An expert system based on BP neural networks for porcelain quality identification

In order to classify porcelain product,; automatically on tine, an expert system based on BP neural networks has been studied and developed, Considering how to optimize the training samples and to simply the structure of neural networks by the method of Fast Fourier Transform (FFT) which is used to convert signals in time domain into frequency domain and by expert list to collect the representative characters from the parameters. Efforts is concentrated on developing intellective unit to imitate human being listening the sound of porcelain vibration and analyzing the signals, therefore BP neural networks is employed to identify statuses of porcelain product quality and an expert system is used to determine the character parameters and to give the report.