JP6662109B2 - Product state prediction device, product state control device, product state prediction method and program - Google Patents

Description

  The present invention relates to a product state prediction device, a product state control device, and a product state prediction method suitable for use in calculating a predicted value of a cooling stop temperature of a steel plate by using a temperature prediction model in a steel plate cooling step, for example. And programs.

In the cooling step downstream of the hot rolling step of the steel sheet, the steel sheet is cooled to a predetermined temperature to obtain the mechanical properties required for the product.
Since the cooling control of the steel sheet is directly connected to the mechanical properties of the product and affects the yield, it is required to improve the accuracy. In the cooling control, a temperature change of the steel sheet due to water cooling is estimated by a heat transfer model calculation, and a cooling water amount, a transport speed of the steel sheet, and the like are determined so as to obtain a desired cooling stop temperature. Although the heat transfer model is created based on the knowledge of thermo-fluid dynamics and heat transfer engineering, there are many factors that are difficult to model, such as the surface properties of steel sheets and changes over time in equipment. It is difficult to accurately match the target value.

Therefore, a method is employed in which a difference (modeling error) between the model calculated value of the cooling stop temperature and the actual value is predicted by a learning model, and the model calculated value of the cooling stop temperature is compensated.
Patent Literature 1 discloses that, for a steel plate to be subjected to a cooling process, a temperature prediction model of the steel plate for predicting a temperature change behavior of the steel plate in the cooling process is used to calculate a cooling stop temperature of the steel plate in the cooling process. A predictive value calculating step of calculating a predictive value, an extracting step of extracting past actual data of a thick steel sheet having similar manufacturing conditions to the thick steel sheet from a database storing past actual data for each slab, and extracting in the extracting step. An estimation step of estimating an error of a predicted value of a cooling stop temperature of the thick steel plate based on the past actual data, and estimating an error of a predicted value of a cooling stop temperature of the thick steel plate. A correction value calculation for calculating a correction value of the cooling stop temperature from a predicted value of the temperature and an error of the predicted value of the cooling stop temperature of the thick steel plate estimated in the estimation step. Step, it is disclosed that has a.

Japanese Patent No. 5682484

International Journal of Pharmaceutics, Vol.421, No.2, pp.269-274,2011 Applied Intelligence, Vol.17, Issue 1, pp.49-60,2002 Iron and steel, 15, 2509-2514, 1981

As described above, when the performance data is extracted from the database, the similarity of the manufacturing conditions is determined based on, for example, the weighted Euclidean distance function of Expression (1). Here, x _i : = (x _{i, 1} ,..., X _{i, m} ) ^T is a manufacturing condition, q _j is a manufacturing condition coefficient, and i is a suffix (i = 0: prediction) indicating a steel plate (manufacturing number). Target material). Euclidean distance d _i actual data in ascending order of a predetermined number extracted.

A local regression model is generated using the performance data extracted in this way as neighborhood teacher data. For example, a local weighted regression model represented by Expressions (2) and (3) is generated. Here, β is a regression parameter, y is a modeled error actual value, y ＾ (＾ is attached to y) is a modeled error predicted value, and n is the number of neighboring teacher data. W _i is a weight coefficient.

  As specific means, local weighted linear multiple regression (Equations (4) and (5)) and local weighted PLS (Partial Least Squares) regression (Equations (6) to (8)) have been proposed. .

The local weighted linear multiple regression is a method in which a modeled error prediction value y ＾ is represented by a linear combination of manufacturing conditions x (Equation (4)), and a coefficient a _reg that minimizes the weighted error sum of squares is obtained. (Equation (5)). This method has a problem of multicollinearity in which regression coefficients become unstable when there is a correlation between manufacturing conditions.
Therefore, in the local weighted PLS regression, a latent variable z that maximizes the covariance between variables is generated so that multicollinearity does not occur (Equation (8)). C _tr is a matrix composed of eigenvectors of the matrix X ^T Wy ^T WX. _{X: = (x 1, ···} , x n) T, W: = diag (w 1, ···, w n), y: = (y 1, ···, y n) is ^T. Thereafter, the problem of multicollinearity is avoided by performing linear multiple regression on the latent variable z (Equations (6) and (7)).

