Oil immersed type reactor fault diagnosis method based on two-way deep network

Abstract

The invention discloses an oil immersed type reactor fault diagnosis method based on a two-way deep network. The method is characterized by comprising the following steps of constructing a reactor fault diagnosis model based on the two-way deep network; selecting training sample data and characteristic variables, performing normalization processing on the sample data, and then dividing the sampledata into a training set and a testing set according to a certain proportion; dividing the reactor fault status, and encoding the reactor fault status; adopting a greedy algorithm for obtaining an initial value of network parameters, and performing pre-training and fine adjustment on the network model; saving the trained network, performing performance analysis, selecting the diagnosis result accuracy as a unique index and performing diagnosis performance contrast with a traditional diagnosis method. The method has the advantages that the oil immersed type reactor fault diagnosis method basedon the two-way deep network is applicable to training large data volume samples, operability is high, compared with a neural network and other traditional fault diagnosis methods, the fault diagnosisaccuracy is higher, and the final result obtained by diagnosis has high reliability.

Description

technical field [0001] The invention relates to a fault diagnosis method for an oil-immersed reactor based on a bidirectional deep network, and belongs to the technical field of state detection and fault diagnosis of electric equipment. Background technique [0002] With the increasing demand in the power market, the system has higher and higher requirements for the operation reliability of power equipment. Reactor has multiple functions such as improving the reactive power related operating conditions of the power system. It is a common and important power equipment in the power system. Its safe and stable operation is a prerequisite for reliable power supply of the power system. [0003] The type and content of dissolved gas in oil-immersed reactor oil correspond to different fault types. Oil chromatographic analysis (Dissolved Gas Analysis, DGA) is to monitor the early failure of equipment by analyzing the characteristic gas that has been dissolved in the oil. At presen...

AI Technical Summary

Problems solved by technology

The learning of two-way network combines the training methods of feedforward network and feedback network, which overcomes the problem that traditional neural network is not suitable for multi-layer network training

Images