Application Of Steel Balls' Surface Crack Detection Based On Digital Signal Processing And Neural Net

At present, the detection to steel balls' surface crack, still using the traditional manual method of detection. With the digital signal processing and neural network technology widely used and gradually increase the economic efficiency requirements, the need for a detection system to evaluate surface quality of the steel balls correctly and efficiently.Using wavelet analysis techniques can effectively remove the steel balls' surface crack detection of noise and interference, and improve the signal to noise ratio, enhance the clarity of the signal, making signal denoising better. Artificial neural networks studied on the importation of samples to find the nonlinear relationship implicit in the network phase of the study, in the learning phase of the network, the neural network approximation is better than traditional math model, provided a guarantee for the detection.The paper using the wavelet transformation and the neural network technology implement the steel balls' crack detection. Using the wavelet technology for signal denoising processing, After network training, automatic decision-making a needed regional to achieve the balls' crack detection. In-depth study of the steel balls' surface crack detection system, and the system combined with the acquisition signals which carried out a large number of experiments to verify the system is feasibility in the signal processing. The system processes including wavelet transform, threshold selection and use of neural network realized balls' surface crack detection eventually.