SYSTEM AND METHOD FOR CONTROLLING ETCHING

Claims

1. A etching control system characterized in that it includes: a module for quantitative assessment of etching quality, designed to obtain data on etching quality and transmit it to the PLC as feedback; PLC designed to receive data on the etching quality transmitted from the quantitative assessment module and to calculate the electric motor torque based on this data; an electric motor designed to regulate the speed of the steel strip in accordance with the specified torque; A PC designed to adjust the relevant parameters of the PLC by means of a neural network model based on the etching quality data transmitted from the quantitative evaluation module and the specified etching quality, and to send the adjusted parameters to the PLC for the purpose of etching control.

2. The etching control system according to paragraph 1, characterized in that the training samples for the neural network model on the PC are collected from the said PLC, and the said samples include r(k), y(k), K P , K I and K D , where r(k) is the specified etching quality at the k-th iteration, y(k) is the etching quality transferred from the quantification module at the k-th iteration, and K P , K I , K D represent the correspondingly adjustable parameters P, I, D of the PLC PID controller.

3. The etching control system according to paragraph 2, characterized in that the collection of training samples is carried out by said PC through communication between the OPC client and the OPC server and communication between the OPC server and the PLC.

4. The etching control system according to paragraph 2, characterized in that the training of said neural network model is carried out using a BP algorithm, where the performance indicator function has the form: where r(k) represents the specified etching quality at the k-th iteration, and y(k) represents the etching quality passed from the quantification module at the k-th iteration.

5. The etching control system according to paragraphs 1-4, characterized in that the number of neurons in the input, hidden and output layers of the said neural network model is 2, 6 and 3, respectively.

6. The etching control system according to paragraph 5, characterized in that the activation function of the hidden layer has the form: .

7. The etching control system according to paragraph 5, characterized in that the output layer activation function has the form: .

8. The etching control system according to paragraph 1, characterized in that the correction and updating of the weight coefficients of the said neural network model are carried out using the gradient descent method.

9. The etching control system according to paragraph 8, characterized in that the said stage of adjusting and updating the weight coefficients of the neural network model using the gradient descent method, in particular, includes the following steps: S1. Initialize the initial values ​​of the weight coefficients of each layer of the specified neural network model, set the iteration counter of the neural network model k=1; S2. Calculate the error between the input and output values ​​of the input layer; if the specified error is less than the first threshold, go to step S3, otherwise, optimize the input and output values ​​of the input layer so that the error becomes less than the first threshold; S3. Calculate the input and output values ​​of the neurons of each layer of the specified neural network model, where the output values ​​of the output layer of the neural network model are the adjustable parameters of the PID controller of the specified PLC; S4. Calculate the output value of the PID controller based on the obtained adjustable parameters of the PID controller; S5. Train the neural network model and adjust the weighting coefficients to implement adaptive tuning of the adjustable parameters of the PID controller; S6. If the adjustable parameters of the PID controller exceed the second threshold value, the training of the neural network model is completed; otherwise, increase the iteration counter of the neural network model k=k+1 and return to step S2.

10. A method for controlling etching, characterized in that it includes the following steps: obtaining data on etching quality using the etching quality quantitative assessment module and transmitting it to the PLC as feedback; receiving data on the etching quality transmitted from the quantitative assessment module into the PLC and calculating the electric motor torque based on this data; regulation of the speed of the steel strip by an electric motor in accordance with the specified torque; adjusting the corresponding PLC parameters on the PC using the neural network model based on the etching quality data transmitted from the quantitative evaluation module and the specified etching quality, and sending the adjusted parameters to the PLC to implement etching control.