Methodological Study On Fault Prediction And Condition Based Maintenance Optimization For Offshore Wind Turbine

Offshore wind turbine is a large-scale multi-functional system with complicated structures and various failure modes.It needs to bear a variety of marine environmental loads in service and its fault rate is high.Once an accident occurs,not only the economic loss of the induced breakdown is serious,but also other adverse effects such as marine environmental pollution will be caused.Moreover,offshore wind turbine needs large maintenance investment because of high cost of hoisting,poor accessibility of ships under serious sea conditions and long time of failure and outage.Fault prediction technology can disclose the development trend of system health state and provide decision for maintenance,which can increase the reliability and utilization of equipments,and reduce the loss of turbines of breakdown time.Condition-based maintenance is conducive to formulate maintenance plan for offshore wind turbine system and change passive maintenance into positive maintenance.Fault prediction and condition-based maintenance are key technologies for equipment and systems to achieve self-support.However,the maintenance methods of offshore wind turbine system still need to be improved to increase its reliability and reduce overall maintenance cost.As a result,in this paper the fault prediction and maintenance method of the offshore wind turbine system are studied.Specifically,in terms of the research object of offshore wind turbine,its main operation and maintenance problems in generation industry are analyzed based on the progressive relationship between structural damage identification and prediction and system maintenance method based on practical situations and comprehensive maintenance strategic application in the offshore wind power plant.The detailed contents include:(1)Since the tower structure is the foudation of an offshore wind turbine system,tower structural damages may bring risks of complete wind turbine overthrowing.The tower structural damage prediction method for offshore wind turbine based on neural network stepwise recognition was studied in this paper.Firstly,six experiments of different tower structural states,including intact structure,different damage locations and different damage degrees,are carried out.The modal calculation of the tower structural damage was performed using the finite element analysis software Abaqus.The finite element model is corrected by the measurement data of the intact tower experimental structural model,and the data is compared with that of damage conditions of other tower structural model.The software Abaqus is used to generate more data groups of damage conditions to train the neural network prediction model.The generalized neural network and improved BP neural network method based on genetic algorithm(GA)are applied to the equipment damage location and damage degree prediction of offshore wind power system step by step.The generalized regression neural network reduces the uncertainty of BP network parameter setting.The BP neural network improved by GA solves the problem that the traditional BP network is easy to be trapped into local minimum,thus improving the prediction accuracy of the network.The training results demonstrate that the stepwise prediction method can reduce the network complexity effectively,decrease training samples of the network significantly,increase the damage prediction accuracy of spud legs,and save training time accordingly.(2)An optimization approach of opportunistic condition-based maintenance for offshore wind turbines is proposed based on the actual operation health status of main equipment and the economic correlation between maintenance of different equipment.Artificial neural network is used to predict life percentage by leveraging the condition monitoring information.The conditional failure probability is calculated according to the predicted failure time distribution of components,which is used to indicate the deterioration degree of offshore wind turbine(OWT)equipment.The state-opportunity maintenance strategy proposed in this paper is determined by two-order failure probability threshold.Its value can be optimized by minimizing the average maintenance cost per unit operating time,thus minimizing the long-term maintenance cost.The results show that the state-opportunity maintenance strategy can consider the historical failure data and actual operation status of OWT equipment comprehensively,and save more maintenance cost than preventive maintenance method of regular intervals,thus the effectiveness of the method is verified.Moreover,this method is proved to be necessary by the comparison of the expense of OWT with onshore WT.(3)Influenced by weather conditions,the equipment maintenance of offshore wind turbine system is challenging with the effects of poor accessibility and serious failure consequences.Therefore,it is necessary to study the optimized strategy of comprehensive maintenance for offshore wind farm,with considerations to the influences of imperfect equipment maintenance and weather environmental accessibility on maintenance of offshore wind turbine.In this paper,the opportunity group maintenance model of offshore wind farm is proposed.The maintenance correlations between systems,equipments as well as breakdown losses are taken into account in the model,thus achieving the coordination of maintenance activities of different systems and different equipments.The proposed model is applied to calculate the maintenance cost of Dafengtian offshore wind farm.The results prove that the proposed model could increase the total availability of the wind power system and reduce total maintenance cost.The influences of the uncertainty of repair degree in the actual engineering on system state were considered in the optimization process.Moreover,the considerations of weather availability not only conform to actual offshore maintenance engineering,but also may rent and schedule ships appropriately,thus reducing the fixed maintenance cost of offshore wind farm.The case study results prove the practicability and superiority of the proposed model.It can realize long-term dynamic optimization of offshore wind farm maintenance activities.And the research conclusions provide important suggestions to actual maintenance of Dafengtian offshore wind farm.