Research On Power Consumption Prediction Model In Cement Production Based On Deep Belief Network

The cement production data has time-varying delay,uncertainty and nonlinearity.The current research methods cannot solve the time-varying delay problem,which makes it difficult to establish accurate cement production power consumption prediction model.Accurate prediction of cement production power consumption can provide a theoretical basis for scientific cement production scheduling and reasonable energy consumption management.Therefore,cement production electricity consumption prediction is of great significance.Aiming at the time-varying delay problem in cement production process,a time-varying delay deep belief network cement production power consumption prediction model is proposed,and the differential evolution algorithm is used to optimize the TVD-DBN model structure to achieve accurate power consumption of cement production.Automatic optimization of forecasts and models.The specific research work is as follows:Firstly,by studying the process of clinker burning process in cement rotary kiln,the candidate influence variables affecting the power consumption of cement production are analyzed.The k-neighborhood mutual information method is used to analyze the degree of correlation between the variables,and the correlation analysis between the cement power input variable and the power consumption is converted into the mutual information value and the mutual information change rate between the research variables.Therefore,the key variables affecting the power consumption in cement production are obtained,which reduces the complexity of the model and lays a foundation for the establishment of cement production electricity consumption prediction model.Secondly,according to the characteristics of time-varying delay,uncertainty and nonlinearity in cement production data,the sliding window method is used to map time series data including time-varying delay to deep belief network,and time-varying delay is established.Deep belief network cement production power consumption prediction model.Through the learning of time-varying delay law by TVD-DBN model,the problem that time-varying delay cannot be determined is solved,and the influence of time-varying delay on power consumption prediction is eliminated,and the prediction accuracy of themodel is improved.Finally,for the problem that the optimal structure of the deep learning model is difficult to select,the forward reconstruction error in the forward training of the time-varying delay depth belief network is the optimization goal,and the time evolution delay deep belief network optimization based on differential evolution is established.The model gets rid of the intervention of artificial experience on the selection of model structure parameters.In the experiment,the actual data in the cement production process was used for verification and analysis.The results show that the proposed method has high precision and strong generalization ability,and it can accurately predict the power consumption of cement production and automatically optimize the model.