JPH0644294A - Signal generating device, simulation device, and controller - Google Patents

Abstract

PURPOSE:To improve simulation and control precision by generating a disturbance signal which is close to an actual phenomenon. CONSTITUTION:A chaos function generating mechanism 11 generates a chaos function on the basis of actually measured time-series data. Several candidates for the chaos function are previously given and the chaos function generating mechanism 11 determines the parameter of a function by a method of least square, etc., and determines the function by using a pattern recognizing method. A chaos signal generating mechanism 12 generates time-series data on the basis of the generated chaos function. Various disturbance signals can be generated by combining functions or varying an initial value. A simulator 13 performs simulation by using the generated time-series data as a disturbance input. Thus, the disturbance signal which is close to an actual phenomenon is easily generated on the basis of the chaos function, so the simulation result and control precision can be improved.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a simulation device for generating a signal close to a disturbance signal for simulation and a control device for actual control, and particularly, to easily generate a signal having a property close to an actual phenomenon. The present invention relates to a signal generation device, a simulation device, and a control device that are suitable for controlling.

[0002]

2. Description of the Related Art When simulating the response of a system to a disturbance signal, it is desirable to use a disturbance that is close to an actual phenomenon. Therefore, there is a method of measuring the disturbance of an actual phenomenon and using the measured time series data as it is for the simulation.

On the other hand, there is also a method in which a probabilistic model is created under the assumption that the disturbance fluctuates to some extent, and time series data is generated based on the stochastic model and used as a disturbance signal. For example, in the case of simulating a frequency fluctuation with respect to a load fluctuation disturbance of a power system, a disturbance signal using a stochastic model based on a Markov process as discussed in IEICE Transactions B, Vol. 111, No. 5, pp. 570-571. There is a method of occurrence.

[0004]

Of the above-mentioned conventional techniques,
The method that uses the measured time series data as it is requires a large amount of data to be stored when simulating a long phenomenon or several cases, and thus a variety of disturbance signals cannot be generated. There is. Further, the method using the probabilistic model has a problem that it is difficult to generate a disturbance signal close to an actual phenomenon.

An object of the present invention is to provide a signal generator or the like which can easily generate a disturbance signal close to an actual phenomenon without the need to store a large amount of data.

[0006]

[Means for Solving the Problems] The above-mentioned object is to create time-series data as a simulated signal based on the created chaotic function based on the actually-measured time-series data. This is achieved by performing simulation and control by generating the time series data from the means and using the generated time series data as a disturbance signal.

[0007]

The time series data that is actually measured is stored in the storage device, and this time series data is read out and input to the chaos function creating means. When only the necessary frequency components are to be the target of the chaos function, the time series data is input to the chaos function creating means through the frequency filter.

The chaotic function creating means selects, from among several prepared chaotic function candidates, one that is closest to the property of the measured time series data. Alternatively, the parameters of the function prepared in advance are optimally set.

The signal generating means selects one from a plurality of chaotic functions or uses a plurality of functions in combination as required. The combination of functions can be changed according to the situation. Further, it is preferable that an auxiliary signal generating means is provided as necessary so that an appropriate value such as a random number can be added to the time series data generated based on the chaotic function.

Since the nature of the original disturbance signal is preserved to some extent in the chaotic function created based on the measured time series data, a signal close to an actual phenomenon can be generated. The time-series data required to create the chaos function does not have to be that long, so a large amount of data storage is not required. If a plurality of chaotic functions are prepared according to the situation, the functions can be switched or combined as needed. Furthermore, by changing the initial value, it is possible to easily generate a time series having the same property but different values.

If the generated time-series data is used as a disturbance input, a simulation close to an actual phenomenon can be easily performed. If used as a disturbance input when designing the adaptive control system, optimum parameters can be easily set according to the situation. Further, if it is incorporated in an online control system and used as a near future disturbance prediction value, it is possible to perform feedforward control for suppressing the influence of the disturbance.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a configuration diagram of a system in which an embodiment of the present invention is applied to frequency control simulation of a power system. The frequency control simulation simulates how the frequency changes in response to a load fluctuation disturbance of several seconds to one minute of the system.

First, the generated power measured by the generated power measuring device 101 installed at each power plant and the frequency measuring device 10
The system frequency measured by 2 is the load fluctuation data creation device 1
It is sent every hour to 03. Load fluctuation data creation device 10
In 3, the load fluctuation disturbance of the entire system is created as time-series data by adding the change amount of the generated power to the change amount of the system frequency. It should be noted that the generated amount may be created only by the frequency change without adding the changed amount. The created time series data is
It is stored in the storage device 104 once.

The data reading device 105 is a storage device 104.
The time series data stored in the
To enter. The filter 106 filters the data so that only the fluctuation of the period of several seconds to 1 minute, which is the target of frequency control, remains. The time series data that has passed through the filter 106 is input to the chaos function creating mechanism 11.

