JPH10207507A - Plant simulator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable adaptation to any plant and to adjust transient response characteristics matching an actual plant in a short time by adding an error model to a simulation model based upon the physical rule of the plant and compensating an error between an actual machine plant and the simulation model with the error model. SOLUTION: A plant transient response characteristic adjusting means 103 automatically generates the error model 105 for compensating the error between the current actual machine and model. A plant transient response characteristic simulating means 104 inputs the manipulated variable of the actual machine and outputs a simulation result. Further, the error model 105 of the plant transient response characteristic adjusting means 103 also inputs the manipulated variable of the actual machine of the actual machine and outputs an error estimated value of plant transient response characteristics based upon the error model 105. This simulation result precisely simulates the actual machine since the error model 105 is compensated into one corresponding to the actual machine at the time of adjustment. Consequently, the transient response characteristics of the actual machine plant can be reproduced with high precision without altering the simulation model which requires a long time for adjustment.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a plant simulator, and more particularly, to a plant simulator having a function of automatically adjusting a transient response characteristic.

[0002]

2. Description of the Related Art A plant simulator includes a development simulator for developing a plant itself and a control method for the plant, and a simulator for operator training for understanding and operation training of a plant by a plant operator. There is. The plant operator training simulator includes:
A full simulator that has the same control panel as the actual machine and can perform operation training similar to the actual machine, and a small simulator that is small in scale but can simulate the plant with a workstation and an input means such as a keyboard or mouse as a console. (Roy Fray, COM
PACT SIMULATORSFOR FOSSIL-FUELED POWER PLANTS, IEEE
Spectrum Volume 32, Number 2, February 1995, pp.4
6-51).

[0003] Some development simulators or small-scale simulators adopt a building block system which enables simulations in which plant components are arbitrarily combined and enables operation training of an original plant or a new plant. is there.

[0004] The development simulator is used for the purpose of adjusting the control system of the actual plant. Therefore, the simulator for development determines the control parameters of the control system of the actual plant by adjusting the steady-state characteristics and the transient response characteristics so as to reproduce the operation results of the actual plant, and applies the determined control parameters to the actual plant. .

[0005] The training simulator simulates a real plant or a virtual plant, and is used for a plant operator to learn operation and maintenance of the plant and the configuration and characteristics of the plant. For this reason, the manufacturer of the training simulator has adjusted the steady-state characteristics and the transient response characteristics of the simulator in advance.

To adjust the steady-state characteristics of the plant simulator, a differential equation based on physical laws such as a law of conservation of mass, a law of conservation of energy, and a law of heat conduction is formulated based on a plan value at the time of plant design. It is general to adjust by a steady solution.

[0007] On the other hand, the adjustment of the transient response characteristic of the plant simulator has conventionally been carried out by an adjustment worker changing the adjustment variable called an adjustment parameter while comparing the calculation result of the plant simulator with the operation result of the real plant or the virtual plant. At present, the adjustment parameter is adjusted to the optimum value so that the two values coincide with each other.

[0008]

In the conventional plant simulator, since the adjustment operator manually adjusts the limits and adjustments of the simulation model, it takes time to adjust the transient response characteristics. The operator training simulator has a problem that every time the component of the plant to be simulated is changed, the plant coordinator must manually adjust the operator training simulator, resulting in a high adjustment cost. Furthermore, in the development simulator, it is necessary to adjust the development simulator within a limited time during the trial operation of the actual plant, and if the adjustment of the transient response characteristics is delayed, the plant trial operation period is prolonged. There is.

Furthermore, in an actual plant, there is a phenomenon called "aging" in which the characteristics of the plant change over time. However, it is difficult to simulate this aging with a simulator for operator training, and plant operators have difficulty. There is also a problem that the characteristics of the current actual plant cannot be grasped using the operator training simulator.

An object of the present invention is to provide a plant simulator that can be applied to any plant and that can adjust a transient response characteristic suitable for an actual plant in a short time.

[0011]

The above object is achieved by adding an error model to a simulation model based on the physical laws of the plant and providing means for compensating the error between the actual plant and the simulation model with the error model. You.

As a result, the transient response characteristics of the actual plant can be reproduced with high accuracy without changing the simulation model requiring a long time for adjustment.

[0013]

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a configuration diagram of a plant simulator according to the first embodiment of the present invention. The plant simulator according to the present embodiment includes an operation unit 100, an operation unit 300 such as a console, and an output unit 200 such as a CRT display.

