Research On Energy Consumption Prediction And Model Optimization Of Cement Calcination Based On Deep Belief Network

The cement calcination process is a complex and continuous process with the characteristics of time-varying delay,nonlinearity and uncertainty.The accurate prediction of electricity consumption and coal consumption in the process of cement calcination can provide sufficient information for the energy saving and consumption reduction of cement production and production management scheduling.Therefore,the prediction of energy consumption in cement calcination has important significance.But the characteristics of the cement calcination process make it difficult to establish an accurate energy consumption prediction model for cement calcination,so a sliding window deep belief network cement calcination energy prediction model(SW-DBN)is proposed.In order to solve the problem of determining the optimal model structure,an improved particle swarm optimization algorithm is proposed to optimize the structural parameters of SW-DBN,which realizes accurate prediction of energy consumption and automatic optimization of the optimal model structure.The specific research work is as follows:Firstly,this paper introduces the new dry cement production process,and analyzes the impact of time-varying delay,nonlinearity and uncertainty.By studying the relationship between the cement production variables and the energy consumption,the main relevant variables are determined.Secondly,in order to solve the influence of cement calcination characteristics on energy consumption prediction,the prediction model of cement calcination energy consumption based on sliding window deep belief network is established.The model uses sliding window technology to map the time-series data containing time-varying delay information to the input layer of the prediction model,then uses the deep belief network to mine the change law between energy consumption and input data.This model avoids the complex problem of time series matching,eliminates the influence of time-varying delay characteristic on energy consumption prediction,and achieves accurate prediction of energy consumption in cement calcination process.Finally,in view of the difficulty in determining the optimal model parameters of theSW-DBN model,an improved particle swarm optimization model is proposed to optimize the model parameters.Based on the basic particle swarm optimization algorithm,this model adds inertial weights and adaptive learning factors,so the speed and position of the particle can be updated according to the fitness value,which reduces the possibility that the particle swarm falls into a local optimum.Then,the training mean square error of the SW-DBN model is used as the optimization goal,and the optimal model parameters are obtained through the optimization process,which not only solves the problem of tedious and inefficient manual selection of model parameters,but also improves the energy consumption prediction accuracy of the SW-DBN model.