JPH1115362A - Simulation method, and simulator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable the follow-up to the rapid input change with excellent accuracy by performing the time increment determination procedure to determine the simulation time increment immediately after an event related to the prescribed input value to a simulator is generated before the simulation is performed. SOLUTION: The time increment determination procedure to determine the simulation time increment for the simulation to be newly performed immediately after generation of an event related to one or two or more input values to a process simulator 4, is first performed according to the change in the input value from a controller 2 immediately after a time increment determining means 5 completes the communication. Then, the process simulator 4 performs the simulation with the simulation time increment obtained by performing the time increment determination procedure. Each simulation time increment of the subsequently continuous simulations is performed by the process simulator 4 with the simulation time increment determined in the immediately previous simulation.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a simulation method and a simulator for changing a simulation time interval according to a change amount of an output value of a simulator.
In particular, the present invention relates to an operation training simulator for operator training of a plant facility that needs to execute a simulation in real time and a simulation method thereof.

[0002]

2. Description of the Related Art Conventionally, in a simulation method and a simulator of this kind, when a simulation target is expressed by a mathematical model that cannot be analytically rigorously solved, generally a multivariable input, a multivariable output and a multivariable Is given, and a simulation is executed by numerically solving a mathematical model in which the relation between each variable is expressed by a state equation and an output equation by a computer. As a simulation time step when solving this state equation by a solution method such as the Euler method or the Runge-Kutta method, for example, 0.
A range such as 1 μs to 10 seconds is set, and the amount of change between the output value calculated by these solution methods and the old output value before simulation is evaluated based on a predetermined evaluation criterion,
If the amount of change as the evaluation result is small, the simulation time step is lengthened, and if the amount of change is large, the simulation time step is shortened and the simulation of the next step is executed sequentially. Is shortened. The output value of the simulator means a result obtained by executing the simulation. More specifically, when the state equation is solved, it means an internal state variable.

[0003]

However, in the conventional simulation method and simulator described above,
When the input value changes suddenly, the simulation time interval becomes inappropriate, the change cannot be made in time, the simulation result may be inconsistent with the actual operation, or the simulation itself may diverge. .
On the other hand, in order to solve these problems, a method of fixing the simulation time interval to a small value that can secure the simulation accuracy in advance has been put to practical use, but in this case, the simulation execution time becomes longer and the actual execution time becomes longer. It becomes difficult to apply to a simulator such as the above-mentioned driving training simulator which needs to execute a simulation in time. Further, even if the speed of the simulation is increased by simplifying the mathematical model, there is a possibility that the accuracy of the simulation is deteriorated by the simplified mathematical model. The present invention has been made in view of the above circumstances, and an object of the present invention is to solve the above-described problems, to execute a simulation at high speed, and to accurately follow a sudden input change. It is to provide a simulation method and a simulator.

[0004]

A first feature of the present invention for achieving this object is a simulation method for changing a simulation time interval according to a change amount of an output value of a simulator. As described in claim 1 in the column of (1), a time step determining procedure for determining a simulation time step immediately after the occurrence of an event relating to one or more input values to the simulator is performed before executing the simulation.

The second characteristic configuration relates to a simulation method of the above-mentioned first characteristic configuration, and as described in claim 2 of the claims, the input values relating to the event occurrence are assumed in advance. A table of the limit simulation time interval at which the simulation does not diverge for the maximum change amount is created in advance, and in the time interval determination procedure, the pre-correction simulation time interval set at the time of executing the simulation immediately before the event occurs is stored in the table. The point is that the correction is made based on the limit simulation time interval to make the simulation time interval immediately after the event occurs.

The third characteristic configuration relates to a simulation method of the above-mentioned second characteristic configuration. As described in claim 3 of the claims, in the time step determining procedure, an input value is set when an event occurs. For each input that has changed, the difference between the limit simulation time step and the pre-correction simulation time step is proportionally divided by the ratio between the actual input value change amount and the maximum change amount assumed in advance, or a predetermined safety coefficient. The multiplied value is added to the pre-correction simulation time step to calculate the post-correction simulation time step.If the input value changes with a single input when an event occurs, the post-correction simulation time step is used immediately after the event occurrence. If the input value changes with multiple inputs at the time of the event,
The point is that the minimum value of the post-correction simulation time increment for each input whose input value has changed or the minimum value multiplied by a predetermined safety coefficient is calculated and used as the simulation time increment immediately after the event occurrence.