The calculating means 100 comprises a plant body simulating means 101 for simulating the transient response characteristics of the plant based on the operating conditions set by the operating means 300 and outputting a simulated result, and a plant control device for simulating a plant control device. Simulating means 102,
A plant operation result storage unit for storing actual machine operation records taken in from the actual plant via the transmission path; This plant operation result storage means 400 is used when adjusting the plant transient response characteristics of the plant simulator.
The actual machine operation amount operation condition is output to the plant control device simulation means 102, and the actual machine state quantity operation result is output to the plant transient response characteristic simulation means 101.

The plant main body simulation means 101 includes a plant transient response characteristic adjustment means 103 constituted by a differential equation based on a physical equation, and a transient response characteristic for outputting a simulation result when an actual operation amount is applied to a physical model. Simulation means 104.

As shown in FIG. 2, the plant main body simulating means 101 includes a plant transient response characteristic adjusting means 103, a plant transient response characteristic simulating means 104, and a subtractor 107. The plant transient response characteristic adjusting means 103 includes the error model 1
05 and the least squares method 106. The operation of the plant body simulation means 101 differs between when adjusting the plant transient response characteristics and when calculating the plant transient response characteristics represented during plant operation training.

FIG. 2 is an explanatory diagram of the operation of the plant main body simulating means 101 when adjusting the transient response characteristics of the plant.
During this adjustment, the plant transient response characteristic simulation means
104 captures the actual machine operation amount in time series and outputs a simulation result. Then, the subtracter 107 subtracts the simulation result at each time from the operation result of the actual machine state quantity, and outputs the subtraction result. The result of the subtraction (that is, a time-series signal of the error between the physical equation and the actual plant) is an error in the transient response characteristic.

On the other hand, plant transient response characteristic adjusting means 103
Captures the actual machine operation amounts in time series and calculates an error estimation value at each time of the transient response characteristic. Least squares method 106
Corrects the error model 105 so that the square error between the time series signal output from the subtracter 107 and indicating the error of the transient response characteristic and the time series signal indicating the error estimation value is minimized. Thus, the plant transient response characteristic adjusting means 10
3 automatically creates an error model 105 that compensates for the error between the current actual machine and the model.

FIG. 3 is an explanatory diagram of the operation of the plant body simulating means 101 when calculating the plant transient response characteristics.
In this calculation, the least squares method 106 is separated and the subtractor 1
07 operates as an adder 108, and the adjusting means 103 holds an error model 105 compensated at the time of adjustment.

At the time of this calculation, the plant transient response characteristic simulation means 104 takes in the operation amount of the actual machine and outputs a simulation result. Further, the plant transient response characteristic adjusting means 10
The third error model 105 also takes in the operation amount of the actual machine, and outputs an error estimated value of the plant transient response characteristic based on the error model 105. The adder 108 adds the simulation result and the error estimation value, and outputs the result as a simulation result of the state quantity. Since the simulation result is compensated by the error model 105 corresponding to the actual machine at the time of the adjustment, the simulation model accurately simulates the actual machine.

Next, a specific example of the error model 105 will be described. It is assumed that the plant body simulation means 101 reproduces the non-linearity of the control amount with respect to the operation amount of the actual plant in the steady state characteristics. At this time, the error model 105 used in the plant transient response characteristic adjusting means 103 needs to reproduce the linearity (deviation) of the control amount with respect to the manipulated variable in the plant transient response characteristic.

In this example, the error model 105 is replaced by a linear model, that is, an auto-regressive-moving average.
rage: ARMA) model. The basic equation of the ARMA model is expressed as in the following Equation 1.

[0023]

(Equation 1)

Here, x '(i): estimated value of control amount error (i = 1,..., M) x (i): true value of control amount error (i = 1,. ) u (i): Time-series sample value of the manipulated variable (i = 0, ..., m) ai, bi: Model parameter m: Order of ARMA model T: Sampling period n: Integer (m ≦ n) .

From Equation 1, the estimated value by the ARMA model is calculated by a linear combination of the past estimated value and the past input. At this time, the model parameters ai and bi are modified by the least squares method so that the evaluation value J of the following Expression 2 is minimized.

[0026]

(Equation 2)

As a method of calculating model parameters by the least square method, there are matrix calculation methods such as an LU decomposition method and a house holder method. The method for calculating model parameters by the Householder method is described in “Fortran 77 Time Series Analysis Programming” pp.173-183 (Genshiro Kitagawa: Iwanami Shoten (Hei 5-
3)).