In the above-described procedure of local weighted regression, the setting of the weighting coefficient w _i generally uses the form of Expression (9). α is a parameter and is a positive real number.

In addition to Patent Literature 1, similar forms are used in Non-Patent Literatures 1 and 2 that use local weighted regression.
This is because the Euclidean distance d _i is a real number taking 0 to ∞, the weight coefficient w _i is 1-0, and the request point (here the prediction target material) focused closer to, and not too large This is thought to be due to the high affinity with the local weighted regression method.

However, when the material to be predicted is a special steel plate or a new steel type, there may be no actual data having similar manufacturing conditions in the vicinity. FIG. 5 is a diagram illustrating the relationship between the distribution of the neighborhood teacher data and the weighting factor. If there is no actual data having similar manufacturing conditions in the vicinity, as shown in FIG. 5, w _i ≒ 0 for all i, the regression calculation diverges, and prediction becomes impossible.
On the other hand, as a remedy the regression calculation to prevent divergence, by decreasing the parameter α of the formula (9), as shown in FIG. 6, it is conceivable to gentle gradients w _i is 1 → 0 . However, conversely, in the case of a prediction target material having a lot of actual data in the vicinity, w _i ≒ 1 for all i, the significance of the locally weighted regression is lost, and there is a concern that the prediction accuracy may be reduced.

In the steel plate cooling process, since the steel plates under various conditions are cooled, it is difficult to know in advance what distribution the result data in the vicinity has. Therefore, it is impossible to manage the weight coefficient with the uniform parameter α.
In Non-Patent Document _{1, α (σ d) =} α'/ σ but _d is a function of the variation sigma _d of d _i as much sigma _d or when large or small compared to the average mu _d of d _i If There is still concern that the above problems may occur.

  The present invention has been made in view of the above points, the regression calculation does not diverge, high accuracy while maintaining the significance of local weighted regression, a model of the state of the product such as the cooling stop temperature of the steel sheet It is an object of the present invention to be able to predict a conversion error.

The gist of the present invention for solving the above problems is as follows.
[1] A product state prediction device that calculates a predicted value of a product state using a state prediction model in a product manufacturing process,
When calculating a predicted value of the state of the product to be predicted, extracting means for extracting neighborhood teacher data based on a distance function from a database storing actual data,
Using the neighborhood teacher data extracted by the extraction means, a local weighted regression model for calculating an error of a predicted value of the state of the product to be predicted calculated by the state prediction model is generated, and the local weighted regression model Calculating means for calculating the error;
Compensation means for compensating for the predicted value of the state of the product to be predicted calculated by the state prediction model, based on the error of the predicted value of the state of the product to be predicted determined by the calculation means,
Said computing means, the weight coefficient w _i of the local weighted regression model (i subscripts indicating the product), the distance d _i by the distance function near the teacher data extracted by the extraction means, the distance d _i by using the average value mu _d and standard deviation sigma _d, the state prediction device products and setting the equation (101).

