Clustering Neural Network Compression Of Railway Power Dispatching Monitoring Data Based On Genetic Algorithm Optimization

The reliability and safety of railway transport facilities are crucial to the development of railway.In order to better monitor and manage railway transport equipment,railway maintenance,rolling stock,power supply,public works,electrical and other industries have established a series of sound monitoring systems to collect a large number of data about the technical status of transport equipment.However,with the rapid increase of data volume,the storage of dispatch monitoring information becomes more complicated,which will seriously hinder the timeliness of dispatch and bring potential risks to the safety of train operation.The use of column storage technology not only has a higher compression ratio than traditional row storage,but also can save I/O operations,so it is highly competitive in the processing of large-scale data,and its cost effectiveness is very significant.The purpose of this paper is to explore how to compress a large number of railway power dispatching monitoring data effectively,especially by using column storage technology to compress the data quickly and accurately.Aiming at the problems of large recording volume and difficult storage of railway power supply scheduling time series data,a new lossless compression method of gated cyclic neural network based on genetic algorithm to improve data clustering was proposed.Firstly,genetic optimization data clustering was used to decompose the multi-dimensional original power data into data clusters with high similarity.The genetic algorithm was used to optimize the clustering to improve the effect of data clustering classification.Then the neural network is used to train the probability distribution model of data encoding,and the characters with higher frequency are endowed with higher character probability.Finally,the arithmetic encoding method is used to encode and compress the data.In this paper,the engineering data of a specific SCADA system is compressed through the experiment,the Hadoop distributed cluster platform is built,and the cluster performance under multiple groups of different compression algorithms is compared.The results show that the genetically optimized clustering gated neural network compression designed in this paper can reduce about 96% cluster memory space and improve about 27% cluster data storage speed.The cluster data query time is increased by 27%.