[0007] The fourth characteristic configuration is the above-mentioned first, second, and third features.
Alternatively, with regard to the simulation method of the third characteristic configuration, as described in claim 4 in the claims section, the event occurs at a fixed time interval regardless of the presence or absence of a change in the input value.

The fifth characteristic configuration relates to a simulation time step variable type simulator that changes the simulation time step according to the amount of change in the output value of the simulator, as described in claim 5 of the claims. ,
The present invention is characterized in that a time interval determining means for determining a simulation time interval immediately after receiving one or more input values before executing the simulation is provided.

The sixth characteristic configuration relates to the simulator having the above-mentioned fifth characteristic configuration. As described in claim 6 of the claims, each input value at the time of receiving the input value is assumed in advance. Storage means for preliminarily storing a table of a limit simulation time interval at which the simulation does not diverge with respect to the maximum change amount, wherein the time interval determination means temporarily sets a pre-correction simulation time interval set at the time of the simulation immediately before receiving the input value; In that the simulation time step before correction is corrected based on the limit simulation time step in the table, and is output as a simulation time step immediately after receiving the input value.

The seventh feature configuration relates to the simulator having the sixth feature configuration described above, and as described in claim 7 in the claims section, the time step determining means sets the time interval for each input value. The difference between the limit simulation time step and the pre-correction simulation time step is proportionally divided by the ratio of the actual input value change amount and the previously assumed maximum change amount, or
A first corrected time step calculating means for calculating a post-correction simulation time step by adding a value obtained by multiplying the multiplied value by a predetermined safety coefficient to the pre-correction simulation time step, and when an input value changes in a plurality of inputs, A second corrected time step calculating means for calculating a corrected simulation time step by multiplying a minimum value or a minimum value of the calculation results for each input value of the post-correction time step calculating means by a predetermined safety coefficient; The point is to prepare.

The eighth characteristic configuration is the fifth, sixth,
Alternatively, regarding the simulator having the seventh characteristic configuration, as described in claim 8 in the claims section, the input value is received at fixed time intervals regardless of the presence or absence of a change in the input value.

The operation will be described below. According to the first or fifth feature configuration, before and after the change of the input value, the simulation time interval used for the simulation immediately after the change of the input value set by executing the simulation immediately before the change of the input value is used as it is. Instead, the simulation time interval is appropriately corrected or a new appropriate value is given in accordance with the change in the input value, so that the occurrence of an extreme simulation error and the divergence of the simulation can be avoided. Simulation of accuracy can be performed. Of course, since the simulation time step can be maintained in a variable simulation time step type simulator in which the simulation time step changes according to the amount of change in the output value of the simulator, it is possible to avoid unnecessary increase in the simulation execution time, and to achieve high accuracy and high speed execution. Sex can be balanced.

According to the second or sixth characteristic configuration, the table stores the limit simulation time interval corresponding to the maximum change amount of each input value.
Based on the limit simulation time step and the previously set simulation time step when there is no change in the input value, the optimum simulation time step according to the actual input value change amount is calculated from the input value change amount. It can be set.

Further, according to the third or seventh feature configuration, it is possible to easily calculate the optimum simulation time interval according to the amount of change of the actual input value by the second or sixth feature configuration. Therefore, the calculation time required for correcting the simulation time increment can be shortened, and the overall execution processing time can be shortened.

According to the fourth or eighth aspect, when a specific process to be simulated in a real system is a control target of a specific controller in the real system and is controlled at a fixed time interval, The control amount of the controller can be applied to the simulation of the system as an input value of the simulation. For example, an operation training simulator for operator training of a plant facility that needs to execute a simulation in real time is composed of a controller that controls the plant facility and a simulator that models the control target of the controller. The configuration can be applied.

[0016]

Embodiments of the present invention will be described below with reference to the drawings.