[2] The database accumulates, as performance data, a difference between a product status performance value and a product status prediction value calculated using the status prediction model in association with manufacturing conditions,
The product state prediction according to [1], wherein the extraction unit extracts actual data having manufacturing conditions similar to those of the product to be predicted as neighborhood teacher data based on the distance function. apparatus.
[3] The product according to [1] or [2], wherein the product is a steel plate, the manufacturing process is a cooling process, the state of the product is a cooling stop temperature of the steel plate, and the state prediction model is a temperature prediction model. Product state prediction device.
[4] A product state prediction device according to [1] or [2],
Control for controlling the operation amount of the manufacturing equipment used in the manufacturing process such that the predicted value of the state of the prediction target product compensated by the compensation unit matches a target value predetermined for each of the prediction target products. And a means for controlling the state of the product.
[5] The product is a steel plate, the manufacturing process is a cooling process, the state of the product is a cooling stop temperature of the steel plate, the state prediction model is a temperature prediction model, and the operation amount is at least one of a cooling water amount and a conveyance speed of the steel plate. The product state control device according to [4], wherein:
[6] A product state prediction method for calculating a predicted value of a product state using a state prediction model in a product manufacturing process,
When calculating a predicted value of the state of the product to be predicted, extracting neighborhood teacher data based on a distance function from a database in which actual data is accumulated;
Using the extracted neighborhood teacher data, a local weighted regression model for calculating an error of a predicted value of the state of the product to be predicted calculated by the state prediction model is generated, and the error is calculated by the local weighted regression model. Steps to
Compensating for the predicted value of the state of the product to be predicted calculated by the state prediction model, based on the error of the predicted value of the state of the product to be predicted calculated,
When calculating the error of the predicted value of the state of the product to be predicted calculated by the state prediction model, the weighting coefficient w _i (i is a subscript indicating the product) of the locally weighted regression model is calculated by using the extracted neighborhood teacher data. wherein the distance d _i by the distance function, the distance mean value of d _i with the mu _d and standard deviation sigma _d, the state estimation method of the product and setting by the formula (101) of.
[7] A program for calculating a predicted value of a product state using a state prediction model in a product manufacturing process,
When calculating a predicted value of the state of the product to be predicted, extracting means for extracting neighborhood teacher data based on a distance function from a database storing actual data,
Using the neighboring teacher data extracted by the extracting means, a local weighted regression model for calculating an error of a predicted value of the state of the product to be predicted calculated by the state prediction model is generated, and the local weighted regression model Calculating means for calculating the error;
Based on the error of the predicted value of the state of the product to be predicted determined by the calculating means, the computer functions as compensation means for compensating the predicted value of the state of the product to be predicted calculated by the state prediction model,
Said computing means, the weight coefficient w _i of the local weighted regression model (i subscripts indicating the product), the distance d _i by the distance function near the teacher data extracted by the extraction means, the distance d _i by using the average value mu _d and standard deviation sigma _d, programs and setting by the formula (101).

  According to the present invention, since the weight coefficient is set so as to decrease in accordance with the distribution of the neighborhood teacher data, the regression calculation does not diverge, and the cooling of the steel sheet is performed with high accuracy while maintaining the significance of the locally weighted regression. It is possible to predict a modeling error of a product state such as a stop temperature.

It is a figure showing the example of composition of the cooling control system of an embodiment. FIG. 4 is a block diagram illustrating a functional configuration of a local regression model generation unit. It is a figure showing distribution of neighborhood teacher data and a relation with a weight coefficient in an embodiment. FIG. 3 is a diagram showing a functional configuration for a numerical experiment. It is a figure which shows the relationship between the distribution of conventional neighborhood teacher data, and a weight coefficient. It is a figure which shows the relationship between the distribution of conventional neighborhood teacher data, and a weight coefficient.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings. In this embodiment, an example in which the present invention is applied to prediction and control of a cooling stop temperature of a steel sheet in a cooling process will be described.
FIG. 1 is a diagram illustrating a configuration example of a cooling control system including a cooling facility.
In the cooling equipment, the steel plate 1 is transported by the transport table 3, and water is injected from the cooling nozzle 2 to cool the steel plate 1.
The cooling nozzle 2 controls the amount of cooling water based on an instruction value output from the cooling process computer 5. As a control method of the cooling water amount, a general cooling water amount control method such as flow rate valve opening adjustment, nozzle ON / OFF control and the like can be applied.
Further, the transfer table 3 controls the transfer speed of the steel sheet based on the instruction value output from the cooling process computer 5. As a control method of the transfer speed of the steel sheet, a general transfer speed control method such as control by a motor driver can be applied.
At the outlet side of the cooling equipment, the temperature at which the steel plate is stopped to cool is measured by the thermometer 4 and input to the cooling process computer 5. As the temperature measuring method, a general temperature measuring method such as a radiation thermometer or a thermography can be applied.

