基于故障-测点互信息的传感器多目标优化配置

Optimization of sensor allocation based on fault-measurement point mutual information

针对大型流程工业系统状态监测故障诊断的传感器优化配置特点,建立能够定量反映测点获取故障信息效率的故障-测点互信息矩阵.提出传感器系统测点-故障关联度优化目标函数,建立兼顾测点获取故障信息效率、传感器系统可靠性、经济性的多目标优化配置模型.设计改进的非支配排序遗传算法(NSGA-II),在保证故障可检测性和可分辨性的前提下,获得兼顾3大优化目标的Pareto前沿解集.电站除氧器的实例分析表明了该方法的可行性和有效性.

A matrix of fault-measurement point mutual information was proposed which can quantitatively reflect the efficiency of the fault diagnose information obtained by the measurement point in order to find the optimized allocation of sensors for large process industries to meet the fault diagnose requirement.An optimized target function based on the correlation of measurement point to fault on the sensor system was created.A multi-objective optimized allocation model was conducted which can consider the following three things： the efficiency of the fault diagnosis information obtained by the measurement point,the reliability of sensor systems and the cost factor.A promoted non-dominated sorted genetic algorithm-II（NSGA-II） was designed,which can gain Pareto front solutions including three above-mentioned optimized-targets under the premise of assuring the detectability and separability of fault.The case of power plant deaerator shows that the method is feasible and effective.