Research on Selective Maintenance Strategy of Box Diesel Generator

Research on Selective Maintenance Strategy of Box Diesel Generator

With the development of the hydropower unit system in the direction of large-scale, complex, and precision, the generator system presents a variety of different operating performance states or failure modes during the service phase. Traditional maintenance methods based on fixed cycles can no longer meet the interests of modern companies in the market competition.

The state maintenance method is based on the information monitored by the advanced state monitoring and fault diagnosis technology, and is scheduled to be repaired in a timely manner so as to avoid the huge loss caused by sudden damage to the equipment. The Markov decision model used in condition-based maintenance methods describes the degradation law of component performance states and predicts the future component reliability trend, and the obtained results are better. The key to this method is to use the Markov process to describe the performance degradation of parts. However, the Markov process usually assumes that the transfer time of equipment between any two states obeys the exponential distribution. This condition seriously affects the application of the Markov process in practice.

Therefore, the semi-Markov model is used to describe the performance degradation of hydropower unit components. The advantage of the semi-Markov process is that the transfer time between the two states of the equipment described can be a non-exponential distribution, which is very important in practice. . Generator failure due to bearing breakage on the generator shaft can be described well with Weibull distribution. Therefore, the performance degradation of the components of the generator set is not suitable for describing using the Markov process. The semi-Markov process is more suitable than the Markov process to describe the performance degradation rules of the generator set components.

In this paper, based on the state maintenance of hydropower generating units, the research on the performance degradation of hydropower unit components and maintenance decision-making optimization are carried out. The research work mainly includes:

(1) For the fact that the Markov process is only applicable to the transition time between the two states of the fitting component obeys the exponential distribution, in order to more accurately reveal the law of performance degradation of the generator set components and predict the reliability change trend in the future time, Establish an effective model that is more in line with the description of the actual system performance state. A semi-Markov process is used to describe the performance degradation of the components of the generator set. A reliability evaluation model that meets the life distribution characteristics of the generator set is constructed. Case studies have verified the accuracy and feasibility of the proposed model.

(2) Common equipment system control models ignore maintenance costs that change with time. For this issue, this study proposes a dynamic threshold control model for state maintenance based on maintenance cost changes. The purpose of this model is to improve the reliability of the system by reducing the probability of preventive maintenance during periods of high downtime costs by achieving lower maintenance cost targets and increasing the probability of preventive maintenance during periods of low downtime costs. Through examples, comparative analysis of the genset equipment maintenance threshold strategy further validates the advantages of the dynamic proportional threshold strategy based on the cost of downtime relative to the constant threshold strategy.

(3) Aiming at the problem of maintenance decisions for the genset components during the mission interval, a selective maintenance decision model based on the component status is proposed. The model evaluates the reliability of the genset system from the perspective of the power system and takes into account the component status. Information and non-perfect maintenance factors, the use of particle swarm optimization algorithm to solve the problem of selective maintenance decision-making optimization of generating units. Finally, through verification, under the same maintenance cost budget, the proposed method can make the task completion rate of the generator set.