Research on the Non-destructive Testing Method of Pod Based on Neural Networks

The non-destructive testing method of pod was proposed based on neural networks according to pod´s vibration signal. The wavelet theory is used to reconstruct chrysalis vibration signals. The characteristic values among chrysalis vibration signal, which were related to chrysalis weights, were extracted respectively. And the characteristic values were selected firstly by fuzzy classification. Then BP model for NDT of silkworm is built, which inputs include seven relevant characteristic values. The output of neural network represents weights of the chrysalis. The weight of cocoon shells can be gain indirectly. The testing result shows that the method is effective and feasible