Research Of Fault Diagnosis Method Of Steel Cord Conveyor Belt Based On Wavelet Singularity And Neural Network

Steel cord conveyor belt for its high tensile strength, little elongation, dynamic performance, long transport distance has been widely used, especially in the coal mining industry as the main equipment of coal continuous conveyor. The transverse fracture accident of steel cord conveyor belt happened sometimes, that will cause significant economic losses and casualties to the corporates. At present, the common detection methods can effectively find the macroscopic geometry type defects of conveyor belt, but the early physical injury of steel cord is difficult to achieve an effective evaluation. This paper has described the development of steel cord conveyor belt fault detection methods in the world, studied the principle of the structure of steel cord conveyor belts and metal magnetic memory testing technology, analysised of the singularity of magnetic memory signal of steel cord, and proposed combination of the normal component crossing zero point and maximum gradient value and singularity of magnetic memory signal to judge the stress concentration area of conveyor belt joint and accurately located the failure point; At last, this paper has solved the identification problem of three states of the conveyor belt steel cord:normal state, the microscopic stress concentration state and macro-defect state, some characteristic quantities of the magnetic memory signal of steel cord was extracted, such as:peak-to-peak of magnetic memory signal, maximum gradient value of magnetic memory signal, the maximum amplitude of first scale detail component with wavelet multiscale decomposition of magnetic memory signal, the diameter of steel cord, the energy of continuous wavelet transform. These characteristic quantities would input to BP neural network designed for data fusion, can synthetically determine the state and type of fault of steel cord conveyor belt. This paper has provided a novel, effective and accurate method for early fault diagnosis of steel cord conveyor belt.In this paper, steel cord conveyor belt was chosen as the research object, metal magnetic memory testing technology was used to detect the steel cord conveyor belt, and the magnetic memory signal of different states of the steel cord could be detected. Firstly, the singularity of magnetic memory signal of steel cord had been analysed, and combined with the normal component crossing zero point and the maximum gradient value of the magnetic memory signal, could accurately judged the failure point of stress concentration area of steel cord conveyor belt. Secondly, with the multi-resolution characteristics and time frequency localization properties of the wavelet transform, the multiple features of magnetic memory signal of steel cord conveyor belt (such as the absolute peak, peak-to-peak, maximum gradient value, the maximum amplitude of first scale detail components with wavelet decomposition, and energy of continuous wavelet transform) would be studied and extracted. Finally, by means of magnetic memory signals obtained on the steel cord conveyor belt fault detection platform, multiple features value of magnetic memory signal extracted formed the vector group, as the input of BP neural network, using MATLAB programming simulation to identificate the state of steel cord conveyor belt. The results showed that the method could meet the requirements of steel cord conveyor belt failure mode diagnosis to a certain extent, and met the requirements of real-time fault diagnosis, it had a strong ability to identify.