  The cooling process computer 5 includes a cooling stop temperature calculator 7, a subtractor 8, a database 9, a local regression model generator 10, and a cooling setting calculator 11. The cooling process computer 5 receives the production conditions of the material to be predicted from the upper rolling process computer 6. The manufacturing conditions include, for example, a steel sheet size, a steel type, a target value of a cooling stop temperature, and the like, but may include elements that are general as information affecting the cooling step, such as an outside air temperature and a rolling roll roughness. In the present embodiment, the cooling process computer 5 functions as the steel plate cooling stop temperature prediction device and the steel plate cooling control device according to the present invention.

The cooling stop temperature calculation unit 7 calculates a cooling stop temperature of the steel sheet by a temperature prediction model including a heat transfer model based on the manufacturing conditions of the steel sheet, the actual value of the water volume in the cooling facility, and the actual value of the transport speed of the steel sheet (“ "Calculated cooling stop temperature").
The subtracter 8 calculates the difference between the actual cooling stop temperature measured by the thermometer 4 and the calculated cooling stop temperature output from the cooling stop temperature calculator 7 (referred to as a “modeling error actual value”).

The database 9 accumulates the modeled error actual value output from the subtractor 8 in association with the manufacturing condition as actual data. Further, the database 9 has a function of, when the production conditions of the material to be predicted are input, extracting actual data having production conditions similar to the production conditions as neighborhood teacher data, and outputting the data to the local regression model generation unit 10. . In the present embodiment, the database 9 functions as an extraction unit according to the present invention.
When the performance data is extracted from the database, the similarity of the manufacturing conditions is determined based on, for example, the weighted Euclidean distance function of Expression (1). In the present embodiment, the Euclidean distance is taken as an example, but a known variable such as a Mahalanobis distance can be applied to the definition of a multivariable distance.

  The local regression model generation unit 10 obtains an error (referred to as a modeling error prediction value) of the predicted value of the cooling stop temperature of the prediction target material calculated by the temperature prediction model using the neighborhood teacher data extracted from the database 9. A local weighted regression model is generated, and a modeled error prediction value is calculated using the local weighted regression model. In the present embodiment, the local regression model generation unit 10 functions as a calculation unit in the present invention.

  The cooling setting calculation unit 11 compensates for the predicted value of the cooling stop temperature of the prediction target material calculated by the temperature prediction model, based on the modeling error predicted value output from the local regression model generation unit 10. Then, the cooling setting calculation unit 11 performs at least one of the cooling water amount and the transport speed of the steel sheet such that the compensated predicted value of the cooling stop temperature of the prediction target material matches a target value predetermined for each prediction target material. After correcting any one of the indicated values, the instruction value of the cooling water amount and the conveying speed of the steel sheet is output to the cooling equipment. In the present embodiment, the cooling setting calculation unit 11 functions as a compensation unit and a control unit in the present invention. In the present embodiment, an example in which the amount of cooling water and the transport speed of the steel plate are set as the operation amounts is described. However, other elements that are common as the operation amounts of the cooling equipment, such as the height of the cooling nozzle 2, may be included.

FIG. 2 is a block diagram illustrating a functional configuration of the local regression model generation unit 10.
Reference numeral 101 denotes an input unit for inputting production conditions of a material to be predicted and neighboring teacher data extracted from the database 9.

