Quantum Neural Network Based On The Dga Transformer Condition Based Maintenance In Applied Research

With the progress of modern society and the rapid economic development, the high reliability and economic requirements of the power system operation promotes the power company throughout the country to start the projects of electrical equipment Condition-based Maintenance. Based on the research and development work of the project-"Power Transmission Equipment Condition-based Maintenance Decision Support System" in the Electric Power Research Institute of Jilin Province, this paper studied the implementation process of condition-based maintenance to the typical power transformer equipment and focused on the application of intelligent algorithm in the state diagnostic module.Firstly, the common types of transformer faults, reasons, Dissolved Gas Analysis (DGA) and traditional diagnostic methods based on DGA are carefully analyzed and researched, and the defects of these methods such as needing to know a fault happened, fault code incompletes, combined effects of a variety of failures can't be judged, failure status can't be fully reflected and so on are specified.Against the deficiencies and shortcomings of traditional diagnostic methods, the emerging intelligent algorithms-Quantum Neural Networks (QNN) is applied to the state diagnostic module for the first time in this paper, and two models of QNN which are respectively based on Quantum superposition state ideology and quantum phase are adopted, for they can show their advantages on the traditional computer. After the comparative analysis with BP neural network under the same input through the simulation tests, we obtained that the sample training results of Quantum state superposition QNN model and Quantum phase QNN model both have better generalization mean square error than BP neural network, and for the result of state diagnosis, the Quantum state superposition QNN model has higher accuracy and credibility than BP neural network while quantum phase QNN model has a fewer number of iterations than BP neural network and Quantum state superposition QNN model ,so it achieved the fast information processing of the status pattern recognition to a certain extent. The actual case analysis proved once again that the two QNN diagnostic models mentioned in this paper can both provide accurate and reliable results of the initial diagnosis, so as to supply a clear direction to the comprehensive analysis of the transformer condition-based maintenance system.Finally, the functional structure and technical architecture of the the overall strategy in transformer condition-based maintenance system is analized in this paper, and the important position and role of the intelligent algorithm analysis to transformer status diagnosis in the system functional structure is reflected. In the end, problems which exist in the realization of the transformer condition-based maintenance are briefly analized and the direction of efforts put into practice is discussed.