The order of the model represents how much the error model can simulate higher-order errors. In general, the higher the model order, the higher the accuracy of the ARMA model,
When the model is mounted on a computer, the model is terminated at an appropriate order in consideration of the processing speed and the storage capacity of the computer. An example of implementation of this method is the Akaike Informati
on Criterion: AIC). The Akaike Information Code is "FOR
TRAN77 Time Series Analysis Programming, ”pp. 90-94 (Genshiro Kitagawa: Iwanami Shoten (Heisei 5-3)).

The sampling period is determined in consideration of the characteristics of the error compensated by the ARMA model. Normally, the spectrum of the plant transient response characteristic error is obtained by using the fast Fourier transform, and from the spectrum of the error compensated by the ARMA model based on the spectrum, a sampling cycle equal to or less than の of the cycle of the waveform is used. The method for determining the sampling period is described in detail in "System Identification Theory for Users", pp.77-87 (Shuichi Adachi: Society of Instrument and Control Engineers (Heisei 5-7)).

FIG. 4 is a configuration diagram of a plant simulator according to a second embodiment of the present invention. In this embodiment,
A plurality of plant body simulation means (10 in the illustrated embodiment)
1a and 101b), and each simulation means is assigned to a plant component to be simulated. Each simulation means includes an adjustment means and a simulation means. In the illustrated embodiment, the simulating unit 101a includes a plant transient response characteristic adjusting unit 103a and a plant transient response simulating unit 104a, and the simulating unit 101b includes a plant transient response characteristic adjusting unit 103b and a plant transient response simulating unit.
104b.

For example, plant transient response characteristic simulation means 104a
When a model of a certain plant component device is obtained and an operation result of the component device is obtained from an actual plant, the plant transient response characteristic adjusting unit 103a identifies the error model of the plant component device. In this way, by providing a plurality of simulation means and allocating the simulation means to each component device, it is possible to make adjustments for each component device even when a plant component device is added or changed.

FIG. 5 is a configuration diagram of a plant simulator according to a third embodiment of the present invention. In this embodiment,
The plant operation result sequential input means 109 is provided in the calculation means 100, and the actual plant 500 and the plant operation result sequential input means are connected by the transmission line 501, and the plant operation result is transmitted through the plant operation result sequential input means 109. The plant operation result storage means 400 is sequentially stored, and the plant body simulation means 1 is stored.
Reference numeral 01 denotes a plant operation result sequential input means 109 which takes in data, and a plant transient response characteristic adjusting means 103 which sequentially adjusts the transient response characteristic of the plant transient response characteristic simulation means 104.

In the present embodiment, the latest operation results of the actual plant 500 are input to the plant simulator, and the transient response characteristics are adjusted as needed. Therefore, operation training can always be performed using the characteristics of the actual plant. The effect can be improved.

FIG. 6 is a configuration diagram of a plant simulator according to a fourth embodiment of the present invention. In this embodiment,
A plant simulator is connected to a plurality of actual plants by transmission lines, and one plant simulator simulates a plurality of actual plants. Therefore, the calculation means
100 includes an actual machine data switching device 110, and plant operation result storage means 410 and 420 are provided for a plant. In this simulator, the error of the transient response characteristic of the actual plant 510 is compensated by the plant transient response adjustment unit 103, and the obtained error information is stored in the plant transient response adjustment result storage unit 600 in association with the actual plant 510. Store it. Similarly, the error of the transient response characteristic of another actual plant 520 is compensated, and the obtained error information is
And stored in the storage means 600. In this way, by selecting an actual plant to be simulated, error information corresponding to the plant can be retrieved from the storage unit 600 and used, so that a single plant simulator can simulate a plurality of plants. Becomes In the present embodiment, a simulation of a plurality of actual plants has been described. However, if error information at each time in one plant is stored in the storage means corresponding to each time, a plant at a certain time in the past is reproduced. It is needless to say that reproduction can be performed by designating the time.

FIG. 7 is a configuration diagram of a plant simulator according to a fifth embodiment of the present invention. In this embodiment,
The application target is a development simulator, which is used to adjust the control system during test operation of an actual plant. The plant operation result is stored in the plant operation result storage means 400 from the actual plant 500 via the transmission line 501. Next, the plant transient response characteristic adjusting unit 103 is adjusted so that the plant operation result matches the calculation result of the plant transient response characteristic simulation unit 104. Then, the control amount of the plant control device simulation means 102 is adjusted by combining the plant body simulation means 101 and the plant control device simulation means 102, and the result is output from the simulation means 102 to the control device 502.

According to the present embodiment, it can be used for adjusting the control system of a large-scale plant having a high operating cost or a plant having a limited trial operation period. It is effective to apply to the attitude control of the structure.

