基于遗传算法的传感器优化布置

为解决空间网格结构模态测试中的传感器位置优化布置问题,提高采集数据信息的完备性,采用遗传算法优化传感器布置位置.在解的编码过程中采用二维数组,每一行存储一个可行解;采用强制变异避免遗传算子操作过程中出现同一个位置重复布置的问题.为了提高收敛速度,将基于模态矩阵QR分解的传感器优化布置结果作为第一代父群.对单层球面网壳进行优化计算后,将基于遗传算法的优化结果和QR分解得到的结果进行了应变能指标和相关性矩阵(MAC)的比较,发现前者不仅在应变能指标有大幅度的提高,而且相关性矩阵也有相应改善.

Optimization of sensor placement by genetic algorithms

The genetic algorithms was adopted to optimize sensor placement and enhance completeness of the test data in the space grid structure modal test. A two-dimension array was used to code the solutions, every feasible solution was denoted by the row in the array. Force mutation was conducted when a feasible solution reapeared in the same position after crossover operating. The first solution group was founded based on the placement optimization from QR decomposition of modal matrix, and the convergence of the algorithms would be faster. The technique was used to optimize the sensor placement of a simple single-layer reticulated shell. Comparing the optimized sensor placement by genetic algorithms with that by QR decomposition of modal matrix about the strain energy and modal assurance criterion, genetic algorithms perform better.