 3, the operation state estimation device 30 includes a control unit 310, an input unit 320, and a display unit 330. The main body 310 includes a CPU 311, ROM 312, RAM 313, a hard disk 315, an input / output interface 316, and a video output interface 317. The CPU 311, ROM 312, RAM 313, a hard disk 315, an input / output interface 316, and a video output interface 317 are buses. Connected by.

The CPU 311 executes a computer program loaded in the RAM 313. Then, when the CPU 311 executes the operation state estimation program 350, which is a computer program for estimating the operation state of the electric furnace 20, the computer functions as the operation state estimation device 30. In addition, the operating state estimation program 350 includes an estimation model 360 used for estimating the operating state.

The RAM 313 is used to read the operation state estimation program 350 recorded on the hard disk 315. Further, the RAM 313 is used as a work area of the CPU 311 when the CPU 311 executes a computer program.

The operating state estimation program 350 is installed on the hard disk 315. Further, the hard disk 315 has an operation record database (operation record DB) 370 that stores operation data for a predetermined target item for the electric furnace 20, and an estimation result database (estimation result DB) that stores the estimation results by the estimation model 360. ) 380 and an evaluation result database (evaluation result DB) 390 that stores the evaluation results of the estimated values obtained by the estimation model 360 are provided. Each of the operation record DB370, the estimation result DB380, and the evaluation result DB390 stores the storage date and time, the batch number, and the data of the material type to be manufactured, and the data stored in each database corresponds to each data. It is attached.

The operation data includes the data output from each of the above sensors 40 to 44 and the data used for the operation of the plant (for example, the component of the discharged metal, the history of raising and lowering the electrode 23, etc.). More specifically, the operation data includes overall data on the entire environment inside the furnace such as the components and temperature of the exhaust gas from the electric furnace 20, the components of the discharged metal, the exhaust gas components and the temperature around the electrode 23, and the components of each electrode 23. Local data regarding the environment around each electrode 23, such as vertical movement and the amount of power used by each electrode 23, is included. The estimation model 360 takes the operation data as an input and estimates the electrode length, which is an item different from the operation data. The electrode length is an item that cannot be detected from the outside, and the estimated value of the electrode length when the operation data stored in the operation record DB 370 is input to the estimation model 360 is stored in the estimation result DB 380. Further, the evaluation result is given as an evaluation value of a ten-step formula as described later. That is, there are evaluation values of "1" to "10", and the lower the evaluation, the lower the evaluation value, and the higher the evaluation, the higher the evaluation. The evaluation result DB 390 stores the evaluation for the estimated value stored in the estimation result DB 380.

但し、Ｙは目的変数であり、Ｘ_ｉ（ｉ＝１，２，…，ｎ：ｎは推定に用いる測定データの項目数）は説明変数であり、ａ（ａ＝［ａ_０，ａ_１，…，ａ_ｎ］）はパラメータである。 Here, the estimation model 360 will be described. The estimation model 360 is a statistical model (multiple regression equation) obtained by multiple regression analysis, which is a statistical method, and is given as the following equation (1).

However, Y is an objective variable, X _i (i = 1, 2, ..., N: n is the number of measurement data items used for estimation) is an explanatory variable, and a (a = [a ₀ , a ₁ , ₁ ,) _{..., a} n]) is a parameter.

An input unit 320 including a keyboard and a mouse is connected to the input / output interface 316, and the user can input data to the operation state estimation device 30 by using the input unit 320. Further, each of the above-mentioned sensors 40 to 44 is connected to the input / output interface 316, and is configured to receive output data from these sensors 40 to 44.

The video output interface 317 is connected to a display unit 330 composed of an LCD, a CRT, or the like, and outputs a video signal corresponding to the image data given by the CPU 311 to the display unit 330. The display unit 330 displays an image (screen) according to the input video signal.

