A fault diagnosis method based on graphic recognition for rotating machinery

Taking a three-dimensional vibration image of rotating machinery as a studying object, the fault diagnosis method based on graphic recognition technology was investigated. This method used the statistical method, the structural method based on texture and the gradient method characterizing the graphic textural direction to form a matrix describing graphic textural characteristics. The matrix rough level, direction and spatial complex level and direction of texture. The graphic grayscale spatial distribution characteristic, spatial statistical dependence and pixel point gradient distribution rule were described accurately. Textural feature information in rotating mechanery state parameter graph was extracted effectively. Rotating machinery fault diagnosis could be conducted by using artificial neural networks after extracting the information of image texture characteristic. This method was validated to have higher diagnosis accuracy with 6 tests on a test table of a 600MW turbine.