Method for predicting power consumption of cement production which implies deep belief network of time series

Abstract

The invention discloses a cement production power consumption prediction method with implicit time series deep belief network, which includes: selecting input variables according to the cement process and performing normalization processing to construct the input layer of the model; determining the initial parameters of the model Complete the initial establishment of the HTS-DBN model, and conduct unsupervised forward training on the model to determine the initial weights and biases; use the BP reverse error correction algorithm to perform supervised reverse fine-tuning on the entire neural network. Use the trained HTS‑DBN model for real-time prediction of electricity consumption in cement production. The HTS-DBN cement energy consumption prediction model established by the present invention solves the time-varying delay problem, can accurately predict the power consumption of cement production, and provides a basis for scientific production scheduling and reasonable energy planning of cement production, thereby providing a basis for cement production. Optimizing production and reducing energy consumption provide the conditions.

Description

technical field [0001] The invention relates to the field of cement production power consumption prediction, in particular to a cement power consumption prediction method based on implicit time series deep belief network. Background technique [0002] Electricity consumption is an important indicator of energy consumption in cement production. Accurate prediction of electricity consumption provides a basis for scientific production scheduling and reasonable energy planning of cement, thus providing conditions for optimizing production and reducing energy consumption in cement manufacturing. Therefore, the accurate prediction of cement electricity consumption is of great significance to the energy saving and consumption reduction of cement production. The cement production process is a complex process with time-varying, hysteresis, uncertainty and nonlinear characteristics, which makes it difficult to establish an accurate cement power consumption prediction model. In respon...

AI Technical Summary

Problems solved by technology

Although the convergence speed of the LSSVM prediction model is fast, it is more suitable for small-scale data samples. It is difficult to find the change law between variables in large data, and it requires complex data cleaning. It does not solve the problem of time-varying delay, resulting in prediction accuracy. Low

Facts show that the regression prediction model and the artificial neural network are difficult to solve the problem of time-varying delay due to their own limitations, resulting in low prediction accuracy

Images