JP2006107256A - Automatic fitting apparatus of model parameter - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an automatic fitting apparatus of a model parameter for precisely updating a parameter by eliminating inappropriate data included in actual value from a plant in updating a simulation model parameter. <P>SOLUTION: The model parameter fitting apparatus comprises: a data storage means 3 for storing data to be used for adjusting a parameter of a mathematical or physical model to simulate plant behavior; a data filtering means 6 for eliminating the inappropriate data from the data to be used for adjusting the mathematical or physical model and its parameter to simulate the plant behavior which are given by the data storage means; a parameter adjusting means 9 for adjusting the parameter of the mathematical or physical model to simulate the plant behavior from data given from the data storage means via the threshold data filtering means; and input/output means 2 and 10 for performing input output processing of the parameter adjusted by the parameter adjusting means. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention is intended to realize an operation satisfying both safe and efficient operation of a plant such as a water and sewage plant, an industrial plant such as a steel plant, a district heating and cooling plant, a heat source plant for each building, and a power plant. In particular, in order to judge whether it is possible to realize the operation that satisfies the plant safety or efficiency or both from the future plant behavior by simulating the plant behavior. The present invention relates to a model parameter automatic fitting apparatus that adjusts a parameter of a plant simulation model using a mathematical model or a physical model from a measured value or numerical data that can be substituted for the measured value.

  When the operator makes a decision on the plant operation policy, the computer simulates the future behavior of the plant and provides support information. In this case, some of the actual measurement data used to adjust the parameters of the simulation model includes inappropriate data, that is, data that includes missing measurement data and measurement errors of the measuring instrument. There is a need.

In addition, after the adjustment is completed, the characteristic parameters may change due to aging of the process or the like during use. In this case, it is necessary to determine whether or not the characteristic parameter needs to be readjusted, and to adjust from the measurement data with specialized knowledge.
JP 2002-328702 A

  Such adjustment takes a lot of time and effort, and some countermeasure is required.

Therefore, accurate parameter adjustment is performed by automatically removing inappropriate data such as missing data and measurement errors in the measured data, and there is a discrepancy between the simulated behavior caused by secular change and the measured value. Therefore, it is desired to realize an automatic fitting device for model parameters that can be adjusted without the hands of a professional engineer.

  The present invention has been made in consideration of the above-mentioned points, and when updating simulation model parameters, it is possible to remove inappropriate data included in the actual measurement values from the plant and update the parameters accurately. Another object of the present invention is to provide an automatic model parameter fitting device that can calculate an error between an actual measurement value and a simulation result to determine when to update a simulation model parameter and automatically update the parameter.

In order to achieve the above object, in the present invention,
Data storage means for storing data used for adjusting parameters of a mathematical or physical model for simulating plant behavior, a mathematical or physical model for simulating plant behavior, which is provided from the data storage means, and Data filtering means for removing inappropriate data included in actual measured values from the plant from data used for adjusting parameters, and plant behavior from data given from the data storage means via the threshold data filtering means. Model parameter fitting characterized by comprising parameter adjusting means for adjusting parameters of a mathematical or physical model to be simulated, and input / output means for performing input / output processing of parameters after being adjusted by the parameter adjusting means apparatus,
Is to provide.

  As described above, the present invention removes inappropriate data contained in actual measured values from a plant by filtering from data used to adjust parameters of a mathematical or physical model that simulates plant behavior. Model that can update the parameters automatically, calculate the error between the measured value and the simulation result by simulating the plant behavior, determine when the simulation model parameter needs to be updated, and update the parameter automatically. An automatic parameter fitting device can be provided.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. Here, a water supply process of a waterworks plant is taken as an example as a specific application target.

Embodiment 1

(Configuration of Embodiment 1)
FIG. 1 is an overall configuration diagram showing Embodiment 1 of the present invention. In the water treatment plant, the treated water is sent from the water purification pond to the water distribution pond, and the water is supplied to each customer from the water distribution pond. FIG. 1 shows this process and the model parameter fitting apparatus in association with each other.

