Design And Application Of 35 KV Transformer Condition Monitoring System Reconstruction Scheme

As one of the important infrastructures of the power system,the stable operation of substation is directly related to people’s production and life.To ensure the stable operation of substation is an important task for relevant departments of electric power.35kV transformer has been widely used,and with the increase of electricity demand,the operating pressure of 35kV substation is increasing,and it is easy to appear abnormal or even fault in the operation process.In view of the transformation of the 35kV substation condition monitoring system,this paper has completed the following work:In order to solve the problem that the current transformer state evaluation index is single and the fuzzy index is not considered,a state evaluation system of analytic hierarchy process（AHP）based on convolutional self-coding neural network is established according to the monitoring data and inspection records.Firstly,4 item layer indicators and 23 index layer indicators are selected,and each indicator is quantified and normalized,and the subjective and objective weight,combined weight and transformer operation state deterioration degree of each indicator are determined by fuzzy algorithm and entropy weight method.Then the transformer data containing the degradation label is used as input and output data to train the convolutional self-coding neural network.Finally,the feasibility of the whole scheme is verified by case analysis.Aiming at the limitations of current fault diagnosis methods,a comprehensive fault diagnosis scheme for 35kV transformer based on convolutional sparse deep confidence network（CNN-SDBN）was proposed.Convolutional neural network was used to extract the features of various monitoring data to construct a new sample set,and sparse deep confidence network was used to classify and recognize the new sample set to complete fault diagnosis.The sparse coefficient is introduced to improve the traditional RBM to improve the recognition accuracy and speed up the convergence of the system.Finally,the fault records strictly selected in recent years are taken as the training set and verification set,and the parameters corresponding to the optimal performance of the network are determined by comparing the accuracy of different network depths and training times,and compared with other intelligent methods.In view of the existing hardware facilities are not perfect,the hardware parts of oil chromatography analysis,partial discharge,insulation bushing,winding temperature and iron core grounding current status monitoring subsystem are designed and reformed,and the intelligent module TMS320F2812 is used to design the data acquisition and data upload unit.In the software part,a comprehensive condition monitoring system and a fault diagnosis system based on CNN-SDBN are developed by Lab VIEW and the tests are completed.The effect of the system is verified by monitoring and diagnosing transformer operation status in a period of time.To sum up,through the transformation of substation monitoring system,the remote monitoring,condition evaluation and fault diagnosis of transformer working conditions were realized,which ensures its safe and stable operation to a certain extent.