As shown in FIG. 1, an operation training simulation system for operator training of a plant facility includes an operation training simulator 1 which is a simulator for executing the simulation method according to the present invention, a controller 2, and a system bus connecting these. 3 The operation training simulator 1 includes a process simulator 4 for simulating various processes generated in an actual plant facility and a time step determining means 5 for executing a time step determining step S2 for supplying an initial value of a simulation time step to the process simulator 4. It consists of. Specifically, the driving training simulator 1, the process simulator 4 and the time step determining means 5 are configured by a general-purpose computer 6 such as an engineering workstation capable of executing a program for executing the simulation method according to the present invention. Is done. The controller 2 is the same as the controller used in the actual plant facility, but the general-purpose computer 6 may be equivalent. In this case, the system bus 3
Substantially, it is formed in the general-purpose computer 6.

The driving training simulator 1 and the controller 2 communicate with each other via the system bus 3 at intervals of one second. The communication interval can be arbitrarily changed according to the actual system. Specifically, as shown in FIG.
A part of the unit time set by the communication interval is allocated to a communication time between the driving training simulator 1 and the controller 2, and a process value which is an output value of the process simulator 4 and a simulation result is obtained within the communication time. The controller 2 outputs an operation amount corresponding to the process value as an input value of the process simulator 4, and causes the process simulator 4 to generate an event S1. Further, the input value is also input to the time step determining means 5.

As shown in FIG. 2, the portion after the communication time in the unit time is mainly used by the driving training simulator 1 for executing a simulation. More specifically, the time step determining means 5 performs a simulation time step for a simulation to be newly executed immediately after the event occurrence S1 in accordance with the change amount of the input value from the controller 2 immediately after the end of the communication. Is executed first, and then the process simulator 4 starts the execution S3 of the simulation at the simulation time interval obtained by executing the time interval determination procedure S2. For each simulation time step of the simulation that continues thereafter within the unit time, the process simulator 4 executes the simulation at the simulation time step determined at the time of the immediately preceding simulation as in the related art, and according to the amount of change in the output value. The procedure for determining the simulation time interval of the next simulation is repeatedly executed, and a series of simulation execution S3 of the process simulator 4 is completed. In the case of the simulation method or the simulator according to the prior art, the use division of the unit time is
It is shown in FIG. 3 for reference.

Hereinafter, a method of determining the simulation time interval by the process simulator 4 will be described.
The process to be simulated by the process simulator 4 is represented by a mathematical model composed of a state equation represented by Equation 1 and an output equation represented by Equation 2. The Euler method is adopted as a method of solving the equation of state.

[0021]

## EQU1 ## dX = A * X + B * U

[0022]

## EQU2 ## Y = C * X + D * U

[0023] However, X is i-th internal state vector with state variables x _i component, dX internal state differential value vector with i-th time derivative dx _i components of the state variable, U
Input vector to the i-th input variable u _i component, Y
Is an output vector having the i-th output variable y _i as a component, A
Denotes a system matrix, B denotes a control matrix, C denotes an observation matrix, and D denotes a direct matrix. For Euler, the differential dx _i is replaced by forward difference _{(x i (t + h)} -x i (t)) / h. H at this time is a simulation time interval.

[0024] The next simulation simulation time step h _NEXT, per differential value dx _i of each state variable of the internal state differential value vector dX of the current simulation results to calculate all the evaluation value e _i shown in equation 3, equation When the determination process of Equation 4 is performed and Expression 4 is satisfied, the simulation time interval h of the current simulation is reduced to half, and when Expression 4 is not satisfied and Expression 5 is satisfied, the simulation time interval h is set to 2 It is derived by doubling. If Equations 4 and 5 are not satisfied, the simulation time step h _NEXT maintains the simulation time step h of the current simulation. Note that in Equation 3 to 5 below, dx _i-OLD is the differential value dx _i at the previous simulation, T is a constant that is set in advance, has a function of adjusting the simulation accuracy. Also, max
(E) is a function value to select the maximum value of the components of the evaluation value vector E to component evaluation value e _i.

[0025]

[Equation 3] e _i = (dx _i −dx _i-OLD ) / (| dx
_i-OLD | + T)

[0026]

## EQU4 ## h * max (E)> 10 * T

[0027]

H * max (E) <T

Next, a method of determining the simulation time step by the time step determining means 5 will be described. Said time increment determining means 5, and the maximum variation Delta] u _MAXi of each input value u _i from the controller 2, the limit simulation time step h _Xi limit the simulation when the maximum change does not diverge measured in advance to each other The limit simulation time increment h _Xi is stored in a table in the storage means 7 provided in the general-purpose computer 6.