Here, the model parameter fitting apparatus has the following means (1) to (8) as main components. That is,
(1) Data storage means 3 for storing data used to adjust parameters of a mathematical or physical model that simulates plant behavior;
(2) Parameter adjusting means 9 for adjusting parameters of a mathematical or physical model for simulating plant behavior from data stored in the data storage means;
(3a) Threshold data filtering means 6 for removing inappropriate data according to a threshold value among mathematical or physical models simulating plant behavior and data used for adjusting parameters thereof;
(3b) Differential data processing means 6 for performing differential data processing based on the difference between the data before an arbitrary time and the current time among the data used to adjust the parameters of a mathematical or physical model that simulates the plant behavior When,
(3c) threshold data filtering means 6 for removing inappropriate data from the data after the differential data processing by the threshold;
(3d) Among the data used to adjust the parameters of mathematical or physical models that simulate plant behavior, statistical processing means for filtering by statistical processing of the data, and the threshold for the data after statistical processing Threshold data filtering means 6 for removing inappropriate data;
(3e) Differential data processing means 6 that performs differential data processing based on the difference between the data before an arbitrary time and the current time among the data used to adjust the parameters of a mathematical or physical model that simulates plant behavior ,
(3f) statistical data processing means for determining a threshold value by statistical processing, and threshold data filtering means 6 for removing inappropriate data using threshold values determined by differential data processing and statistical data processing;
(3g) Frequency data filtering means 6 that removes inappropriate data using a frequency filter among data used to adjust parameters of a mathematical or physical model that simulates plant behavior;
(3h) self-organizing map data filtering means 6 for removing inappropriate data by self-organizing map among data used for adjusting parameters of a mathematical or physical model that simulates plant behavior;
(4) There are multiple means for filtering the data used to adjust the parameters of the mathematical or physical model that simulates plant behavior, and which means should be used depending on the characteristics of the data to be processed. Filtering manual selection means 5 that enables selection;
(5) Input / output means 2 and 10 for performing input / output processing of parameters after adjustment;
(6) Plant simulation means 7 for simulating plant behavior by a mathematical or physical model,
(7) Update determination means 8 for determining based on a square error and a threshold value between measured data of a plant that simulates whether or not model parameter adjustment is required and a simulation result simulated by a parameter before parameter adjustment;
(8) Update determination means 8 capable of updating model parameters at an arbitrary timing;
It is.

  With the configuration as described above, it is possible to perform accurate parameter update by removing inappropriate data included in the actual measurement values from the plant used when updating the simulation model parameters. The parameter can be automatically updated by calculating the error between the simulation result and the simulation result and determining when the simulation model parameter needs to be updated.

(Operation of Embodiment 1)
In the waterworks plant, raw water is taken from rivers, etc. with a water pump, and after passing through the wells, chemicals are injected into the mixing basin for coagulation, sedimentation, and filtration. Gradually enough flocs are formed to settle (floc ponds) and filth settles (sedimentation basins). Supernatant water is filtered through a filtration pond, sterilized with chlorine, etc., passed through a water purification pond, and distributed to a distribution area (customer) via a distribution pond by a water pump.

  In the operation management of waterworks plants, it is an absolute proposition for operators to stably supply water without shutting off water demand. For this reason, the operator must supply and supply raw water that matches the amount of chemical injection for water treatment at the water treatment plant and the required amount of water (transmission and distribution) (the capacity and water level of each pond, the pump The plant is operated, controlled, and managed in consideration of the characteristics of the component equipment such as pipes and valves. At that time, since the operation policy is decided while assuming the demand for purified water in the future, if the behavior of the process can be grasped in advance, the decision can be made easily.

  In order to grasp the behavior of the process in the future in advance, a mathematical or physical model that can simulate the process behavior is required. The model has characteristic parameters that describe the characteristics of each process, and it requires expert knowledge about the process model to adjust it so that it can follow the current process behavior with sufficient accuracy. Absent.

  Therefore, when such a function is incorporated in the system, it is necessary to adjust the measurement data with specialized knowledge when simulating the process behavior for the first time. However, some measurement data includes missing values and measurement errors of the measuring instrument, and it is necessary to make adjustments by removing the inappropriate data. In addition, after the adjustment is completed, the characteristic parameter may change due to the secular change of the process during use. In this case, it is necessary to determine whether or not the characteristic parameter needs to be readjusted, and to adjust from the measurement data with specialized knowledge.

  FIG. 1 shows the configuration of Embodiment 1 of the present invention. In this case, the plant modeling device 1 is applied to a water supply process of a water supply plant. That is, in the water supply process, the pump 22, the electromagnetic valve 23, the flow meter 24, and the water level meter 25 are provided in the water supply system from the distribution reservoir 21a to the distribution reservoir 21b.

