Prediction of Acid Concentration in Continuous Pickling of Strip Steel Based on Deep Learning Optimized by Artificial Bee Colony

Pickling is an important part of the strip treatment process. Effectively predict the acid concentration in the next time period and make reasonable judgments and treatments, which can provide a certain reference for the control of the acid concentration. In order to accurately predict the acid concentration, an artificial bee colony algorithm (ABC) based on an improved search strategy to optimize a deep learning acid concentration prediction model is proposed in this paper. Based on the joint network model of convolutional neural network (CNN) and long-short-term memory neural network (LSTM), this model introduces the ABC algorithm that improves the search strategy. The ABC algorithm with improved search strategy is used to optimize the key parameters of the CNN-LSTM model, in order to match the acid concentration data features with the model structure and improve the prediction accuracy of the model. The experiment builds an ABC-CNN-LSTM model based on the pickling data of the No. 1 pickling tank of a steel plant, and analyzes the prediction results of the model. The experimental results show that the CNN-LSTM acid concentration prediction model optimized by the ABC algorithm based on the improved search strategy reduces the influence of manual parameter adjustment and significantly improves the prediction accuracy of acid concentration.