Research And Visual Design Of User Energy Consumption Prediction

In recent years,with the development of intelligent perception and wireless communication technology,and the continuous expansion of the scale of smart grid construction,a large amount of electricity-related data has been accumulated.Using data mining,pattern recognition and other methods to mine and analyze the law of energy consumption changes from these massive historical data,and establish an accurate and reliable energy consumption prediction method system,which will provide a scientific decision-making basis for power resource dispatch.However,affected by external factors such as holidays,seasonal effects,and climate,energy consumption data presents complex non-linear characteristics.In addition,the characteristics of large data scale,high dimensionality,and strong timeliness greatly increase the difficulty of data analysis and bring great challenges to energy consumption prediction based on data-driven methods.Based on the above content,this paper takes the power consumption data as the research object,and based on the cloud computing platform,uses the clustering analysis,deep learning and time series analysis methods to carry out the systematic research of short-term and medium and long-term energy consumption prediction methods.The main research contents and innovations are as follows:1.Aiming at the characteristics of short-term electricity energy consumption data,first propose a clustering algorithm based on improved iterative self-organization(ISODATA),which clusters and analyzes the user’s electricity consumption behavior,and then uses similar sample data as input to establish a support vector Machine regression(SVR)energy consumption prediction model,and use the cuckoo algorithm(CS)to optimize the kernel function parameters,and then combine with Map Reduce in the cloud computing platform for parallel computing.Numerical simulation results show that the use of various sample sets obtained after cluster analysis for modeling can obtain better energy consumption prediction results and reduce the calculation time.2.According to the characteristics of medium and long-term power consumption data,a power consumption prediction model based on improved long-term and short-term memory neural network(LSTM)and prophet is established.Firstly,convolution neural network(CNN)is used to extract features from historical energy consumption data,and the extracted features are used as input to establish energy consumption prediction model based on LSTM.Due to the influence of weather,holidays and other factors of medium and long-term energy consumption data,a hierarchical model of historical energy consumption data is established by using prophet algorithm.Finally,the two groups of predicted values are weighted to get the final predicted value.Numerical simulation shows that the first mock exam model has a higher general accuracy than the single model.3.Aiming at the problem that the traditional stand-alone data processing platform is not enough to deal with massive data,build a cloud computing platform,and use the existing energy consumption prediction method to parallel calculate the data,so as to reduce the running time of the algorithm.Then,combined with the characteristics of massive data and multi-source,the workflow of power big data visualization is arranged,and the results of the two energy consumption prediction methods mentioned in this paper are visualized through Datav.The experimental results show that the power big data visualization based on cloud computing platform can provide decision support for improving the efficiency and reliability of energy consumption.