  The plant modeling apparatus 1 includes an input unit 2, a data storage unit 3, a GUI 4, a filtering unit selection unit 5, a filtering unit 6, a plant simulation unit 7, an update determination unit 8, a parameter adjustment unit 9, an output unit 10, and remote data. Input means 11 is provided. Hereinafter, main ones will be described.

で表すこととする。なお、

Nは計測データセットの数である。 Parameter adjusting means 9
The parameter adjustment means 9 used in the present invention adjusts the parameters of a mathematical or physical model that simulates the plant behavior from the measurement data of the process stored in the data storage means 3 so as to match the actual behavior. Various methods are conceivable. For example, the following generalized linear model can be used as one of mathematical or physical models for simulating plant behavior.

It shall be expressed as In addition,

N is the number of measurement data sets.

とおくと、一般化線形モデルは

となる。この式(2)の誤差ｅが最小となるようなθを求める方法の１つとして、最小2乗法が知られている。実際のプロセス方程式をこの形へ変形する方法については、次に示す説明中で示すこととする。 Now

A generalized linear model is

It becomes. As one of the methods for obtaining θ that minimizes the error e in the equation (2), the least square method is known. A method for transforming an actual process equation into this form will be described in the following description.

  The plant simulation means 7 used in the present invention simulates the plant behavior using the result obtained by adjusting the parameter θ of a mathematical or physical model such as a generalized linear model by the parameter adjustment means 9. The result simulated by the plant simulation means 7 is stored in the data storage means 3 (# 2).

For example, in the water supply process of FIG. 1, water is supplied from the pond 21a to the pond 21b and the pond 21c by the water supply pump 22. Discharge amount and head of water pump 22 are characteristic equations,
H = k ₁ Q ² + k ₂ Q + k ₃ (3)
(However, water head H, discharge amount _Q, k 1 to k ₃ are to be adjusted parameter)
It shall be expressed as At this time, the pressure loss at the solenoid valve 23 is
h _I −h _O = f _v (u) · Q ² (4)
It is represented by Here, h _I and h _O are the head values before and after the valve, Q is the discharge amount, and u is the valve opening.

Note that f _v (u) in equation (4) is expressed by equation (5).
f _v (u) = 1 / ( ² gA ² ) * (1 / u−1) ⁿ (5)
At this time, the parameter to be adjusted is n.

Similarly, the pressure loss in the pipeline is from the Hazen Williams equation:
_{h I -h O = f l ·} Q 1.85 (6)
It is represented by However, h _I and h _O are the head value before and after the pipeline and the flow rate Q.

Incidentally, f _l in the formula (6) is represented by the formula (7).
f _l = 10.666 x C ^-1.85 x D ^-4.87 x L (7)
At this time, D is the diameter of the pipeline, L is the pipeline length, and the parameter to be adjusted is C. From these, the pressure balance equation for the process is created as follows.

H ₂ the water level of the distributing reservoir 21b, the water level of the distributing reservoir 21a and h _1, when the flow rate of the conduit leading to 21b from distributing reservoir 21a to is Q, the pressure loss can be expressed by general formula (6) and (7) . Further, the relationship between the pump discharge pressure and the flow rate is expressed by equation (3), and the pressure loss of the solenoid valve 23 is expressed by equations (4) and (5). Then, from these,
h ₂ -h ₁ = k ₁ Q ² + k ₂ Q + k ₃ + (10.666 × C ^-1.85 × D ^-4.87 × L) × Q ^1.85
-1 / (2 × g × A ² ) × (1 / u−1) ⁿ × Q ² (8)
It becomes. However, here, the characteristics of the pressure loss of the pipe line are summarized in one for easy explanation, but in reality, the material and the diameter of the pipe are different in the middle of the pipe, or the primary of the solenoid valve 23 Therefore, each time the pipes are different, it is necessary to set and adjust the parameters corresponding to equations (6) and (7).

より、Cを算出する。 In the equation (8), the parameters k ₁ , k ₂ , k _3, C ^−1.85 , and n are estimated by applying the least square method with respect to h ₂ −h _1, Q, Q ² , Q ^1.85 , and u. For C- ^1.85 , let x be the estimated value.

From the above, C is calculated.

  The parameter adjustment method explained here can be applied to various combinations of pipes, valves, pumps, and ponds by creating balanced equations for pressure and flow corresponding to equations (3) to (8). It is possible.