<Operation of operating state estimation system>
Hereinafter, the operation of the operating state estimation system according to the present embodiment will be described. The CPU 311 of the operation state estimation device 30 executes the operation state estimation program 350 to execute the operation state estimation process described below. FIG. 4 is a flowchart showing a procedure of the operation state estimation process by the operation state estimation device 30 according to the present embodiment. During the operation of the electric furnace 20, data is output by the sensors 40 to 44, and the operation data including these is given to the operation state estimation device 30.

When the operation state estimation device 30 receives the operation data (step S101), the CPU 311 stores these operation data in the operation record DB 370 together with the storage date and time, the batch number, and the material type (step S102). Next, the CPU 311 executes the model update process (step S103).

FIG. 5 is a flowchart showing the procedure of the model update process. In the model update process, the CPU 311 first acquires the past operation data from the operation record DB 370 (step S201). At this time, operation data for a plurality of storage dates and times is acquired. Further, the CPU 311 acquires the estimated value of the past electrode length from the estimation result DB 380 (step S202), and acquires the past evaluation result from the evaluation result DB 390 (step S203). Data for a plurality of storage dates and times are also acquired for these estimated values and evaluation results.

As described above, the evaluation result of the estimated value is given as an evaluation value of 10 stages. The CPU 311 sets a weight for each evaluation result (step S204). In this process, the higher the evaluation value, the larger the weight, and the lower the evaluation value, the smaller the weight. The weight set here is assigned to the operation data associated with the evaluation result by the storage date and time, the batch number, and the material type.

Next, the CPU 311 extracts a part of the operation data acquired in step S201 as the value of the model variable x _0i (step S205). The operation data used as the model variable is n items (n <m) of the operation data (number of items m) acquired in step S201, and in this process, the operation of n items at each storage date and time is performed. The data is extracted.

The following table is for explaining the extracted operation data. In the following table, each data in one row is associated with each other by storage date, batch number, and material type. That is, y ₁ , x ₁₁ , x ₁₂ , ..., X _1n correspond to each other. This means that the estimated value y ₁ was obtained in the past using the operation data of x ₁₁ , x ₁₂ , ..., X _1n . Similarly, y ₂ , x ₂₁ , x ₂₂ , ..., X _2n correspond to each other, and y _N , x _N1 , x _N2 , ..., X _Nn correspond to each other. However, N is the number of records acquired from the operation record DB370, the estimation result DB380, and the evaluation result DB390 in steps S201 to S203. The extracted operation data and the estimated values and evaluation values obtained in steps S202 and S203 are associated with each other by the storage date and time, the batch number, and the material type. Hereinafter, the combination of operation data, estimated value, and evaluation value corresponding to each other is referred to as a data set.

また、ｔより次のＷが式（３）で与えられる。

The CPU 311 then updates the estimation model 360 by weighted multiple regression analysis (step S206). This process will be described below. The objective variable Y, the explanatory variable X, and the weight t are expressed by the following equation (2).

Further, the following W from t is given by the equation (3).

このようにしてパラメータａ（ａ＝［ａ_０，ａ_１，…，ａ_ｎ］）を求めることで、新たな推定モデル３６０が構築される。得られたパラメータａ（ａ＝［ａ_０，ａ_１，…，ａ_ｎ］）は、ハードディスク３１５に記憶される。以上で、モデル更新処理が終了する。 By weighted multiple regression analysis using Y, X, W, the parameter a (a = [a ₀ , a ₁ , ..., _An ]) is obtained by the equation (4).

By obtaining the parameter a (a = [a ₀ , a ₁ , ..., _An ]) in this way, a new estimation model 360 is constructed. The obtained parameter a (a = [a ₀ , a ₁ , ..., _An ]) is stored in the hard disk 315. This completes the model update process.

See FIG. 4 again. After the model update process, the CPU 311 executes an estimation process for estimating the electrode length from the latest operation data (step S104).

FIG. 6 is a flowchart showing the procedure of the estimation process. In the estimation process, the CPU 311 first reads the parameter a (a = [a ₀ , a ₁ , ..., _An ]) from the hard disk 315 (step S301). Further, the CPU 311 extracts the above n items (hereinafter, referred to as “estimation target data”) of the latest (that is, the latest storage date and time) data from the operation data acquired in step S201 (step S302). ..

