Regular maintenance inspection and testing is essential in extending cable life and reducing failure probability. This can be achieved by improving the conduit conditions and taking corrective actions on faulty cable components and accessories. Regulators and corporate governance among power utilities require the implementation of risk-based approaches to asset management. However, practitioners lack sufficient historical event data and knowledge that allow them to determine the failure probability of individual cable components, which is an essential component for risk assessment, due to that the high voltage (HV) cable population are relatively young, and many have not yet reached the end of their design life. This paper presents a novel holistic approach to allow the risk based maintenance strategy to be conveniently implemented for the cable conduit, cable terminations, joints, main bodies and the earthing systems separately for each cable circuit. Contributions include: (i) a failure frequency model which accounts for every past failure record of individual cable circuit components to approximate the probability of failure. This, when multiplying with the cable importance or failure consequence, yields the risk level of an individual cable component or a cable circuit; and (ii) a method of optimally scheduling the maintenance activities by setting the objective functions as the minimal cable system risk. The benefit of the simple failure frequency model has the advantage of not having to depend on human intervention and it does not need a large sample to generate valid results, as is the case with other statistical methods. Results of applying the proposed maintenance scheduling model to 21 selected High Voltage (HV) cable circuits show that the average risk can be significantly reduced while continuing with the same number of inspections and test operations.INDEX TERMS Power cable, risk based maintenance, probability of failure, maintenance scheduling. I. INTRODUCTION

No abstract

Abstract. According to prediction of landslide based on grey vector machine model, the landslide is affected by multiple factors such as soil layer structure and these factors show complicated non-linear relation with landslide, while it is hard for the traditional model to reflect the non-linear complicated relation. The prediction model for law of landslide established by grey model and neural network model can better predict the complicated surface subsidence and can effectively improve the predicted precision compared with the single model. Not only the reliability of prediction system is improved, but also the cost of the whole system is reduced.