传感器多故障诊断的信息融合方法研究.pdfVIP

27 卷 16 期 中 国 电 机 工 程 学 报 Vol.27 No.16 Jun. 2007 2007 年6 月 Proceedings of the CSEE 2007 Chin.Soc.for Elec.Eng. 文章编号：0258-8013 (2007) 16-0104-05 中图分类号：TP206 文献标识码：A 学科分类号：47040 传感器多故障诊断的信息融合方法研究 冀，王兵树，邸 剑，于 浩，鲁 斌 (华北电力大学计算机系，河北省 保定市 071003) Research on Information Fusion for Sensors Multiple Fault Diagnosis ZHANG Ji, WANG Bing-shu, DI Jian, YU Hao, Lu Bin (North China Electric Power University, Baoding 071003, Hebei Province, China) ABSTRACT: A novel scheme for multiple fault diagnosis of 行。因此，对传感器的状态检测和故障诊断尤为重 multisensor under different frame of discernment based on [1-2] 要 。 evidence theory is proposed. The redundant or complementary 由于生产设备的复杂性、传感器的可靠性和运 information of multisensor in space or time is used to detect and isolate the fault sensors. The evidence of sensor state is 行环境的不稳定性，单个传感器获得的信息具有局 obtained through a high modularity of the sensor state 部性和不确定性。根据信息论的原理，由单维信息 recognizer network with modularized RBF, so the advantage of 融合起来的多维信息，其信息含量比任何一个单维 neural net is promoted and the bug of its unusable for change of 的或几个单维信息简单相加的信息量都要大的多。 input parameter is resolved. The evidential combination rule 多传感器信息融合充分利用多个传感器资源， under different frame of discernment is presented, and the 通过对各种传感器及其观测信息的合理支配与使 operation of refinement and coarsening is adopted, so the maximum of available information is exploited and the 用，以各种传感器在空间和时间上的互补与冗余信 uncertainty of the effective state between the potential states of [3-4] 息可为故障的检测和分离提供诊