Condition Assessment For Power Transformer Based On Improved Grey System Theory

Power transformers are very important equipment of substation,power plant and electric management departments,the running conditions of which will affect the safety and stability of system. According to the transformer evaluation results, formulating the corresponding maintenance strategy, can reduce maintenance costs and improve the reliability of the transformer, extend the service deadline and working life of transformer, so as to ensure the safty and stability of the power system.Maintenance based on condition of the transformer is a kind of new forms of management mode, to reduce the maintenance costs and improve the safety index, must establish new concept. The transformer condition assessment is the core of condition based maintenance and the basic guarantee of state administration. So realizing the condition assessment of transformer can further achieve the condition based maintenance.This paper improves the grey model GM (1,1) based on the grey theory, establishes unequal-interval GM (1,1) model, which using the space between the sequence as multiplicator to generate a new sequence. The new sequence can broaden the applicable scope of the original model and improve the accuracy of the predictions. Considering the influence on calculated value of gas dissolved in oil by the emergency events, this paper establishes GM(1,1) model including buffer operator,which can avoid the predicting distortion resulted from abrupt changing data.Applying two model to condition assessment of transformer can obtain better forecasting results. Using scoring model of centesimal system in condition assessment of power transformer is better for conditions based maintenance than"true or false rule"model,which just divided the conditions of transformer into two type——conformity and nonconformity. According to the results of rating,conditions of power transformer will be divided into five categories in order to propose specific maintenance strategy, which has obvious advantages.This paper establishes the model of transformer conditions assessment,considering both unequal-interval of sequence and the effects of abrupt changing data, in order to improve accuracy of transformer conditions assessment model. Using scoring model of centesimal system divides conditions of transformer into five categories can propose specific maintenance suggestions. For transformers which need immediate repairment or maintenance as soon as possible,this paper also provide possible fault for maintenance personnel, offering the theory basis for their decision-making. Finally,this paper use two examples to prove that this model of transformer conditions assessment are effective and superior .