Research On Evaluation Method Of Bushing Condition On Valve-side Of Converter Transformer

The converter transformer is the key equipment of the DC transmission system and occupies an important position in the power system.The valve-side bushing is an outlet device that leads the inner winding lead to the outside of the oil tank,and must be operated within the scope of reliable regulations.However,the valve-side bushing is affected by AC and DC voltage or even reverse voltage for a long time,and the possibility of failure is greater.At present,there is no effective method to evaluate the condition of the valve-side bushing.The research in this paper is carried out to accurately predict the operation situation of the valve-side bushing.According to the operating experience and technical guidelines,the selection principle of the bushing index on the converter valve-side is determined.At the same time,various information on the converter valve-side bushing and fault statistics cases are analyzed,and the evaluation index set that effectively reflects the casing status is summarized.Then,considering the fuzziness and randomness of the evaluation boundary,the extension analytic hierarchy process is used to construct the judgment matrix,and the subjective constant weight is determined by calculating the normalized eigenvectors with positive components.In order to avoid relying too much on expert experience,the objective constant weight is obtained by improving the entropy weight method,and then the comprehensive constant weight factor is determined by weighted integration.Aiming at the situation that individual indicators are seriously degraded and the constant weight cannot be sensitively reflected,a penalty-type state variable weight function is proposed,and the variable weight principle is used to modify the constant weight to make the weight distribution more reasonable.A standard cloud matter-element model is established based on the extension cloud theory,and the cloud correlation degree between the matter-evaluation and the standard cloud is calculated,which realizes the accurate evaluation of the converter valve-side bushing,and introduces the confidence factor to verify the credibility of the results.The effectiveness of the method is verified by an example.In order to avoid relying too much on empirical evaluation and make better use of historical data,a deep belief network based deep learning state evaluation method is proposed.The method is data-driven.First,the model is initialized by setting the parameters and hyperparameters of the deep belief network,and model training and verification.The model was optimized by adjusting the hyperparameters.After the completion of the modeling and the initialization experiment,it was shown that the correct recognition rate of the optimized model increased by 9%.At the same time,the method in this chapter is used to verify the example in Chapter 3,and the results show that this method is feasible.Based on the above evaluation methods,a set of converter variable valve-side bushing condition evaluation system based on SQL Server 2012 database and Java programming language was developed,and each module of the system was introduced.The system runs reliably,saves resources,improves the evaluation efficiency of the converter valve-side bushing,and has certain guiding significance for its condition maintenance.