Of the input values u _i , the controller 2
Only those actually changed during communication with, and calculates an input value u _i by the simulation time step h _i to each other by calculating Equation 6. Here, in Expression 6, h _0i is a simulation time interval determined by the procedure of Expressions 1 to 5 in a simulation immediately before communication with the controller 2, and Δu _i is an actual input value at the time of communication with the controller 2. a variation of u _i, a ₁ is one or more valued in the safety factor.

[0030]

H _i = h _0i + ((h _Xi −h _0i ) * (Δu _i / Δ
u _MAXi )) * a ₁

[0031] Of the input values u _i, the controller 2
If the input that has actually changed at the time of communication with is one input, let h _i in Equation 6 be the simulation time interval h. If two or more inputs change, the simulation time increment h is further determined based on Equation 7. However, in Equation 7, min (H) is a function value for selecting the minimum value among the components of the simulation time step vector H having the simulation time step h _i as a component for each input value u _i , and a ₂ is a safe value. The coefficient is a value of 1 or less. .

[0032]

H = min (H) * a ₂

Note that in Equations 6 and 7, when the safety coefficient a ₁ or a ₂ is fixed at 1, each multiplication is unnecessary.

Hereinafter, another embodiment will be described. <1> In the above embodiment, the communication interval is a fixed time
Occurs at intervals and necessarily output by the controller 2
Configuration in which the input value of the process simulator 4 does not change
But at least one of the input values changes
The driving training simulator 1 and the controller
Event is generated at irregular intervals so that communication between
You may let it. <2> Formulas incorporated in the process simulator 4
The model is not necessarily expressed in the form of Equations 1 and 2.
It does not matter. The solution of the equation of state is less than Euler's method.
It can be outside. For example, using the Runge-Kutta method
Is also preferred. <3> Each input value u from the controller 2_iMaximum change
Amount Δu_MAXiLimit simulation time step corresponding to
h_XiAnd the stain determined in the simulation immediately before communication.
Hour time increment_0iFrom simulation time
The method for determining h is not necessarily based on Equations 6 and 7.
You don't have to. For example, the input value u _iAccording to the characteristics of
Therefore, without performing the linear interpolation as in Equation 6, the nonlinear interpolation is performed.
You can go between them. Also, the safety coefficient a₁All
Input value u_iTo the input value u_iAnother
It may be changed to. Further, the safety factor a_TwoAlso real
Input value u that has changed_iOr change according to the number of inputs
You can do it. <4> The time step determination step S2 is not necessarily performed by each input.
Value u_iMaximum change amount Δu _MAXiAnother said limit simulation
Time interval_XiYou don't have to use
Absent. For example, each input value u_iInput pattern when the maximum change of
Measure the limit simulation time interval for each
May be used. Also, each input value u_iAmount of change
Simulation time increment h directly from the neural network
It may be a configuration that can be determined by applying
No.

<5> In the above embodiment, the process simulator 4 and the time step determining means 5 are constituted by a general-purpose computer 6 such as an engineering workstation capable of executing a program for executing the simulation method according to the present invention. Instead, a dedicated processor may be used separately or integrally.

<6> In the above-described embodiment, a system in which the controller 2 exists is assumed by taking an operation training simulation system for operator training of a plant facility as an example. However, the controller 2 does not necessarily need to exist. Alternatively, the mathematical expression model itself may include a feedback system.

[0037]

As described above, according to the present invention,
Even if the input value to the simulator changes abruptly, the simulation can be executed with the simulation time interval appropriately set by the execution of the time interval determination means or the time interval determination procedure. As in the past, a simulation method that changes the simulation time interval according to the amount of change in the output value of the simulator could be used. FIG.
FIG. 3 illustrates the difference between the output value of the simulation result according to the present embodiment and the simulation result obtained by the conventional method shown in FIG. As a result, according to the present invention, it is possible to provide a simulation method and a simulator that can execute a simulation at a high speed and can accurately follow a sudden input change. Furthermore, when the time interval of the event occurrence is fixed to a fixed time, the effect of shortening the simulation time is that the idle time after the execution of the simulation can be effectively used by the computer for executing the simulation for executing other functions. The effect can be expected.