  In addition, the adjustment method described here is an example, and the method is not only the least square method, but also the sequential least square method (a method known as a special case of the Kalman filter), auxiliary variable method, neural network, minimization viewpoint Then, mathematical techniques such as the steepest descent method, genetic algorithms, simulated annealing, tab search, and ant algorithm methods such as Ant algorithm may be used. The data adjusted by the parameter adjusting means 9 is stored in the data storage means 3 (# 1).

Data filtering means 6
FIG. 2 shows the structural relationship between the filtering means selection means 5 and the data filtering means 6. Data filtering can be performed by selecting several filtering means as will be described later by the operation of the filtering means selection means 5 in response to a command from the GUI 4. Below, six examples are shown per filtering means.

により計算し、不適切データの候補を選定する。次に、後述する任意の閾値Ｓを設定し、この不適切データの候補のうち閾値Ｓを越えたものを不適切データと判定し除去する。 Example 1 of the data filtering means 6 used in the present invention removes inappropriate data included in measurement data used to adjust a parameter of a mathematical or physical model that simulates plant behavior by difference calculation and a threshold value. The difference between the current measured value x (t) and the measured value x (ta) before any time

Calculate candidates and select candidates for inappropriate data. Next, an arbitrary threshold value S to be described later is set, and among the inappropriate data candidates, those exceeding the threshold value S are determined as inappropriate data and removed.

  Example 2 of the data filtering means 6 used in the present invention removes inappropriate data included in measurement data used for adjusting parameters of a mathematical or physical model that simulates plant behavior by statistical processing and threshold values. As an example of statistical processing, filtering using variance is shown below.

により計算し、実測データの移動平均から+nσ、-nσ（n：正の整数）の値を閾値として、その閾値を超える実測データを不適切データとして除去する。 Distribution of measured data

And the values of + nσ and −nσ (n: positive integer) are taken as threshold values from the moving average of the measured data, and the measured data exceeding the threshold value are removed as inappropriate data.

  Example 3 of the data filtering means 6 used in the present invention is inappropriate data included in measurement data used for adjusting parameters of a mathematical or physical model that simulates plant behavior by difference calculation, statistical processing, and threshold values. Remove. As an example of statistical processing, filtering using distributed calculation is shown below.

により計算し、不適切データの候補を選定する。次に、実測データの分散を、

により計算する。次に実測データの移動平均を計算し、その値から+nσ、-nσ（n：正の整数）を閾値とし、不適切データの候補のうち、この閾値を超えたものを不適切データとして除去する。 The difference between the current measured value x (t) and the measured value x (ta) before any time

Calculate candidates and select candidates for inappropriate data. Next, the variance of the measured data is

Calculate with Next, calculate the moving average of the measured data, and use + nσ and -nσ (n: positive integer) as threshold values from those values, and remove inappropriate data candidates that exceed this threshold as inappropriate data. To do.

  Example 4 of the data filtering means 6 used in the present invention removes inappropriate data included in measurement data used to adjust parameters of a mathematical or physical model that simulates plant behavior by a frequency filter. Various frequency filters can be used for removing inappropriate data, but a frequency filter method such as a low-pass filter may be used alone, or a plurality of methods may be used in combination.

  Example 5 of the data filtering means 6 used in the present invention is included in measurement data used for adjusting parameters of a mathematical or physical model that simulates plant behavior by a self-organizing map subjected to self-organizing learning. Remove inappropriate data. Here, a learning method of the self-organizing map and a method of removing inappropriate data will be described.

出力ベクトルを

とし、重みベクトルをWｊとする。 FIG. 3 shows the configuration of the self-organizing map. Here, the input vector is

Output vector

And the weight vector is Wj.

である。例えば、図１に示した上水道プラントの送水プロセスにおいては、入力Ｕには圧力や流量が入力され、出力Ｙには、水位などが入力される。この入出力ベクトルより学習ベクトル

を生成し、これを基に重みベクトル

を求める。次に、

より入出力ベクトルと最も類似した自己組織化マップ上のユニットを検索し、勝者ユニットとする。その勝者ユニットおよびその近傍のユニットを

により更新する。なお、この式におけるε_selfは、学習係数を表す。この勝者ユニットの決定と更新を繰り返し行うことにより、対象の特徴を写像したマップが形成される。 The input / output vector to the self-organizing map is

It is. For example, in the water supply process of the water supply plant shown in FIG. 1, pressure and flow rate are input to the input U, and water level and the like are input to the output Y. Learning vector from this input / output vector

And the weight vector based on this

Ask for. next,

The unit on the self-organizing map that is most similar to the input / output vector is searched for as a winner unit. The winner unit and nearby units

Update with Note that ε _self in this equation represents a learning coefficient. By repeatedly determining and updating the winner unit, a map that maps the features of the object is formed.

