Parallel Diagnosis Of Transformer Faults Based On Variable Prediction Model Under Spark Framework

Online monitoring data of power equipment presents the characteristics of big data,and transformer fault diagnosis has problems such as poor classification effect of small samples.Traditional stand-alone tools cannot effectively deal with the above situation,while emerging big data storage and processing technology can effectively solve the problem.This project builds cloud computing platform through Spark,Hadoop and other platforms,selects a large amount of DGA data(a large amount of data has no label,a small amount of data has label)for experimental analysis,studies and optimizes methods of transformer fault diagnosis.This paper proposes parallel variable predictive model based on the Spark computing framework.Firstly,HDFS was used as memory storage system,RowMatrix was stored as a distributed matrix storage structure,and broadcast variable and adjustment of the partition number were used to optimize the degree of parallelism.Secondly,the optimal model type and model parameters were obtained by training four mathematical models to diagnostic the transformer fault.The standard data set was used to verify that the algorithm has good adaptability to high dimensional feature vectors.Taking oil chromatographic data as an example,the recognition accuracy was compared with SVM algorithm.This paper proposes a hybrid fault diagnosis method based on variable prediction model(VPMCD)and stacked denoising autoencoder(SDAE).Firstly,the chromatographic data of transformer oil was collected and normalized.Secondly,SDAE was trained layer by layer to achieve the high-level feature representation and the parameters of network structure;Then,the optimal model type and model parameters of fault type were obtained by training four mathematical models of VPMCD.Finally,the model was fined-tuned by a small amount of labeled data to obtain the optimal network parameters and complete the fault diagnosis.The performance and robustness of the hybrid model were verified by applying it to feature extraction and fault diagnosis of oil chromatographic data.