摘要
从分形理论和神经网络原理出发 ,基于信号分形计算维数定义 ,以采样周期和分形计算维数两个坐标对信号特征进行合理地模糊化处理 ,提出分形维数隶属度特征量概念·以多时间尺度采样所获得的分形维数隶属度作为网络输入 ,单位矩阵为网络输出的分形模糊神经网络 ,建立了时域精确诊断新方法 ,通过对典型齿轮系统故障进行精确诊断 ,结果表明分形模糊神经网络诊断方法的有效性·
Based on fractal principle and neural network theory,a new concept of signal computing dimension and fractal dimension subordinating degree was proposed. In both sampling period and fractal computing dimension,the signal is fuzzed to construct fractal dimension subordinating degree. A novel correct method,the fuzzy neural network diagnosis method,was proposed,in which the input is the fractal dimension subordinating degree, and the output is the unit matrix. A gearing system was diagnosed by using the modeling.
出处
《东北大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2002年第2期195-198,共4页
Journal of Northeastern University(Natural Science)
基金
辽宁省博士起动基金资助项目 (2 0 0 110 2 0 17) .