  As shown in FIG. 4, the self-organizing map has a clustering function in which data having similar features are gathered nearby as described above, and data having different features are gathered in a place away from the set. First, among the process data stored in the data storage means, data that does not include missing data or measurement errors is input to the self-organizing map, and self-organizing learning is performed, so that a feature map of appropriate process data can be obtained. Form.

  Next, the actual measurement data used to adjust the process model parameters is input to the self-organizing map that has been learned. At that time, if the unit away from the pre-configured feature map becomes the winner unit, it means that the feature is different from the appropriate process data, so that data is removed as inappropriate data. Do.

  Example 6 of the data filtering means 6 used in the present invention is that the user can select from a plurality of filtering means when the properties of inappropriate data included in the measured data differ depending on conditions such as the measurement location and the accuracy of the measuring instrument. It is. Also, this device can be selected by this means. When this apparatus selects, it selects based on the past performance of the filtering accuracy with respect to conditions, such as a measurement location.

5 and 6 show an example of the data filtering means 6, and each filtering means shown by the switching method in FIG. 2 is shown separately. 5 (a) to (d) and FIGS. 6 (a) to (d), the filtering means 6 shown in FIG. 2 is replaced with a threshold value (FIG. 5 (a)) and a difference (FIG. 5 (b)). , Difference + threshold (FIG. 5 (c)), statistical processing (FIG. 5 (d)), statistical processing + threshold (FIG. 6 (a)), difference + statistical processing + threshold (FIG. 6 (b)), frequency filter This is shown as a self-organizing map (FIG. 6D).
Here, −S in FIG. 5 (a) is a threshold, Δx in FIGS. 5 (b), 5 (c) and 6 (b) is a difference, and ± nσ in FIGS. 5 (d) and 6 (b) is a statistics. In the above difference, σ ² in FIGS. 6A and 6B indicates statistical processing (dispersion).

Update determination means 8
FIG. 7 shows the configuration of the update determination means 8 used in the present invention. The update determination means 8 can select either the manual mode or the automatic (square error + threshold) mode in response to a command from the simulation means 7, and obtains input from the filtering means 6 regardless of which mode is selected. The output is given to the parameter adjusting means 9.

の２乗誤差を下記の式に従い計算する。

Then, whether or not the adjustment of the model parameter is required is determined by comparing the measured data of the simulated plant and the square error of the simulation result simulated by the parameter before adjustment. First, the actual measurement value x _i and the simulation value

Is calculated according to the following equation.

がその閾値を越えた時点で、シミュレーションのモデルパラメータが実際のプラントと乖離したと判断し、パラメータ調整を実行する。 Set an arbitrary threshold value ε and sum of squared errors up to the present

When the value exceeds the threshold, it is determined that the model parameter of the simulation has deviated from the actual plant, and parameter adjustment is executed.

Embodiment 2

  FIG. 8 shows a second embodiment of the present invention. As shown in FIG. 8, the configuration of the present invention is a configuration in which all of the configuration of the present invention also includes remote data input / output means for transmitting plant measurement values and adjusted model parameters through a communication line such as a dedicated line, a public line, and a radio. is there. The ASP (Application Service Provider) side stores process values and performs adjustment calculation of model parameters based on them, and the result is distributed via a communication line.

  By adopting such a configuration, the plant side as a user can receive only information and use it as support information, instead of purchasing the model parameter fitting device itself.

Other embodiments

  The present invention can be applied not only to a waterworks plant but also to a sewerage plant, an industrial plant such as a steel plant, a district heating and cooling plant, a heat source plant for each building, and an operation monitoring control system related to plant operations such as a power plant.

  In addition, if a measurement value is given even offline, it is possible to automatically adjust the model parameter, so that it can be used as a model identification tool for control system design.