Next, the CPU 311 applies the extracted latest operation data to the estimation model 360, and calculates an estimated value of the electrode length (step S303). _{Assuming that the} estimation target data x _0i (i = 1, 2, ..., N) and the estimated value y _p0 , the estimated value y _p0 is calculated by the following equation (5). The calculation of the estimated value is performed for each electrode 23.

The CPU 311 registers the estimated value calculated as described above in the estimation result DB 380 together with the storage date / time, the batch number, and the material type (step S304). This completes the estimation process.

See FIG. 4 again. After the estimation process, the CPU 311 executes an estimation result output process that outputs an estimated value of the electrode length (step S105).

FIG. 7 is a flowchart showing the procedure of the estimation result output processing. In the estimation result output process, the CPU 311 first reads the latest estimated value from the estimation result DB 380 (step S401). Further, the CPU 311 causes the display unit 330 to display the estimation result screen including the read estimated value (step S402). This completes the estimation result output processing.

See FIG. 4 again. After the estimation result output processing, the CPU 311 executes an evaluation processing for evaluating the estimated value of the electrode length (step S106).

FIG. 8 is a flowchart showing the procedure of the evaluation process. In the evaluation process, the CPU 311 accepts the input of the evaluation value from the user (step S501). FIG. 9 is a diagram showing an example of the estimation result screen. The estimation result screen 600 includes an estimated value 601 of the electrode length of each electrode 23. Further, input buttons 602a and 602b for inputting an evaluation value and an evaluation value display unit 603 for displaying the input evaluation value are provided corresponding to each estimated value. After confirming the outputs of various sensors provided in the electric furnace 20, the user evaluates whether the estimated electrode length 601 is an appropriate value based on the know-how gained from his / her own experience. Then, by operating the input buttons 602a and 602b, the evaluation value is input to the operation state estimation device 30. The evaluation value is a step formula of 10 steps, and the user selects the input button 602a by the number when he / she wants to further increase the evaluation value from the value displayed on the evaluation value display unit 603, and when he / she wants to decrease the evaluation value. Select the input buttons 602b for that number. In this way, the evaluation value of 10 steps is input to the operating state estimation device 30.

Further, the estimation result screen 600 is provided with a save button 604 for saving the estimated value, and the user operates the input unit 320 to select the save button 604 to save the input estimated value. The instruction is given to the CPU 311. See FIG. 8 again. When the evaluation value is input and the storage instruction is given, the CPU 311 registers the evaluation value together with the storage date and time, the batch number, and the material type in the evaluation result DB 390 (step S502). This completes the evaluation process.

See FIG. 4 again. After the evaluation process, the CPU 311 returns the process to step S101, and repeats steps S101 to S106 again.

As described above, when the evaluation value is high, the weight is large, and when the evaluation value is low, the weight is small. Such weights are assigned to each dataset and multiple regression analysis is performed using the dataset and the corresponding weights. Therefore, the larger the weight, that is, the higher the evaluation value, the larger the influence on the updated estimation model 360, and the smaller the weight, that is, the lower the evaluation value, the smaller the influence. When the user (expert) evaluates the estimated value, the evaluation value is set according to the knowledge (know-how) obtained from his / her past experience, so that the user's know-how is reflected as a weight in the construction of the estimation model 360. .. Therefore, by repeating the update, the estimation model 360 can perform estimation close to the know-how of the user.

(Other embodiments)
In the above-described embodiment, the configuration in which the estimation model 360 is used as a statistical model has been described, but the present invention is not limited thereto. It is also possible to use the estimation model 360 as a physical model that simulates a physical phenomenon in the electric furnace 20. However, the estimation model 360 needs to include adjustable parameters. For example, as a physical model including parameters, the physical relationship between the electric power W (or electric energy) supplied to the electrode 23 and the consumption amount C of the electrode 23 is a model such as C = bW (b: parameter). It can be simulated by an equation. Further, the estimation model 360 may be a combination of a statistical model and a physical model.

