Research On Condition Evaluation And Condition-based Maintenance For Distribution Equipment Based On Multi Information Source

In order to ensure safe operation of distribution equipment and cut down the maintenance cost, condition-based maintenance should be adopted instead of traditional time-based maintenance.Some basic theories of condition-based maintenance are still not mature for further application. For example, current researches emphasized information of inspection or experiment but did not pay enough attention on other kinds of available data, especially real-time or history operation data. This has led to lower timeliness and availability of condition-based maintenance.By utilizing available data as basis, this thesis classified different types of distribution information (including regular-interval inspection information, real-time operation information and history record information), and integrated the information using data mining technology and information fusion technology. Then condition evaluation models based on integration multi source information was established. By using the evaluation result, further research on condition-based maintenance decision-making was studied. The following achievements were obtained in this thesis.(1) Condition evaluation model based on real-time operation information was established. The model is able to early warn some faults and has good timeliness and universality so that it can be a beneficial supplement to current condition evaluation scheme.(2) History fault or failure records were utilized. The results of text mining were then calculated into condition evaluation score, forming a condition evaluation method based on history operation information. Then condition evaluation model based on multi information source was established by integrating different types of information. The model utilizes real-time data and history data to comprehensively reflect the operation performance, history condition and real-time condition of equipment. The evaluation result meets the requirement of timeliness and comprehensiveness and accords with fault records and theoretical rules.(3) Condition-based maintenance decision-making method utilizing condition evaluation result based on multi information source was proposed. The method adopts D-S evidence theory to establish a further application of condition evaluation result. The proposed method realizes comprehensive group decision-making of both maintenance type and maintenance intervals and was able to be easily adjusted according to preferences of decision-makers. The decision-making results can realize global optimization under comprehensive indicators.