The block diagram which shows one Embodiment of this invention. Explanatory drawing which shows the structure of the filtering means in the whole structure of FIG. Explanatory drawing which shows the structure of the self-organization map which is one method of the data filtering in this invention. Explanatory drawing which shows the update etc. of the unit of the self-organization map shown in FIG. FIGS. 5A to 5D are block diagrams individually showing the filtering means shown in FIG. 6 (a) to 6 (d) are block diagrams showing the filtering means shown in FIG. 2 individually. Explanatory drawing which shows the structure of the update determination means in the whole structure of FIG. The block diagram which shows another embodiment of this invention.

Explanation of symbols

1 plant modeling device, 2 input means, 3 data storage means, 4 GUI,
5 filtering method selection means, 6 data filtering means,
7 plant simulation means, 8 update determination means, 9 parameter adjustment means,
10 output means, 11 remote data input means, 21 ponds (reservoir pond, wash pond, etc.),
22 water pump, 23 solenoid valve, 24 flow meter, 25 water level meter.

Claims (11)

Data storage means for storing data used to adjust parameters of a mathematical or physical model simulating plant behavior;
A data filtering means for removing inappropriate data contained in the actual measurement value from the plant from the mathematical or physical model for simulating the plant behavior and the data used for adjusting the parameters given from the data storage means; ,
Parameter adjusting means for adjusting parameters of a mathematical or physical model that simulates plant behavior from data given from the data storage means via the data filtering means;
A model parameter fitting apparatus comprising: input / output means for performing input / output processing of a parameter after being adjusted by the parameter adjusting means. The model parameter fitting device according to claim 1, wherein
Of the data used to adjust the parameters of the mathematical or physical model that simulates the plant behavior, differential data processing means for performing differential data processing by the difference between the data before an arbitrary time and the data at the current time;
A model parameter fitting device comprising: threshold data filtering means for removing inappropriate data from the data after the differential data processing by a threshold. The model parameter fitting device according to claim 1, wherein
Among the data used to adjust the parameters of a mathematical or physical model that simulates plant behavior, statistical processing means for filtering by statistical processing of the data,
A model parameter fitting device, comprising: threshold data filtering means for removing inappropriate data from the data after statistical processing using a threshold. The model parameter fitting device according to claim 1, wherein
Of the data used to adjust the parameters of the mathematical or physical model that simulates the plant behavior, differential data processing means for performing differential data processing by the difference between the data before an arbitrary time and the data at the current time;
Statistical data processing means for determining a threshold value by statistical processing;
A model parameter fitting apparatus comprising: threshold data filtering means for removing inappropriate data using threshold values determined by differential data processing and statistical data processing. The model parameter fitting device according to claim 1, wherein
A model parameter fitting device comprising a frequency data filtering means for removing inappropriate data by a frequency filter among data used for adjusting parameters of a mathematical or physical model simulating plant behavior . The model parameter fitting device according to claim 1, wherein
It is characterized by a self-organizing map data filtering means that removes inappropriate data from self-organizing maps among data used to adjust parameters of mathematical or physical models that simulate plant behavior. Model parameter fitting device. The model parameter fitting device according to any one of claims 1 to 6,
Filtering means selection means for selecting a means according to the characteristics of data to be processed, having a plurality of means for filtering data used to adjust parameters of a mathematical or physical model that simulates plant behavior A model parameter fitting device characterized by being provided. The model parameter fitting device according to any one of claims 1 to 7,
It is characterized by comprising an update determination means for judging by a square error and a threshold value between measured data of a plant that simulates whether or not model parameter adjustment is required and a simulation result simulated by a parameter before parameter adjustment. Model parameter fitting device. The model parameter fitting device according to any one of claims 1 to 7,
A model parameter fitting device, characterized by comprising an update determination means for updating a model parameter at an arbitrary timing. The model parameter fitting device according to any one of claims 1 to 9,
A model parameter fitting device comprising a remote data input / output means for transmitting plant measurement values and adjusted model parameters through a communication line such as a dedicated line, public line, or wireless line. The data storage means stores data used to adjust the parameters of a mathematical or physical model that simulates plant behavior,
Of the data used to adjust the mathematical or physical model that simulates plant behavior and its parameters, remove inappropriate data by means of data filtering,
Adjusting parameters of a mathematical or physical model that simulates plant behavior by parameter adjusting means from data stored in the data storage means;
Model parameter fitting method that performs input / output processing of adjusted parameters by input / output means.

Images