Research On Transformer State Detection And It’s Application

Maintenance of transformer has great significance to the security and stability ofpower system, as well as extending equipment life, reducing outage losses and improvingeconomic benefits. With the voltage level of transformer gradually rising, equipmentimportance, outage losses, maintenance manpower and material costs have significantlyincreased, and the shortcomings of periodic maintenance gradually become obvious. Inorder to enhance pertinence of maintenance scheduling and system reliability benefits,and reduce maintenance costs simultaneously, this thesis researches the transformercondition based maintenance scheduling problem from two aspects. The first is toenhance pertinence of maintenance scheduling by researching the multi-state equipmentmaintenance reliability model based on condition monitoring and assessment techniquesfor device status information. The second is to add the system operation information toequipment maintenance scheduling problem to harmony the relationship betweenmaintenance and the system operation, and balance the system reliability and economy ofmaintenance scheduling in system operation vision to gain the optimal strategy for thetransformer maintenance scheduling.Firstly, on basis of the different natures and modes of failure, this thesis divides thefault into random failure and accumulative failure. Based on equipment conditionmonitoring and assessment results, accumulative failures can be perceived, assessed,tracked and predicted. According to maintenance practices the multi-state maintenanceprobabilistic reliability model which can consider the effects of maintenance decision isestablished. This model can contacts equipment condition assessment with reliabilityevaluation depicts the effect of randomness and uncertainty of the equipment statetransition process and maintenance effects. Furthermore, based on timely varyingequipment state sequence which can be simulated by using state duration samplingmethod among the maintenance cycle, system reliability indexes in maintenance periodscan be calculated, and the economic indexes of manpower and resources costs in relationwith maintenance decisions can be got at the same time by statistic method.Secondly, this issue establishs transformer maintenance scheduling optimizationmodel. The objective function of this model is to minimize the sum total costs of thetransimission system maintenance schedule stategy which includes manpower andresources costs, equipment replacement costs and reliability indexes which is converted tothe economic indexes using the interrupted energy assessment rate. Then, maintenancestrategies constraints and system operation constraints is added to this model respectively.In order to solve this model, this thesis proposed a new encoding method of transformermaintenance decision which codes the start time and type of maintenance to variablelength chromosomes (Messy Genetic Algorithm). Afterwards, using filters and penaltyfunction to deal with constraints. It can be proved that this method can improves the efficiency of genetic algorithm and reduces storage space.This thesis implement proposed model by C++programming language. FinallyGuangdong220kV grid was used to test the model and algorithm. It shows that theproposed model and algorithm can make reasonable maintenance Scheduling under thecondition monitoring. The example results verify the validity and superiority of thealgorithm.