Reference numeral 102 denotes a weight coefficient setting unit that sets a weight coefficient w _i of a locally weighted regression model by Expression (10). Here, mu _d average of the Euclidean distance d _i in the vicinity of the teacher data, the sigma _d is the standard deviation of the Euclidean distance d _i in the vicinity of the teacher data.

Reference numeral 103 denotes a local weighted regression calculation unit based on the neighborhood teacher data and the weighting coefficient w _i , a local weighted linear multiple regression (equations (4) and (5)), and a locally weighted PLS (Partial Least Squares). ) Generate a locally weighted regression model by regression (equations (6) to (8)) and the like.

  Reference numeral 104 denotes a modeling error calculation unit that inputs the manufacturing conditions of the material to be predicted into the local weighted regression model generated by the local weighted regression calculation unit 103 and calculates a modeled error prediction value of the material to be predicted.

  Reference numeral 105 denotes an output unit, which outputs a modeling error prediction value of the prediction target material calculated by the modeling error calculation unit 104 to the cooling setting calculation unit 11.

Here, Equation (10) includes an error function erf () for the Euclidean distance d _i in the vicinity of the teacher data, the average value mu _d and standard deviation sigma _d of the Euclidean distance d _i and variables, the cumulative distribution function of the normal distribution a function that mirror-reversed with respect to the average value mu _d. Thus, as shown in FIG. 3, regardless of the distribution of the neighboring teacher data, the weighting coefficient w _i becomes a decreasing function adapted to the distribution. As described above, since the weighting coefficient w _i is set to decrease in accordance with the distribution of the neighborhood teacher data, the regression calculation does not diverge, and the cooling stop temperature model is maintained with high accuracy while maintaining the significance of the locally weighted regression. The prediction error can be predicted.

The present invention can be applied not only to the prediction and control of the cooling stop temperature of the steel sheet in the cooling process, but also to a general purpose, and in a manufacturing process of manufacturing a product from raw materials at a certain manufacturing facility, a state prediction model is used. The present invention is widely effective when calculating a predicted value of a product state and controlling a manufacturing facility based on the predicted value.
That is, the steel sheet in the embodiment is an example of a product, the cooling process and equipment are an example of a manufacturing process and equipment, the cooling stop temperature of the steel sheet is an example of a product state, and the temperature prediction model is an example of a state prediction model. When the present invention is applied to a general manufacturing process, the state control system may have, for example, the same configuration as that of FIG. 1. The cooling stop temperature calculator 7 in the embodiment is a state calculator, and the cooling setting calculator is 11 is a setting calculation unit, and the instruction values of the cooling water amount and the transport speed of the steel plate may be replaced with the instruction values of the operation amount of the manufacturing equipment.
The state calculation unit predicts the state of the product from the actual values of the manufacturing conditions and the manipulated variables using a state prediction model, and a modeled error actual value, which is an error between the state calculation value and the actual state value, is a database. 9 is accumulated. Then, when calculating the predicted value of the state of the product to be predicted, the neighborhood teacher data is extracted from the database 9 storing the actual data based on the distance function, and the local model generation unit 10 is used by using the neighborhood teacher data. A local weighted regression model for calculating the modeling error, and calculating the local weighted regression model based on the predicted value of the modeling error of the product to be predicted calculated by giving manufacturing conditions to the local weighted regression model. The predicted value of the predicted state of the product to be predicted is compensated. By compensating in this way, the predicted value of the state can be made closer to the actual value. In addition, the setting calculation unit calculates the operation amount instruction value of the manufacturing equipment and controls the manufacturing equipment so that the predicted value of the compensated state matches the target value determined for each product to be predicted. By doing so, the actual value of the state of the product can be made closer to the target value.