In the claims, reference numerals are provided for convenience of comparison with the drawings, but the present invention is not limited to the configuration shown in the attached drawings.

[Brief description of the drawings]

FIG. 1 is a block diagram of a simulator according to the present invention.

FIG. 2 Time step determination procedure and simulation according to the present invention
Diagram showing the time relationship between the action execution and the event occurrence

FIG. 3 is a timing chart showing a time relationship between a simulation execution and an event occurrence in the related art.

FIG. 4 is a comparison diagram of simulation results between a simulation method according to the present invention and a conventional simulation method.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Driving training simulator 2 Controller 3 System bus 4 Process simulator 5 Time step determining means 6 General-purpose computer 7 Storage means S1 Event occurrence S2 Time step determining procedure S3 Simulation execution

Claims (8)

[Claims] 1. A simulation method for changing a simulation time step according to a change amount of an output value of a simulator, wherein the simulation time step is determined immediately after an event (S1) relating to one or more input values to the simulator. A simulation method characterized in that a time step determining step (S2) to be performed is executed before executing a simulation (S3). 2. A limit simulation time interval table in which the simulation does not diverge for a previously assumed maximum change amount of each input value related to the event occurrence (S1) is created in advance, and the time interval determination procedure ( S2)
In the above, the simulation time interval before correction set at the time of the simulation execution (S3) immediately before the event occurrence (S1) is corrected based on the limit simulation time interval of the table to be the simulation time interval immediately after the event occurrence (S1). The simulation method according to claim 1. 3. In the time step determining step (S2), a difference between the limit simulation time step and the pre-correction simulation time step is calculated for each input whose input value has changed at the time of the event occurrence (S1). And proportionally divided by the ratio of the maximum change assumed in advance, or
A value obtained by multiplying the calculated value by a predetermined safety coefficient is added to the simulation time step before correction to calculate the simulation time step after correction. If the input value changes with a single input when the event occurs (S1), the correction is performed. The post-simulation time step is a simulation time step immediately after the event occurrence (S1). If the input value changes by a plurality of inputs at the time of the event occurrence (S1), the post-correction simulation time for each input whose input value has changed 3. The simulation method according to claim 2, wherein a minimum value of the step or a value obtained by multiplying the minimum value by a predetermined safety coefficient is calculated, and the calculated value is used as a simulation time step immediately after the occurrence of the event (S1). 4. The event according to claim 1, wherein the event occurs at a fixed time interval regardless of whether or not the input value has changed.
The described simulation method. 5. A simulation time step variable type simulator that changes a simulation time step according to a change amount of an output value of the simulator.
A simulator provided with time step determining means (5) for determining a simulation time step immediately before receiving the above input value before executing the simulation (S3). 6. A storage means (7) for storing in advance a limit simulation time step table in which the simulation does not diverge with respect to a previously assumed maximum change amount of each input value at the time of receiving the input value, and the time step determination is performed. Means (5)
Temporarily stores the pre-correction simulation time step set at the time of the simulation execution (S3) immediately before receiving the input value, and corrects the pre-correction simulation time step based on the limit simulation time step in the table to input the input value. The simulator according to claim 5, wherein the simulator outputs the simulation time interval immediately after reception. 7. The time step determining means (5) calculates, for each input value, a difference between the limit simulation time step and the pre-correction simulation time step between an actual input value change amount and a previously assumed maximum change amount. A first corrected time step calculating means for calculating the corrected simulation time step by adding a value proportionally divided by the ratio or a value obtained by multiplying the ratio by a predetermined safety coefficient to the pre-correction simulation time step; In the case of a change in the input of (1), a minimum value of the calculation results for each input value of the post-correction time step calculation means or a minimum safety value is multiplied by a predetermined safety coefficient to obtain a post-correction simulation time step. 7. The simulator according to claim 6, comprising a time step calculating means after two corrections. 8. The simulator according to claim 5, wherein the input value is received at a fixed time interval regardless of whether the input value has changed.