小波包能熵谱和证据融合推理的电梯急停诊断

Elevator Fault Stop Diagnosis Based on Wavelet Packet Energy Spectrum and Evidences Fusion Reasoning

采用加速度时域信号的峭度系数和小波包变换后的频域信号能熵谱定义特征向量，获得证据融合法推理证据；利用测试样本统计得到特征值区间数；基于待测样本特征值与区间数之间的距离设计相似度，并归一化得到证据的基本概率分配函数；而后对证据进行时空融合得到证据融合合成规则，并采用可信度分配函数进行证据融合推理诊断；最后开展电梯实测实验，利用3层sym8小波包变换和9个证据对急停进行诊断。实验表明，该方法电梯急停故障诊断的准确率大于98％。

The feature vector will be got by using the kurtosis coefficient of time domain signal and the entropy spectral of the frequency domain signal after wavelet packet transformation, and this is defined as DS evidence. The eigenvalues interval through sample statistics will be obtained. The similarity based on the eigenvalues and the length between the intervals of the test sample was designed, and the basic probability assignment function by normalization will be obtained. Then the DS combination rule by spatio-temporal combination of the evidence was got. And the DS diagnosis results by using reliability allocation function can be obtained. Finally, the experiments will be carried out and diagnosis results by using sym8 for three-layer wavelet packet transformation and 9 evidences are received. The experimental results show that the diagnostic accuracy of this method about elevator fault stop is more than 98%.