  For example, when the present invention is applied to prediction and control of a sheet width in hot rolling, the cooling equipment in the first embodiment is a steel sheet width processing equipment such as an edger and a sizing press. Further, the manufacturing conditions are factors considered to affect the sheet width, such as the steel sheet size, the steel type, and the temperature of the steel sheet on the apparatus entrance side. In addition, the cooling nozzle 2 and the transfer table 3 are actuators for controlling a plate width such as a hydraulic pressure reduction device for an edger and a mechanical crank for a sizing press. Further, the thermometer 4 is a plate width measuring device such as a line sensor. The cooling process computer 5 is a rolling process computer, and the upper rolling computer 6 is a heating process computer or a business computer. Further, the cooling stop temperature calculating unit 7 is a plate width calculating unit that calculates a device outlet side plate width based on a model described in Non-Patent Document 3, for example. Further, the cooling setting calculation unit 11 is based on a model described in Non-Patent Document 3, for example, and a width reduction setting calculation unit that sets the opening degree and speed of the edger or sizing press so that the apparatus outlet plate width becomes a desired value. Become.

The effect of the method to which the present invention was applied was verified by numerical experiments.
FIG. 4 shows a functional configuration for a numerical experiment. The database 9 and the local regression model generation unit 10 are reproduced by a personal computer to perform a numerical experiment. Manufacturing conditions are input to a personal computer from a factory database storing operation results at the factory, and a modeling error prediction value is output by the database 9 and the local regression model generation unit 10.
Since the factory database also stores the modeled error actual values, the factory database was compared with the modeled error predicted values by the database 9 and the local regression model generator 10 to evaluate the accuracy. Unlike FIG. 1, the modeling error prediction value is not reflected in the actual operation, but it is considered that by accurately predicting the modeling error, the accuracy of the cooling stop temperature control is improved.

Based on the contents described in Patent Literature 1, Non-Patent Literature 1, and Non-Patent Literature 2, the local regression model generation unit 10 uses the locally weighted PLS regression, and the weight coefficient is set to 1 (Comparative method 1 ), The weighting factor determined by equation (9) (Comparative method 2), and the weighting factor determined by equation (10) (invention method), the prediction accuracy of the modeling error was evaluated.
Numerical experiments were performed on the operation results of 10115 lines for 6 months, and the accuracy evaluation was performed using the square error average RMSE, which is an index capable of evaluating both the error average and the variation. Table 1 shows the results of the accuracy evaluations of the comparative methods 1 and 2 and the inventive method.

  According to Table 1, in the comparative method 1, since all the weighting factors are 1, regression cannot be performed while emphasizing the neighborhood teacher data close to the material to be predicted, and the prediction accuracy of the modeling error is not good. In Comparative Method 2, there was a steel sheet whose regression calculation diverged and could not calculate the modeling error prediction value. On the other hand, in the invention method, the modeling error can be predicted with good accuracy without diverging the regression calculation.

The steel plate cooling stop temperature prediction device and the steel plate cooling control device to which the present invention is applied are realized by, for example, a computer device having a CPU, a ROM, a RAM, and the like. FIG. 1 illustrates an example in which the cooling stop temperature calculation unit 7, the local regression model generation unit 10, and the cooling setting calculation unit 11 operate on one process computer (cooling process computer 5). , And input / output via a network. In addition, the present invention is applied to a configuration that is generally used as a form of accelerated cooling control in hot rolling, for example, although the manufacturing conditions are input from the rolling process computer 6 and are input from an upper business computer. Applicable.
The present invention can also be realized by supplying software (program) for realizing the functions of the present invention to a system or apparatus via a network or various storage media, and a computer of the system or apparatus reads and executes the program. It is feasible.

