A Neural Network Method for Quantity-quality Prediction in Lead-zinc Sintering Process

Based on some features in the lead-zinc sintering process, such as strong non-linearity and a large time delay, a variable-learning-rate-based back propagation neural network (BPNN) is proposed to predict quantity and quality in the sintering agglomeration. First, the factors influencing quantity and quality are determined by analyzing the correlation of operation parameters. Then, the quantity-quality predictive models of agglomerations are established applying a BPNN based on the variable-learning-rate method. Finally, compared with usual BP training algorithm, this algorithm provides a better convergence rate and the obtained quantity-quality predictive models possess a higher accuracy. Actual results show that the proposed predictive method settles the modeling problem of the quantity and quality in the lead-zinc sintering process.