Study On Method To Decision-Making Of Condition-Based Maintenance For Power Transformer

The operation reliability of the power transformer, the key equipment in the power system, is one of the key factors related with the safety and stability of the power system. For a long time, its time-based tests, checks and maintenances out of service, which played a very important role in guaranteeing safety of the equipment operation. are carried out based on the "Regulation on the preventive test of electric equipment" . This kind of maintenance mechanism has The main problem of the widely used maintenance code of the transformer is that the transformer-needed maintenance has not been maintained because the overhaul period has not come; the transformer not needed maintenance has been maintained in accordance with the overhaul period. This made a lot of manpower and material wasted. The Condition-Based Maintenance(CBM) is a method that maintenances are arranged according to the health state of equipment by monitoring the condition of the equipment. The CBM technology has strong pertinency and is economically rationable. With the widely accepted technology of the Condition-Based Maintenance, it replaces Time-Based Maintenance will be the inevitable developing trend.The basic flow chart of Condition-Based Maintenance to the transformer is to assess the health state of transformer under service based on data obtained through the timed preventive test or the on-line monitoring, and the history recordings of the transformer, then to give some judgments in terms of the codes of Condition-Based Maintenance, and to put forward some maintenance scenarios in term of its actual situation, finally to select the optimal scenario from limited scenarios according to the result of the technical and the economical comparison. In the process, it involves many assessment targets and many decision makers. Based on a lot of relevant references, the paper mainly finished the reserches of the following contents:The rationale and method of adopting grey clustering decision-making to health state of transformer are presented. Firstly, the states of transformer are divided into five grey classes, then the whitenization weight function of the five grey classes are established, the state information involved in assessing the state of transformer are analyzed, and finally the steps of grey clustering decision-making to the state of transformer are presented.In the operation of transformer and all kinds of tests, there is a lot of information reflecting the state of transformer and the rules of Condition-Based Maintenance can be shaped according to these information. But in this information, some are independent and some are unnecessary. Based on deep research of Rough Set Theory, the paper puts forward the method of knowledge reduct to mine rules from this information and consequently obtained the rules of Condition-Based Maintenance to transformer.The paper introduces the grey incidence decision-making method to Condition- Based Maintenance decision-making of transformer. The computing method of degree of grey incidence and some countermeasures are analyzed. The decision-making targets of Condition-Based Maintenance to transformer are assessed. Steps of grey incidence decision-making to Condition-Based Maintenance decision-making of transformer are presented.The decision-making of the Condition-Based Maintenance of the transformer involves many assessment targets in decision-making. And the judgments of these targets are made by domain experts and engineers. The judgment cannot be measured by a strict standard but a fuzzy concept. So, the author introduces the method of the fuzzy decision-making to Condition-Based Maintenance decision-making of transformer, puts forward to adopt the method of the fuzzy multiple attribute decision-making in order to analyzing the Condition-Based Maintenance decision-making of transformer, and researches the decision-making targets and maintenance scenarios. By analyzing an actual example, the paper assesses the maintenance scenarios using the fuzzy simple mean-weighted method and the fuzz...