  1: steel plate, 2: cooling nozzle, 3: transport table, 4: thermometer, 5: cooling process computer, 7: cooling stop temperature calculator, 8: subtractor, 9: database, 10: local regression model generator, 11: cooling setting calculation unit, 101: input unit, 102: weight coefficient setting unit, 103: regression calculation unit with local weight, 104: modeling error calculation unit, 105: output unit

Claims (7)

In a product manufacturing process, a product state prediction device that calculates a predicted value of the state of the product using a state prediction model,
When calculating a predicted value of the state of the product to be predicted, extracting means for extracting neighborhood teacher data based on a distance function from a database storing actual data,
Using the neighborhood teacher data extracted by the extraction means, a local weighted regression model for calculating an error of a predicted value of the state of the product to be predicted calculated by the state prediction model is generated, and the local weighted regression model Calculating means for calculating the error;
Compensation means for compensating for the predicted value of the state of the product to be predicted calculated by the state prediction model, based on the error of the predicted value of the state of the product to be predicted determined by the calculation means,
Said computing means, the weight coefficient w _i of the local weighted regression model (i subscripts of an article), the distance d _i by the distance function near the teacher data extracted by the extraction means, the distance d _i Using the average value μ _d and the standard deviation σ _d , equation (101)

An apparatus for predicting the state of a product, wherein the apparatus is set by: The database, as actual data, accumulates the difference between the actual state value of the product and the predicted value of the state of the product calculated using the state prediction model in association with the manufacturing conditions,
2. The product state prediction according to claim 1, wherein the extraction unit extracts actual data having manufacturing conditions similar to those of the product to be predicted as neighborhood teacher data based on the distance function. 3. apparatus.   3. The product state prediction according to claim 1, wherein the product is a steel sheet, the manufacturing process is a cooling step, the product state is a cooling stop temperature of the steel sheet, and the state prediction model is a temperature prediction model. 4. apparatus. A product state prediction device according to claim 1 or 2,
Control for controlling the operation amount of the manufacturing equipment used in the manufacturing process such that the predicted value of the state of the prediction target product compensated by the compensation unit matches a target value predetermined for each of the prediction target products. And a means for controlling the state of the product.   The product is a steel sheet, the manufacturing step is a cooling step, the state of the product is a cooling stop temperature of the steel sheet, the state prediction model is a temperature prediction model, and the operation amount is at least one of a cooling water amount and a conveyance speed of the steel sheet. The product state control device according to claim 4, wherein: In a product manufacturing process, a product state prediction method for calculating a predicted value of the state of the product using a state prediction model,
When calculating a predicted value of the state of the product to be predicted, extracting neighborhood teacher data based on a distance function from a database in which actual data is accumulated;
Using the extracted neighborhood teacher data, a local weighted regression model for calculating an error of a predicted value of the state of the product to be predicted calculated by the state prediction model is generated, and the error is calculated by the local weighted regression model. Steps to
Compensating for the predicted value of the state of the product to be predicted calculated by the state prediction model, based on the error of the predicted value of the state of the product to be predicted calculated,
When calculating the error of the predicted value of the state of the product to be predicted calculated by the state prediction model, the weighting coefficient w _i (i is a subscript indicating the product) of the locally weighted regression model is calculated by using the extracted neighborhood teacher data. the distance and the distance d _i by a function, by using the average value mu _d and standard deviation sigma _d of the distance d _i in the formula (101)

A method for predicting the state of a product, characterized in that: In a product manufacturing process, a program for calculating a predicted value of the state of the product using a state prediction model,
When calculating a predicted value of the state of the product to be predicted, extracting means for extracting neighborhood teacher data based on a distance function from a database storing actual data,
Using the neighborhood teacher data extracted by the extraction means, a local weighted regression model for calculating an error of a predicted value of the state of the product to be predicted calculated by the state prediction model is generated, and the local weighted regression model Calculating means for calculating the error;
Based on the error of the predicted value of the state of the product to be predicted determined by the calculating means, a computer functions as a compensation means for compensating the predicted value of the state of the product to be predicted calculated by the state prediction model,
Said computing means, the weight coefficient w _i of the local weighted regression model (i subscripts indicating the product), the distance d _i by the distance function near the teacher data extracted by the extraction means, the distance d _i Using the average value μ _d and the standard deviation σ _d , equation (101)

A program characterized by being set by:

Images