JP2010225022A - Simulation system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a simulation system that achieves efficient reuse of a model parameter set in the past. <P>SOLUTION: The simulation system for adjusting a parameter so that an output of an simulation object matches a simulation result while the simulation is performed on the basis of data input to an actual simulation object has: a simulation result database for storing the result of the simulation; a parameter history database for storing history of the parameter set during the simulation; and a precision calculation part for calculating degree of matching between the result of the simulation and the output of the simulation with respect to a model parameter corresponding to an output value of the simulation object according to the parameter history database. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a simulation system, and more particularly to improvement of parameter adjustment in various simulators including a process simulator used in a process industry such as petroleum and petrochemicals.

  For example, in the process industries such as petroleum and petrochemicals, there is an interest in building a simulator that faithfully reproduces the actual process on a computer and utilizing the results of this simulation for operation in order to improve the safety and efficiency of plant operations. It has increased rapidly in recent years.

By building a simulator that faithfully simulates the plant on the computer,
(1) Visualization of internal conditions that cannot be actually measured (2) Predictive simulation under various operating conditions (3) Model-based control incorporating process simulation into automatic control (4) Realization of optimization of production costs and product quality is expected Because it can.

  In this way, the use of a process simulator for actual plant operation can provide significant benefits, but whether or not the process simulator can be applied to actual plant operation is to adapt the parameters of the process model to the actual process. It depends on whether you can.

  However, the process model in which the entire plant is modeled has a huge number of parameters, and in the strict model constructed based on the physicochemical principle, each parameter has a physical meaning. Parameter tuning requires an understanding of the entire process and a great deal of experience, which is a difficult task for a skilled engineer with trial and error.

  Therefore, in order to reduce the difficulty of process modeling as described above, proposals for supporting process model parameter setting have been made in Patent Document 1, Patent Document 2, and the like. In these proposals, model parameters that are highly sensitive to a certain process output are extracted from a large number of parameters of the process model and adjusted.

  Patent Document 1 relates to a process model parameter adjustment apparatus, an adjustment support apparatus, and a method in which a process engineer who is not familiar with the internal structure of a process can easily perform parameter adjustment.

  Patent Document 2 relates to a sewage treatment process simulation method and apparatus that automatically perform parameter setting, reduce prediction errors with high sensitivity, and reduce time.

  Non-Patent Document 1 relates to a calculation method of an evaluation method.

JP 2002-328702 A JP 2004-121953 A

Shuichi Adachi, “Advanced System Identification for MATLAB Control”, 1st edition, 1st edition, Tokyo Denki University Press, March 10, 2004, p.13

  FIG. 7 is a configuration diagram showing an example of a plant simulation system used in a process industry such as conventional petroleum / petrochemicals. In FIG. 7, the plant simulation system includes a plant 1, a plant operation data DB 2, a console screen 3, and a plant simulator 4.

  The plant 1 outputs data corresponding to the input data. In the plant operation data DB2, input data and output data of the plant 1 are sequentially written.

  On the other hand, the operator of the plant 1 sets the parameters of the process model via the console screen 3 and outputs a simulation execution instruction to the plant simulator 4.

  Then, the plant simulator 4 receives the execution instruction, sets the parameters of the process model, reads the input to the plant 1 necessary for the simulation from the plant operation data DB 2, and executes the simulation.

  After executing the simulation, the operator of the plant 1 adjusts the parameters of the process model while comparing the output data of the plant 1 stored in the plant operation data DB 2 with the simulation output calculated by the plant simulator 4 on the console screen 3. To do.

  However, since the current plant simulation system does not have a function for managing and reusing parameters set in the past, the user must pull out the desired model parameters while checking the huge history of parameters. There is a problem that it takes a lot of work time.

  The present invention solves these problems, and an object thereof is to realize a simulation system capable of efficiently reusing model parameters set in the past.

In order to achieve such a problem, the invention according to claim 1 of the present invention is:
In a simulation system that adjusts parameters so that the output of the simulation target matches the result of the simulation while performing simulation based on the data input to the actual simulation target,
A simulation result database for storing the simulation results;
A parameter history database for storing a history of the parameters set during the simulation;
A simulation result of a model parameter corresponding to the output value of the simulation target from the parameter history database and a matching rate calculation unit that calculates the output matching degree of the simulation target in a time series are provided.

Claim 2 is the simulation system according to claim 1,
The relevance ratio calculation unit calculates from the amount of change in the relevance ratio between the simulation target output and the simulation result before and after the simulation model parameter change.