Further, in the above-described embodiment, the configuration for updating the estimation model 360 using the multiple regression analysis has been described, but the present invention is not limited to this. The estimation model 360 can be updated by using other statistical analysis methods such as PLS (Partial Least Squares) and support vector regression.

Further, in the above-described embodiment, the configuration for estimating the electrode length as the operating state of the electric furnace 20 has been described, but the present invention is not limited to this. Although it is possible to estimate the operating conditions other than the electrode length, such as the state of equipment such as blast furnaces, heating furnaces, sintering machines, and sintering furnaces, the authenticity cannot be easily confirmed, and field operators. It is also possible to estimate the operational soundness of large-scale equipment that tends to depend on the know-how of.

Further, in the above-described embodiment, the configuration in which all the processes of the operation state estimation program 350 are executed by the single operation state estimation device 30 has been described, but the present invention is not limited to this, and the operation is not limited to this. It is also possible to make a distributed system in which the same processing as that of the state estimation program 350 is distributed and executed by a plurality of devices (computers).

The plant operating state estimation device and the operating state estimation method of the present invention are useful as an operating state estimating device and an operating state estimating method for estimating the operating state of a plant by using field know-how.

10 Operating state estimation system 20 Electric furnace 30 Operating state estimation device 310 Control unit 311 CPU
320 Input section 330 Display section 350 Operation status estimation program 360 Estimate model 370 Operation record database 380 Estimate result database 390 Evaluation result database 600 Estimate result screen 601 Estimated value

Claims (7)

Input the operation data into the estimation model that inputs the operation data for a predetermined target item of the plant and outputs the estimated value of the operation state for the item different from the target item of the plant, and inputs the operation data from the estimation model. An estimation means for acquiring the output estimated value, and
An evaluation means that accepts the evaluation of the estimated value acquired by the estimation means from the user , and
An operation record storage unit that stores the operation data of the plant,
An estimation result storage unit that stores an estimated value when the operation data stored in the operation record storage unit is input to the estimation model, and an estimation result storage unit.
An evaluation result storage unit that stores the evaluation of the estimated value stored in the estimation result storage unit, and an evaluation result storage unit.
A model update that updates the estimation model based on the operation data stored in the operation record storage unit, the estimated value stored in the estimation result storage unit, and the evaluation stored in the evaluation result storage unit. Equipped with means,
Plant operation status estimation device. The estimation model contains adjustable parameters
The model updating means is configured to update the estimated model by adjusting the parameters based on the operational data, the estimated values and the evaluation.
The plant operating state estimation device according to claim 1. The estimation model includes a statistical model obtained by a statistical analysis method.
The operating state estimation device for the plant according to claim 2. The estimation model includes a physical model that mimics a physical phenomenon in the plant.
The operating state estimation device for the plant according to claim 2 or 3. The model updating means is configured to set weights on the operation data and update the estimation model based on the operation data weighted by the set weights.
The operating state estimation device for a plant according to any one of claims 1 to 4. The plant is an electric furnace used for steelmaking.
The operation data includes information on the resistance value of the electrode of the electric furnace, the raw material to be charged into the electric furnace, the by-product gas component generated by the electric furnace, and the temperature of the by-product gas.
The estimated value is the length of the electrode.
The operating state estimation device for a plant according to any one of claims 1 to 5. The computer inputs the operation data to the estimation model that inputs the operation data for a predetermined target item of the plant and outputs the estimated value of the operation state for the item different from the target item of the plant, and inputs the operation data to the estimation. The step of acquiring the estimated value output from the model and
A step in which the computer receives an evaluation of the acquired estimated value from the user ,
The computer has a step of updating the estimated model based on the past operation data of the plant, the estimated value when the operation data is input to the estimated value, and the evaluation of the estimated value.
How to estimate the operating status of a plant.

Images