In the above equation (1), an ARMA model is used as an example of a linear model as an error model. However, a nonlinear ARMA model or a nonlinear ARMA model is used as a model that includes a nonlinear element in the error between the actual plant and the plant simulation means. It goes without saying that a neural network may be used.

FIG. 8A is an external view of a small simulator. This small simulator includes a main body 10, a display 20, and a keyboard 30, and has a specific key on the keyboard 30 as an instruction input key 40 for automatic adjustment.

FIG. 8B is an external view of an operator training simulator. This operator training simulator has a configuration that simulates a central control unit of an actual plant, and includes a simulator main body 50, a large display 60, and a console 70, and an automatic adjustment instruction is provided on the console 70. An instruction input key 80 for performing is provided.

The plant operation trainer or the plant operation instructor presses the automatic adjustment instruction input keys 40, 80 when the operation results of the actual plant displayed on the displays 20, 60 are different from the calculation results of the simulator. Then, the simulator automatically adjusts the transient response characteristics described in the above embodiment, and updates the error model 105. Then, the simulator performs recalculation using the updated error model, and displays the recalculation result on the same screen as the above calculation result. As a result, the simulator can perform a highly accurate simulation of the actual plant, and can recognize at a glance how the simulation result has been corrected by the automatic adjustment.

Note that a touch panel,
It goes without saying that a light pen, tablet, mouse, etc. can be used.

[0042]

According to the present invention, the transient response characteristic of the plant model is automatically adjusted from the actual equipment characteristic and the conventional transient response characteristic model. Therefore, the transient response characteristic of the plant equipment model, which is conventionally difficult to adjust, is changed to the actual equipment characteristic. It can be easily adjusted. Furthermore, the transient response characteristics of the plant model are automatically adjusted by a combination of the physical equation based on the physical law and the error model obtained from the operation results of the actual plant. Can be output.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a simulator according to a first embodiment of the present invention.

FIG. 2 is an operation explanatory diagram of the plant body simulation means shown in FIG. 1 at the time of adjusting transient response characteristics.

FIG. 3 is an explanatory diagram of an operation when calculating a transient response characteristic of the plant body simulation means shown in FIG. 1;

FIG. 4 is a configuration diagram of a simulator according to a second embodiment of the present invention.

FIG. 5 is a configuration diagram of a simulator according to a third embodiment of the present invention.

FIG. 6 is a configuration diagram of a simulator according to a fourth embodiment of the present invention.

FIG. 7 is a configuration diagram of a simulator according to a fifth embodiment of the present invention.

FIG. 8 is an external view of a small simulator (a) and a simulator for operator training (b).

[Explanation of symbols]

10, 50: Simulator main body, 20, 60: Display, 30: Keyboard, 40, 80: Auto-adjustment instruction input key, 70: Operation console, 101: Plant body simulation means, 102: Plant control apparatus simulation means, 103 ... Plant transient response characteristic adjustment means, 104: Plant transient response characteristic simulation means, 105: Error model, 106: Least square method, 107: Subtractor (adder).

 ──────────────────────────────────────────────────の Continued on the front page (72) Inventor Yoshio Sato 7-1-1, Omika-cho, Hitachi City, Ibaraki Prefecture Within Hitachi Research Laboratory, Hitachi, Ltd.

Claims (4)

[Claims] In a plant simulator for simulating transient response characteristics of an actual plant, an error of the transient response characteristics is obtained from a difference between a simulation result of the transient response characteristics of the plant and an operation result of the actual plant, and an error model is obtained from the error. A plant simulator comprising: means for creating; and output means for outputting a result obtained by correcting the simulation result with an error estimated value obtained from the error model. 2. A plant simulator for simulating a transient response characteristic for each of a plurality of actual plants, each system of an actual plant, or each constituent device, wherein the transient response characteristic is obtained from a difference between the simulation result of the transient response characteristics and the operation result of the actual plant. Means for obtaining an error in the response characteristic, creating an error model from the error, output means for outputting a result obtained by correcting the simulation result with an error estimation value obtained from the error model, an actual plant or system to be simulated A plant simulator comprising: means for switching constituent devices. 3. The display unit according to claim 1, wherein a simulation result output from the output unit before updating the error model and a simulation result output from the output unit after updating are displayed on the same screen. A plant simulator comprising: 4. A plant simulator for simulating a transient response characteristic of a plant using a simulation model based on physical rules of an actual plant, wherein an error model is added to the simulation model, and an error between the actual plant and the simulation model is calculated by the error model. A plant simulator provided with a compensation means.