基于多重优化策略的风力机叶片传感器布置方法

OPTIMAL SENSOR PLACEMENT FOR WIND TURBINE BLADE BASED ON MULTIPLE STRATEGIES

针对叶片结构健康监测系统中不合理的测点布置影响传感器数据采集的有效性,降低监测系统精度的问题,提出了一种运用多种优化策略组合布设的方法。首先为了提高收敛速度,利用QR分解法筛选初始测点群;其次基于模态置信准则(MAC准则)建立优化目标函数,并提出了通过MAC最大非对角元素设定阀值作为确定最佳测点数量的方法;最后为解决传统算法易陷入局部最优解、收敛速度慢的问题,将模拟退火算法作为一个算子加入遗传算法中,把两种成熟的智能算法有机地结合到Memetic框架下,利用互补优势形成Memetic算法实现传感器的优化布设。以2MW级风机叶片为例,利用该算法进行传感器优化布置并与传统QR分解法和标准遗传算法进行比较,结果表明,该方法能有效解决叶片传感器优化布置问题,并较QR分解法和遗传算法有更好的收敛性能和优化效果。

In order to collect date by sensors effectively and improve the accuracy of wind turbine blade＇s monitoring system,a new sensor placement method based on the multiple optimization strategies was presented. Firstly,to improve the convergence rate,QR-factorization method was used to screen the initial test point group. Secondly,the modal assurance criterion（ MAC） matrix was taken as the target function,and a method of determining sensors number was proposed by the maximum off-diagonal element of MAC. Finally,in order to solve the problem that the traditional genetic algorithm is easy to fall into,local optimal solution and slow convergence SA was used as an operator of the genetic algorithm. The two kinds of mature intelligent algorithm was combined under the framework of Memetic by complementary advantage. The plan of optimal sensor placement for a 2-MW wind turbine blade was performed and compared with the traditional QR method and the standard genetic algorithm. The final results indicate that the arrangement through the new optimal method could solve the optimal sensor placement of blade and have better performance in convergence and optimization effect than the QR method and genetic algorithm.