Claim 3 is the simulation system according to claim 1 or 2,
The result of the precision calculation unit is displayed as a bar graph of the precision for the history.

Claim 4 is the simulation system according to claim 1,
The simulation result is displayed as a bar graph of the history for each arbitrarily selected parameter.

Claim 5 is the simulation system according to claim 1,
Corresponding to the output of the simulation target, an arbitrarily selected history is displayed as a trend.

Claim 6 is the simulation system according to any one of claims 1 to 5,
The simulation target is a process plant.

  By configuring in this way, model parameters set in the past can be efficiently reused.

It is a block diagram which shows one Example of the simulation system based on this invention. It is a flowchart which shows the operation | movement of FIG. It is explanatory drawing which shows an example of the data stored in parameter log | history DB6. It is a trend screen example figure which shows the specific example of the model parameter selected from parameter log | history DB6. It is the example of a graph screen which showed the precision calculation result of each log | history a to f in the bar graph. It is an example figure of a trend display screen of a process output and a model parameter output of a selected history. It is a block diagram which shows an example of the plant simulation system used in process industries, such as the conventional petroleum and petrochemical.

  Hereinafter, the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the simulation system of the present invention, and the same reference numerals are given to the parts common to FIG.

  The difference between the simulation system of the present invention shown in FIG. 1 and the conventional simulation system shown in FIG. 7 is that a simulation result database (hereinafter referred to as DB) 5 for storing the simulation results of the plant simulator 4 and the simulation of the plant simulator 4 A parameter history DB 6 for storing a history of setting parameters, and a precision calculation unit 7 for calculating the degree of coincidence between the simulation result of the plant simulator 4 and the output of the plant 1 in time series. It is that you are.

  FIG. 2 is a flowchart showing the operation of the simulation system of FIG.

  At the start, the operator designates a section 1 ≦ k ≦ L in which simulation is performed from the console screen 3 and issues a simulation execution command.

  In step S1, the operator selects whether to reset the model parameters on the console screen 3. If the model parameter is reset, the process proceeds to the next step S2. If the model parameter is not reset, an end process is performed.

  In addition, when not resetting, the plant simulator 4 performs a simulation process in the state of the present model parameter. That is, the input / output data of the plant 1 is sequentially written from the plant 1 to the plant operation data DB 2, and the output of the plant 1 and the simulation output of the plant simulator 4 are displayed on the console screen 3.

  In step S2, the operator selects process data to be used for evaluation from the process data stored in the plant operation data 2. For example, temperature, flow rate, etc. are selected as process data.

  In step S3, the plant simulator 4 reads the history parameters from the parameter history DB 6 and displays them on the console screen 3. That is, when the operator determines in advance a model parameter to be changed by some method, the plant simulator 4 extracts a history of past parameters arbitrarily set by the operator from the parameter history DB 6. On the other hand, for parameters that have not been changed, the parameter history is read as many times as the number of times arbitrarily set by the operator for the same history as the current model parameter.

  FIG. 3 is an explanatory diagram showing an example of data stored in the parameter history DB 6, and shows that there are 1 to m parameters for the histories a to f, respectively. The history data stored in the parameter history DB 6 is configured as such a combination of model parameter values in FIG. When the value of the model parameter is changed by tuning or the like at each time, it is recorded in the parameter history DB 6 every time the parameter is changed.

  FIG. 4 is an example of a trend screen showing a specific example of model parameters selected from the parameter history DB 6, where the operator arbitrarily selects and changes parameters 1, 2, n−1, and n from the parameter history DB 6. The trend of the values of the history a to f of the model parameter corresponding to each of these is shown as a bar graph. These example trend screens shown in FIG. 4 are displayed on the console screen 3.

  With the example of the trend screen as shown in FIG. 4, the operator can know the parameter values investigated in the past, and can grasp the transition direction of the parameter to be investigated.

  Returning to FIG. 2, in step S <b> 4, the plant simulator 4 sets a combination of model parameters in each history read from the parameter history DB 6 as model parameters, and executes simulation for each history.

  In step S5, the precision calculated by the precision ratio calculator 7 is displayed. That is, the relevance ratio calculation unit 7 calculates the relevance ratio in each history a to f and displays on the console screen 3 a graph screen showing the relevance ratio calculation results for each history a to f as shown in FIG. To do. In addition, the symbol display 8 of “◯” provided at the end of the relevance ratio display of each history a to f in the graph of FIG. 5 is a trend confirmation button operated when it is desired to confirm the trend for each history. Function.

