EPLA electro-pneumatic conversion valve degradation prediction method based on wavelet neural network

[0023]The present invention will be described in detail below in conjunction with the drawings and specific embodiments of the specification.

[0024]A method for predicting EPLA electro-pneumatic conversion valve degradation based on wavelet neural network, the steps include:

[0025]1) Collect and process the coil voltage data, and use the voltage values ​​at the previous 4 moments to predict the voltage at the current moment;

[0026]2) Initialize the parameters of the neural network and wavelet function;

[0027]3) Using particle swarm optimization algorithm to optimize the initial network weights after initial network training;

[0028]4) Use the optimized weights for network retraining, and use the trained network to predict the test set.

[0029]Such asfigure 1 As shown, the specific steps include:

[0030]Step 1. Data preprocessing

[0031]Voltage Since the time series prediction algorithm uses the voltage data of the first four minutes to predict the voltage change in the next minute, the voltage dat