The chaos function creating mechanism 11 creates a chaos function most suitable for the input time series data. There are various creation methods, and an example of the procedure will be described with reference to FIG. First, the time series data is considered as a function of time v (t) as shown in FIG. An arbitrary value v0 is selected, and times t1, t2, ... Of points crossing v0 in the positive direction are obtained (step 2).
1). FIG. 3 shows t1 and t2 when v0 = 0. Since the actual time series data is a sample value, it is obtained by using an interpolation method such as linear interpolation. Next, select an appropriate time τ, v (t1 + τ), v (t2 + τ),
Is obtained (step 22). These values can also be obtained from given time series data by some interpolation method. In addition, regarding how to select the optimum τ, see the literature,
Fraser and Et Swinney's "Independent Coordinates for Strange Attractors From Mutual Information" Physical Review A, 33, 1134 (A.
Fraser and H. Swinney, "Independent coordinates
for strange attractors from mutual informat
ion, "Physical Review A, vol.33, p.1134).

Here, x (n) is defined as in the following equation 1 and the coordinate value (x (n), x (n + 1)) is obtained (step 2).
3).

[0017]

## EQU1 ## x (n) ≡ v (tn + τ) x (n) and x (n + 1) are plotted on a plane having axes as axes, and as shown in FIG. Then,
If this can be described by a function of the form shown in the following Expression 2, a chaotic function can be created.

[0018]

## EQU00002 ## x (n + 1) = f {x (n)} However, it is quite difficult to automatically determine the functional form of f in practice, so prepare some function candidates,
A possible method is to select the most suitable one.

Therefore, a function including some parameters is prepared, and the value of the parameter is set so as to best fit the coordinate value obtained by the above method for each function. As the function, for example, the following expression 3 can be considered.

[0020]

## EQU00003 ## x (n + 1) = a.x (n) .multidot. {1-x (n)} where a is a parameter.

As a parameter setting method, for example, the least square method or the like may be used so that the sum of squares of the difference between the function value and the coordinate value is minimized (step 2).
4). At that time, it is also possible to calculate how much the identification error between the obtained function and the coordinate value is. Finally, the identification error of each function is compared, and the chaotic function that best fits is selected (step 25). As described above, the method of presetting some function candidates has an effect of easily creating a chaotic function.

In the above example, a method for obtaining a chaotic function from a simple two-dimensional attractor was shown, but as to a method for identifying a chaotic function from a three-dimensional attractor, a document "Nonlinear Mechanics and Chaos", Translated by Musashi Toshimitsu, Hashiguchi See Sumikyu, Ohmsha, pp. 329-333 for details.

In the above example, the chaotic function is given in the form of equation 2, but any other form such as the one given by the time function or the differential equation can be used as long as it can be identified from the time series data. Needless to say, the function of can be used. Also, the number of variables is not limited to one, and there may be two or more.

Next, the chaotic signal generating mechanism 12 generates time series data based on the chaotic function created as described above. For example, if the chaos function is defined in the form of Equation 2, if only the initial value x (0) is determined, then the time series data x (1), x (2), ... Is created based on the formula. be able to. Also, if given as a differential equation, this can also be numerically solved to generate time series data if an initial value is determined. As the initial value, a method of determining using a random number, a method of setting by a user, etc. can be considered. By changing the initial value variously, various disturbance signals can be generated.

As the chaos function, one function may be used, but it is also possible to use a combination of several functions depending on the case. For example, when the power load fluctuation is represented by the sum of chaotic functions determined for each region, the combination of chaotic functions is changed according to the connection state of the system.

Next, the filter 10 is used when creating the chaos function.
In order to supplement the component removed in 6, the disturbance components are generated in the long-period disturbance generation mechanism 107 and the short-period disturbance generation mechanism 108, respectively, and added to the time series data. Since these auxiliary signal generation mechanisms do not need to be based on chaotic functions, for example, the short-term disturbance generation mechanism 107 is a probabilistic model-based method.
The long-period disturbance generation mechanism 108 generates disturbance by a method based on load prediction. Of course, it may be generated by a method based on a chaotic function.

As described above, if a method is used in which a chaos function is passed through a filter and the disturbance is compensated for by another auxiliary signal generating mechanism, only the necessary frequency component can be represented by the chaos function, which is more realistic. There is an effect that a chaotic function is easily obtained.

The time series data generated by the above method is
It is given to the simulator 13 as a disturbance power fluctuation input, and the simulator 13 calculates a frequency fluctuation corresponding to the disturbance.

According to the present embodiment, there is an effect that the frequency control simulation of the power system using the load power fluctuation disturbance which is close to the actual one can be easily performed.

In the above example, a mathematical programming method such as the least squares method is used for obtaining the chaotic function, but the chaotic function can also be obtained by the pattern recognition method. For example, select an appropriate time τ and set time series data to v (t) and v (t).
A locus called an attractor is drawn by plotting it on a plane consisting of two axes (t + τ). By comparing the attractor of the chaotic function prepared in advance with the attractor drawn based on the time series data,
The closest chaotic function can be selected. When comparing attractors, a pattern recognition method such as a neural network can be used. Not only two-dimensional attractors but also three-dimensional and four-dimensional attractors can be considered.

Furthermore, by using a neural network that inputs time series data and outputs the type and parameters of a chaotic function, it is possible to directly select a chaotic function without drawing an attractor.

