旋转机械状态参数图形识别的免疫-模糊形态学方法

State parameter image identification method based on fuzzy mathematical morphology and immune intelligence for rotating machinery

以旋转机械振动多维图形为对象,研究了直接提取和挖掘图形特征信息的模糊形态学方法,提出了基于模糊数学形态学及免疫智能的旋转机械振动参数图形识别方法.利用模糊形态滤波方法实现图形滤波,研究了模糊形态边缘检测算子,并结合旋转机械振动参数图形进行形态学梯度的边缘纹理特征提取,最后利用人工免疫算法对图形特征进行诊断识别.在600 MW模化汽轮机转子试验台上进行了转子正常、转子不平衡故障、转子不对中故障及汽流激振故障的试验,诊断结果表明所提出的方法可以获得较高的诊断精度.

Taking the multi-dimensional image information of rotating machinery vibration as the research object, this paper investigated the fuzzy morphology method of directly extracting and mining the texture feature in vibration parameter image of rotating machinery, and then presented a state parameter image identification method of rotating machinery based on fuzzy mathematical morphology and immune intelligence. Using the method of fuzzy mathematical morphology filter to implement image filter, this paper studied fuzzy morphology edge detection operator and extracted the edge texture feature based on the morphology gradient from the vibration parameter image of rotating machinery, and finally diagnosed and identified the image feature using the artificial immune algorithm. On the modeling of 600MW turbine rotor experimental bench, the rotor normal state, unbalanced fault, misalignment fault and steam exciting vibration were tested. The diagnosis result shows that the method can accurately obtain image feature and improve the diagnosis accuracy.