| Electricity is an important economic index in the power market.The electricity at a given time can directly reflect the level of economic development in the region.On the other hand,it also reflects the sales and operation status of local power companies.A good electricity forecast will help to do a good job in the distribution of electricity in various industries,and play a very important role in the improvement and transformation of power facilities and the operation management of the company.Firstly,this paper classifies and analyzes the internal and external factors affecting electricity.External factors mainly come from five factors: population change,economic and social development,fixed investment,environment and electricity price policy.The internal is mainly the unified classification of the whole industry and the statistical data source of regional electricity.Industry information is divided into residents’ life,large industry,etc.Compared with the actual situation and forecast time requirements,the influencing factors of monthly power consumption are mainly analyzed from internal factors,and the influencing factors of annual power consumption are mainly analyzed from internal and external factors.Secondly,considering the influencing factors of improving annual and monthly power consumption,this paper applies adaptive and self-learning power consumption prediction to complex nonlinear mapping,and improves the prediction accuracy.The neural network is used to study,and the electricity prediction model is designed by using the electricity prediction method based on LSTM neural network.The annual and monthly electricity consumption in Beichuan county is simulated by MATLAB programming.Finally,taking the annual and monthly electricity prediction of Beichuan County as an example,the prediction results of neural network electricity prediction method and traditional common electricity prediction methods are analyzed and compared.The comparison shows that the new method based on neural network has better prediction accuracy than other traditional methods,and can meet the actual needs.The test results show that the new method of power forecasting based on neural network is feasible. |