The plant simulator 4 inputs the input {u _s (k)} (1 ≦ s ≦ q) of the plant 1 stored in the plant operation data DB 2 and the model parameter set {θ _j } (1 ≦ j ≦ M) before the change. The simulation result {y _i ^~ (k)} (1 ≦ i ≦ N) is stored in the simulation result DB 5. Note that the model parameter means, for example, the height and size of the reactor when the plant 1 has a reactor.

Based on the output {y _i (k)} of the plant 1 stored in the process operation data DB 2 and the simulation result {y _i ^~ (k)} stored in the simulation result DB 5, the degree of coincidence of both over time Is calculated by the precision ratio calculation unit 7 as a time-series precision ratio {A _i } (1 ≦ i ≦ N).

  Here, as a calculation method of the relevance ratio Ai, for example, the following equation described in Non-Patent Document 1 can be cited.

  Returning to FIG. 2, in step S6, the trend display of the process output and the model parameter output as shown in FIG. 6 is displayed on the console screen 3 for the history of the trend confirmation button 8 operated on the conformity ratio calculation result graph screen of FIG. I do.

  The trend confirmation button 8 can be selected for a plurality of histories, and FIG. 6 shows an example in which the histories e and f are selected and the trends of the histories e and f are superimposed and displayed.

  As shown in FIG. 5, by graphing the precision of the parameters investigated in the past, as shown in FIG. 6, it is possible to grasp at a glance the trend of an arbitrarily selected history corresponding to the actual output of the plant 1.

  Returning to FIG. 2, in step S7, the model parameter setting is changed.

  When the operator inputs a desired response history into the history window 9 on the screen of FIG. 6 and presses the change button 10, the parameter combination at the time of pressing is set as the model parameter of the plant simulator 4. Further, the plant simulator 4 additionally stores the history of model parameters set in the plant simulator 4 in the parameter history DB 6.

  Then, a series of processing ends.

  With this configuration, the operator can determine the parameter initial value when the parameter is set.

  Further, by storing the parameters searched during parameter tuning on the basis of the process response at an arbitrary time in the parameter history database, it is possible to know the parameter values investigated in the past from the trend screen as shown in FIG. The operator can grasp the direction in which the parameter to be investigated changes.

  Furthermore, by graphing the precision of the parameters investigated in the past as shown in FIG. 5, it is possible to grasp at a glance the trend of an arbitrarily selected history corresponding to the actual output of the plant 1 as shown in FIG. .

  In the above embodiment, an example of parameter adjustment in a process simulator used in the process industry such as petroleum and petrochemicals has been described. However, the present invention also applies to parameter adjustment of a simulator used in an electronic circuit or a semiconductor device. It is valid.

  As described above, according to the present invention, the model parameter values set so far in the database are accumulated as history, the model parameters set in the past are reset, and the comparison of each history with the current data is visualized. As a result, the labor for searching for a desired model parameter from the history can be greatly reduced, so that the time required for parameter setting can be greatly shortened, and various simulation systems capable of executing simulation efficiently can be realized.

1 Plant 2 Plant Operation Data DB
3 Console screen 4 Plant simulator 5 Simulation result DB
6 Parameter history DB
7 Compliance rate calculator

Claims (6)

In a simulation system that adjusts parameters so that the output of the simulation target matches the result of the simulation while performing simulation based on the data input to the actual simulation target,
A simulation result database for storing the simulation results;
A parameter history database for storing a history of the parameters set during the simulation;
A simulation comprising: a result of the simulation performed on a model parameter corresponding to the output value of the simulation target from the parameter history database; and a matching rate calculation unit that calculates the output matching degree of the simulation target in time series system.   The simulation system according to claim 1, wherein the relevance ratio calculation unit calculates the amount of change in the relevance ratio between the simulation target output and the simulation result before and after the simulation model parameter change.   3. The simulation system according to claim 1, wherein the result of the precision ratio calculation unit is displayed as a bar graph of the precision ratio for the history.   2. The simulation system according to claim 1, wherein the simulation result is displayed as a bar graph of the history for each arbitrarily selected parameter.   The simulation system according to claim 1, wherein an arbitrarily selected history is displayed in a trend corresponding to the output of the simulation target.   The simulation system according to claim 1, wherein the simulation target is a process plant.

Images