Strip steel thickness prediction method employing shuffled frog leaping feedback extreme learning machine

Abstract

The invention discloses a strip steel thickness prediction method employing a shuffled frog leaping feedback extreme learning machine. The strip steel thickness prediction method comprises the following steps: 1) analyzing acquired steel plate data signals; 2) performing feature extraction; 3) introducing a Kalman filtering idea into the extreme learning machine, and feeding back a difference value between the actual output and the expected output of the network to an input layer to form a feedback extreme learning machine algorithm, optimizing random parameters of the feedback extreme learning machine algorithm by applying a shuffled frog leaping algorithm, and constructing a shuffled frog leaping feedback extreme learning machine prediction model; and 4) applying the shuffled frog leaping feedback extreme learning machine obtained in the step 3) to prediction of the strip steel outlet thickness, and performing result comparison with a traditional extreme learning machine and a shuffled frog leaping extreme learning machine to verify the effectiveness of the method. Through the steps, the prediction method is small in prediction error, high in prediction precision and good in robustness.

Description

technical field [0001] The invention relates to a method for predicting the outlet thickness of strip steel, which is a method for predicting the thickness of strip steel with a mixed leapfrog feedback extreme learning machine. Background technique [0002] Strip thickness plays an important role in the rolling process, and the accuracy of the exit thickness has become an important indicator to measure the quality of the finished steel plate. However, in the actual rolling process, there are many factors affecting the thickness of the strip exit, and each factor has different effects on the thickness of the strip according to the tension control method. At present, the Automatic Gauge Control (AGC) method is widely used in the prediction of strip thickness control, but the control accuracy of this method depends entirely on the accuracy of the control model, so the thickness prediction accuracy is limited. In recent years, neural network has been widely used in the predicti...

AI Technical Summary

Problems solved by technology

[0004] The random selection of input weights and hidden layer bias values ​​for extreme learning machines will affect the calculation of output weights, and make extreme learning machines require more hidden layer nodes than traditional learning algorithms based on parameter adjustment , causing the technical problem that the generalization performance is reduced due to the ill state of the system. This method uses the hybrid leapfrog algorithm to optimize the feedback extreme learning machine, and applies it to the prediction of the thickness of the strip exit. A hybrid leapfrog feedback extreme learning machine is proposed. Steel thickness prediction method to reduce prediction error and improve prediction accuracy and robustness

Images