摘要
针对Fuzzy ART神经网络在多工况设备状态监测中存在的学习时间过长问题,提出一种基于粗集约简的RS-Fuzzy ART状态监测方法.该方法利用粗集的信息决策表和决策矩阵对设备的监测参数进行约简提取,降低Fuzzy ART输入向量的维数.监测实例结果表明,采用粗集约简提取的监测参数与原监测参数具有相同的监测能力,且可极大缩短网络的学习时间.若将约简提取的监测参数进行联合监测,还可间接消除由于传感器故障或信号传输错误引起的误报.
Multiple work state monitoring method based on RS - Fuzzy ART neural network is presented for complexity equipment system. The method uses rough information decisionmaking form and decision - making matrix. To reduce choose monitoring parameters of the system, thus can effectively reduce dimension of FuzzyART input vectors, improve study efficiency of network. The example test result show that it has same monitoring capability either using monitoring vectors of rough reduction distilling or former monitoring vectors, if united monitor the monitoring parameters of reduction distilling, it can diagnose indirectly partial sensor failure or signal transmit error, decrease misinformation.
出处
《哈尔滨商业大学学报（自然科学版）》
CAS
2008年第6期733-737,共5页
Journal of Harbin University of Commerce：Natural Sciences Edition
基金
黑龙江省教育厅科学技术研究项目(11511099)
黑龙江省自然科学基金项目(E200615)资助