Next, an example in which the signal generator of this embodiment is applied to the parameter setting of the controller will be described with reference to FIG. The simulator 13 has a controlled object model and a control device model, and performs simulation for disturbance fluctuations. The parameter setting mechanism 52 optimally sets the parameters of the controller model while watching the simulation output. As the disturbance, the chaotic signal generating mechanism 12 generates time series data by using the chaotic function created by the chaotic function creating mechanism 11. The initial value of the chaos function is automatically set by the initial value setting mechanism 51 using a random number.

According to this method, since a disturbance that is close to the actual one can be easily generated, there is an effect that the parameters of the control device suitable for the actual control can be easily set.

Further, the above method is applied to the frequency control of the power system to generate disturbances in some situations such as morning, daytime, night, weekdays, and holidays by using a chaotic function, and in each situation, It is also conceivable to find the optimum control policy for the above and selectively use the control policy according to the situation. According to this method, there is an effect that the optimum control design and adaptation according to the situation can be easily performed.

If the time series data of the disturbance is obtained online, the chaos function is created online,
It is also possible to automatically select and control the control policy corresponding to the created chaotic function. According to this method, there is an effect that the optimum control measure corresponding to the situation can be automatically selected online.

Next, an example in which the present invention is incorporated into an online feedforward control system will be described with reference to FIG. The disturbance measuring device 63 measures the disturbance signal input to the controlled object 62 and supplies it to the chaos function creating mechanism 11 as time series data. The chaotic function measuring mechanism 11 creates a chaotic function based on the time series data and gives it to the chaotic signal generating mechanism 12. Since the chaos signal generation mechanism 12 knows the value of the disturbance at the current time point from the signal from the disturbance measuring device 63, it can generate the predicted value of the disturbance in the near future by using the chaotic function. The control device 61 performs feedforward control that cancels the influence based on the disturbance prediction value generated by the chaotic signal generation mechanism 12.

Incorporation into an online control system in this way makes it possible to execute control that predicts near-future disturbances and cancels out their effects.

[0038]

According to the present invention, there is an effect that a disturbance having a property close to an actual phenomenon can be easily generated. Nor does it require a large amount of data storage. Furthermore, there is an effect that various disturbances can be easily generated by combining chaotic functions and changing various initial values.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a system in which a signal generator according to an embodiment of the present invention is applied to frequency control simulation.

FIG. 2 is a flowchart showing a procedure of creating a chaotic function in the chaotic function creating mechanism shown in FIG.

FIG. 3 is a diagram showing an example of time variation of time series data.

FIG. 4 is a diagram showing an example of mapping of a chaotic function.

FIG. 5 is a configuration diagram when the signal generator according to the embodiment is applied to parameter setting of an adaptive control system.

FIG. 6 is a configuration diagram when the signal generator according to the embodiment is applied to an online control system.

[Explanation of symbols]

11 ... Chaotic function creating mechanism, 12 ... Chaotic signal generating mechanism, 13 ... Simulator, 61 ... Control device.

Front Page Continuation (72) Inventor Hiroo Konishi 4026 Kuji Town, Hitachi City, Hitachi, Ibaraki Prefecture Hitachi Research Laboratory, Inc. (72) Inventor Yasushi Harada 4026 Kuji Town, Hitachi City, Ibaraki Hitachi Research Laboratory, Ltd.

Claims (10)

[Claims] 1. A chaotic function creating means for creating a chaotic function based on actually measured time series data, and time series data as a simulated signal is generated based on the chaotic function created by the chaotic function creating means. A signal generating device comprising: a signal generating means. 2. The chaotic function creating means according to claim 1, further comprising means for selecting one or a plurality of chaotic functions from a plurality of pre-given chaotic functions based on actually measured time series data. A signal generator comprising: 3. The signal generating apparatus according to claim 1, wherein the chaos function creating means includes means for determining a parameter of a chaos function given in advance, based on actually measured time series data. . 4. The chaotic function generating device according to claim 1, wherein a filter is provided in the preceding stage of the chaotic function generating means, and the actually measured time series data is passed through the filter for removing components in a certain frequency region. A signal generator characterized by inputting to a means. 5. The signal generating device according to claim 1, wherein the signal generating means includes means for combining a plurality of chaotic functions to generate time series data as a simulated signal. 6. The signal generating device according to claim 5, wherein the signal generating means includes means for changing a combination of chaotic functions depending on a situation. 7. The apparatus according to claim 1, further comprising: auxiliary signal generating means, and means for adding the value generated by the auxiliary signal generating means to the time series data generated by the signal generating means based on a chaotic function. Signal generator. 8. A simulation apparatus comprising the signal generator according to claim 1, wherein the simulation is performed by using a simulated signal output from the signal generator as a disturbance. 9. The signal generator according to claim 1, wherein the time-series data as the simulated signal generated by the signal generator is used as an input signal for setting a parameter of the controller. 10. A time-series data as a simulated signal, which is equipped with the signal generator according to claim 1 and is output online by the signal generator based on time-series data measured online. A control device characterized in that a control signal is applied to a control system as a feedforward signal by using.