Studies And Applications On The On-line Decision-making System Of Condition-based Maintenance For Power Transmission And Transformation Equipment In Urban Power Grid

As there are a great number of power equipment,the workload is heavy as well.In order to reduce the failure probability and improve its performance,which has become an improtant measure of enhancing the euiqment performance and operation reliability of power system.Condition-based maintenance has got extensive publicity of its studies and applications,which has become the main development trend of modern maitenance mode.Because the euipment condition,influence consequence has obvious time-varying characteristics,which involves many different parts such as the achievement of power system operation state and topologic structure,topologic structure analysis of power system,failure development and evolution,load shedding and so on.It is hard to acheve the quantitative calculation,especially the on-line decision-making,which needs the supports of high efficiecy calculation methods and abundant computing resources.Thus,there are many unsolved problems of theory and implementation technology.With the support of the National High Technology Develoment and Research Project(863 project)and other projects,in order to meet the timeliness requirements of on-line analysis,this dissertation carries out the theoretical studies and simulation analysis on the high efficiency calculation methods of function modules,short-term and long-term maintenance strategies optimaziton,engineering applicaions of condition-based maintenance system and some other aspects with the help of the multi-source data fusion and efficient computing engine on the established big data platform.According to the practical requirements,This dissertation provides a structure schema for the on-line condition-based maintenance systemthe,which has three layers: underlying data layer,function analysis layer and application presentation layer.For the underlying data layer,the multi-source data fusion and the data normalization processing are the main problems to be solved.In order to resolve the CIM and E documents,which are the key information of power system modeling,the unified data specification and interface formats for the multi-source data are established.For the function analysis layer,the main purposes and efficient computing methods of each module are studied,which pave the roads for system development.For the application presentation layer,from the perspectives of basic information and maintenance strategies for power transmission and transformation equipment,the presentation content and requirements are studied,and the presentation schemes are proposed.These above studies lay the solid foundations of practical system development and engineering applicaiton.As there are a great number of power euipment in the urban power system,as well as the complicated bus connection modes,the condition-based maintenance,especially the short-term maintenance involves a lot of analysis and calculation.The key problerm is to improve the calculation efficiency of each model.This dissertation proposes some theoretical research of the efficient computing methods such as power system topologic analysis,failure development and evolution studies,load-shedding model and parallel computing.For the topologic,a efficient topologic analysis method based on breaker classfication is proposed.For the failure development and evolution,a deduciton model is established by taking the action behavior of both the protection and the safety automatic device into consideration,which can correctly deduce the evolution process of each different failure.By sensitivity analysis,the load-shedding model is brought out,which can effectively improve the calculation speed of load shedding.From the perspectives of inter-task and within-task,the parallel computing strategy based on Spark platform is proposed,which can solve the problerms of timeliness by the serial calculating methods.Since the health state evaluation based maintenance strategy for the power transmission and transformation equipment cannot quantitatively determine the best maintenance cycle,it cannot effectively guide the engineering practice.According to the view of better engineering applications,this dissertation proposes a novel condition based maintenance strategy for power equipment based on failure probability analysis.This method firstly sets up equipment’s basic failure rate model between failure rate and time based on its historical data.In accordance with the periodic maintenance experience,it determines the equipment’s failure probability accumulated threshold value.Taking equipment’s familial defects,individual defects,maintenance behavior and other modifying factors into account,it puts forward the real failure rate model for the different individual equipment.Based on the real failure rate model,combining the equipment’s current health state evaluation value and the failure probability accumulated threshold value,this paper works out its maintenance strategy.The proposed maintenance cycle can provide a more intuitive and effectively support when making the mid-term or long-term maintenance plans for power equipment.The current research on condition-based maintenance of the power equipment mainly concerns on mid-term or long-term maintenance,which cannot meet the requirements of short-term especially temporary maintenance.The short-term maintenance strategy can make dynamic on-line decision according to the work condition of both the equipment and power system,which can effectively cover the shortages of mid-term and long-term mainteannce,which is an urgent unsolved problerm in the field of condition-based maintenance.According to the engineering application,this dissertation proposes a short-term maintenance strategy based on risk-cost analysis.For the equipment working in bad health condition,its operation risk costs and maintenance costs are calculated,then the latest maintenance time is calculated based on the principle that its operation risk costs are no higher than maintenance costs.Using of the latest maintenance time,the best maintenance time is calculated by setting the maximum relative earnings of postponing maintenance as the target,which can provide the operation staffs with comprehensive maintenance-decision support.In the end,simulaiton cases are used to verify the validity of the proposed short-term maintenance strategy.Based on the above theoretical analysis,and in accordance with the requirements of demonstration projece engineering,the engineering applications of the proposed high efficiency calculation modulars and condition-based maintenance strategies for power transmission and transformation equipment are carried out with the help of the established big-data platform and efficient computing engine.The engineering application results demonstrate that the established on-line decision-making system can provide instructive strategies with abundant functions,which has a positive influence in promoting the development and engineering application of condition-based maintenance.