| With the rapid development of the economy,issues such as energy shortage and environmental pollution have become increasingly prominent.Seeking new energy development models and achieving sustainable energy development have become the focus of global attention.As a renewable and clean energy resource,wind energy is abundant and widely distributed,and its commercial development value has attracted the attention of countries all over the world.However,the structure of wind turbine is complex and the fault frequency is high.The frequent operation failures not only greatly reduce the operating efficiency of the wind turbine,but also affect the safety and stability of the grid.The maintenance quality of wind turbine equipment directly determines the reliability and economic benefits of the whole wind farm.The wind farm is proposed as the research object,and the reliability assessment methods,maintenance strategy optimization,spare parts management and maintenance resource allocation are studied in this paper.A reliability assessment model for wind turbine components based on SCADA data and BP neural network is established in this paper.The BP neural network model is trained with SCADA data under initial normal conditions to obtain the normal state parameter prediction model and prediction error distribution of key components of the wind turbine.Based on the prediction error,statistical methods are used to quantitatively evaluate the reliability of components in real time.The validity of the proposed method in wind turbine reliability assessment is verified by using real SCADA data collected from a wind farm.An opportunistic maintenance approach is proposed for wind turbines considering an imperfect maintenance schedule that is based on reliability.Taking the key components of a single wind turbine,such as blade,gearbox,generator and bearing,as the research object,this paper studies the maintenance correlation between each component and the multilevel maintenance control limits,and establishes an opportunistic maintenance model considering the multilevel maintenance limits and multiple maintenance methods.Based on the wind turbine average maintenance cost,the economic advantages of the opportunistic maintenance and the optimal maintenance threshold are discussed.In the specific simulation example,the average maintenance cost of a wind turbine system considering opportunistic maintenance strategy is 158.73$/day,which is lower than that not considering opportunistic maintenance strategy(209.81$/day).Through the fruit fly optimization algorithm,the minimal average maintenance cost in this example is obtained to be 139.86$/day,and the optimal opportunistic maintenance reliability threshold for wind turbine components is 0.9825.An opportunistic maintenance strategy for wind farm considering stochastic weather conditions and spare parts management is proposed.The Markov chain model is adopted to generate wind speed time series and obtain the maintenance wait time owing to weather restrictions.The opportunistic maintenance threshold is a dynamic value related to the wind speed to reduce power generation losses.Moreover,the(s,S)inventory policy is introduced to establish joint optimization of maintenance strategy and spare parts inventory management.Taking the operation and maintenance cost of the wind farm during operation period as the objective function,the economic advantages,optimal maintenance threshold and inventory management strategies of the proposed maintenance strategy model are discussed.Compared with a static opportunistic maintenance strategy and a strategy not considering opportunistic maintenance,the life cycle total cost for the proposed dynamic opportunistic maintenance strategy shows a decline of 10.86%and 18.29%,respectively.The minimal cost is obtained to be 3.36×10~7$with the opportunistic maintenance reliability threshold of 0.99 and reorder stock level of 0.This paper establishes an optimization model for wind farm maintenance resource allocation based on the dynamic programming theory of complex systems.The wind farm maintenance resource demand model is established according to the reliability model of wind power equipment and its current health status.Combined with the current maintenance resource quantity constraint,the fast elitist non-dominated sorting genetic algorithm(NSGA-Ⅱ)is used to establish the maintenance resources dynamic allocation model of regional wind farm groups.The optimal resource allocation scheme is solved for the specific simulation example,and the influences of related parameters are further discussed.The simulation results show that the maintenance resource allocation scheme given by the NSGA-Ⅱalgorithm after optimization is significantly improved compared with the scheme after initial decoding,and the total resources and maintenance loss cost required for carrying out maintenance tasks are basically at very low level.On the other hand,the more the initial resources are,the closer the optimal solution distribution is to the left lower,that is,the smaller the objective function value is and the better the resource allocation scheme is.Finally,a wind farm maintenance and resource allocation support system is established in this paper,which is compiled by the Matlab-GUI.A group of running examples have been given to illustrate the practicality of the system in wind farm maintenance task planning and resource allocation. |
PDF Full Text Request