Research On Multi-energy Collaborative Scheduling Method For Industrial Internet

With the integration and development of the Industrial Internet technology and the energy industry,the traditional energy industry has begun to transform into digital,networked and intelligent.As an efficient energy scheduling method,multi-energy collaborative scheduling can effectively improve the management and scheduling level of industrial energy scheduling systems,which has important research significance.Aiming at the problems of low forecasting accuracy of energy load demand,high energy scheduling cost,large pollutant emission,and serious waste of clean energy under the multi-energy collaborative scheduling scenario oriented to the Industrial Internet,this thesis establishes a short-term energy load forecasting model by collecting historical power load data,which effectively improves the accuracy of short-term energy load forecast.Based on the load forecasting results,a multi-energy collaborative scheduling model for the Industrial Internet is constructed,which reduces the economic cost,pollutant emissions and clean energy waste in the process of multi-energy collaborative scheduling.This thesis mainly studies the short-term energy load forecasting problem and the multi-energy collaborative scheduling problem based on load demand forecasting in the Industrial Internet scenario.The specific research contents are as follows:(1)A short-term energy load forecasting model based on IWOA and Bi LSTM-Attention neural network is proposed.In this thesis,a short-term energy load forecasting model based on IWOA and Bi LSTM-Attention neural network is established.The model introduces an attention mechanism based on the Bi LSTM neural network model,and uses the improved whale optimization algorithm to optimize the hyperparameters of the neural network.Thus,the load forecasting accuracy of the model is improved.The experimental results show that the short-term energy load forecasting model based on IWOA and Bi LSTM-Attention neural network is superior to other basic neural network load forecasting models in forecasting accuracy.(2)A multi-energy collaborative scheduling model for the Industrial Internet is proposed.In this thesis,a multi-energy collaborative scheduling model for the Industrial Internet is constructed.The model considers the cost,pollutant emissions and clean energy use of the wind power,photovoltaic power,thermal power and power storage systems in the power supply process.The collaborative scheduling process is transformed into a multi-objective optimization problem with economic cost and clean energy waste rate as the goal.In order to solve the proposed problem,this thesis improves the traditional multi-objective particle swarm optimization algorithm,improves the search ability of the algorithm by using the adaptive grid mechanism and the memory repository mechanism,and improves the performance of the algorithm by using the mutation strategy and the improved inertia weight.The experimental results show that the multi-energy collaborative scheduling model for the Industrial Internet proposed in this thesis can significantly reduce the economic cost of energy scheduling and the waste rate of clean energy in industrial scenarios,and the improved algorithm has good optimization ability and search performance.