基于IBOA-ERF的风力发电机齿轮箱故障检测

Fault Detection Method of Wind Turbine Gearbox Based on IBOA-ERF

针对风力发电机齿轮箱故障检测模型参数优化难度大的问题,提出一种基于改进的蝶形优化算法的极端随机森林故障检测模型。将故障检测模型虚警率与漏警率的代数和构造为适应度函数,改进个体初始位置和位置更新策略。引入混沌映射策略代替原有的种群初始化方法,增强初始种群分布的随机性。提出一种自适应惯性权重因子,结合鸽群优化算法的地标算子更新种群位置迭代方程,加快收敛速度,提高蝴蝶优化算法的多样性和鲁棒性。采用局部搜索阶段和全局搜索阶段的动态切换方法,实现全局搜索与局部搜索的动态平衡,避免陷入局部最优。建立极端随机森林故障检测模型,利用改进的蝶形优化算法获取最优参数,实现所提模型在高维数据下具有良好的鲁棒性和泛化性的快速响应。与其他优化算法相比,所提风力发电机组齿轮箱故障检测方法具有较低的误报率和漏报率。

Considering the difficulty of parameter optimization of wind turbine gearbox fault detection model,this paper proposes an extreme random forest fault detection model based on improved butterfly optimization algorithm(IBOA-ERF)optimization.The algebraic sum of the false alarm rate and the missed alarm rate of the fault detection model is constructed as the fitness function,and the individual initial position and position update strategy are improved.The chaotic mapping strategy is introduced to replace the original population initialization method to enhance the randomness of the initial population distribution.An adaptive inertia weight factor is proposed,which combines the landmark operator of the pigeon swarm optimization algorithm to update the population position iteration equation,accelerates the convergence speed,and improves the diversity and robustness of the butterfly optimization algorithm.The dynamic switching method of local search stage and global search stage is adopted to realize the dynamic balance between global exploration and local search and avoid falling into local optimum.The ERF fault detection model is established,and the improved butterfly optimization algorithm is used to obtain the optimal parametersto realize the fast response of the proposed model with good robustness and generalization under high-dimensional data.Compared with other optimization algorithms,the proposed wind turbine gearbox fault detection method has lower false alarm rate and false negative rate.