Research On Big Data Analysis And Prediction Method Of Regional Water Consumption Based On Cloud Platform

The urban water demand prediction problem based on the distributed big data system is one of the cores of urban water supply system scheduling and management.The article designs an improved algorithm based on BP neural network on the cloud platform and applies it to the prediction of urban area water demand.This method not only can scientifically and accurately predict the regional water demand,but also can effectively solve the problems existing in the traditional stand-alone machine,and provide important support for relevant departments in decision-making.According to the background that Yunnan Water Investment Corporation's “Internet +” thinking,the cloud computing platform will be used to aggregate the massive water data obtained by smart water meters through the mobile broadband network to the cloud in real time,and then combined with relevant algorithms to analyze and predict the water demand in urban areas,etc.The distributed data storage system based on the cloud platform implements the prediction algorithm of this article,that is,a big data platform is built in the cloud environment,and the distributed file system HDFS of the platform is used to store the massive water data obtained by Flume from the smart terminal in real time,and then Use Spark to parallelize the proposed improved algorithm,and then predict the water demand in urban areas.According to research,there is a non-linear relationship between the influencing factors of water consumption in urban areas,and BP neural network has a strong nonlinear mapping ability,therefore,the article focuses on the in-depth study of the urban area water demand prediction algorithm based on BP neural network.At the same time in the research process,it was found that the BP neural network has some shortcomings such as poor global search ability and sensitivity to initial parameters,and then used particle swarm optimization to make up.Upon further in-depth research,it was found that the combined model has the disadvantages of being easily trapped in local optimum and low in search accuracy,therefore,the idea of genetic algorithm was introduced to improve the particle swarm optimization algorithm,and the improved particle swarm optimization BP neural network combined algorithm model is proposed to predict the urban area water demand.Finally,the experiment shows that this scheme can not only improve the accuracy of the urban area water demand prediction model,but also effectively solve the limitations of the traditional single-machine model,and provide an accurate basis for the urban area water demand scheduling and scientific prediction.