电子系统健康状态监测数据优化算法

为解决电子系统健康状态监测数据的冗余性和高维性问题,提出了一种将样本优化和特征优化相结合的监测数据优化算法。首先,采用特征空间样本选择算法对监测数据进行样本优化,找出最具代表性的样本;然后,采用核主成分分析—分布估计算法(KPCA-EDA)对样本优化后的监测数据进行特征优化,在保证特征信息充足的情况下,保留更多的识别信息;最后,以某滤波电路为例进行了验证,仿真结果表明,该算法同KPCA等优化算法相比,在训练时间和识别率上能达到更好的平衡。

Threshold decision for instrument control system and its application in nuclear power plant

Threshold decision is an important function of nuclear instrument control system based on physical parameters threshold decision. Because the conventional decision methods lack correlation with time and conditions, by analyzing the existing methods, some optimized methods are adopted. Considering safety, those methods are improved in data processing algorithms, floating threshold with multiple values, association with specific working condition, etc. These measures im- prove the nuclear instrument control system in fault tolerance and fault diagnosis, especially, the shutdown number of nucle- ar power plant decreases.