[A method of recognizing biology surface spectrum using cascade-connection artificial neural nets].

A method of recognizing the visible spectrum of micro-areas on the biological surface with cascade-connection artificial neural nets is presented in the present paper. The visible spectra of spots on apples' pericarp, ranging from 500 to 730 nm, were obtained with a fiber-probe spectrometer, and a new spectrum recognition system consisting of three-level cascade-connection neural nets was set up. The experiments show that the spectra of rotten, scar and bumped spot on an apple's pericarp can be recognized by the spectrum recognition system, and the recognition accuracy is higher than 85% even when noise level is 15%. The new recognition system overcomes the disadvantages of poor accuracy and poor anti-noise with the traditional system based on single cascade neural nets. Finally, a new method of expression of recognition results was proved. The method is based on the conception of degree of membership in fuzzing mathematics, and through it the recognition results can be expressed exactly and objectively.

[Study on the method to adjust optical fiber's spectrum attenuation with the Fourier transform].

A method based on Fourier transform to compensate the non-linear attenuation of optical fiber used as a probe in a spectrum-collecting system was proposed. First the output electric currents of photoelectric tube with and without fiber were transformed to the frequency field. So an adjustable function in frequency field was obtained, and the adjustable function was transformed to the spectrum field, so the final adjustable function was obtained. A photoelectric system was designed for testing. With visible light, this method can make the error rate of fiber transmission as low as less than 1.54%. It is proved that the method is fit for adjusting some optical fiber spectrum attenuation.

[Improving partial least square regression precision in NIR multi-component analysis using artificial neural network].

The present paper presents a new NIR multi-component analysis method with Artificial Neural Network(ANN) and Partial Least Square Regression(PLS). First, this method divides the concentration range of training samples into some sub-ranges, and respectively computes a PLS correlation model in each sub-range with the sub-range's training samples. Then, the authors classify prediction samples according to its concentration sub-range with ANN and judge which sub-range theprediction sample belongs to. Finally, the authors compute the concentration of prediction component with the PLS correlation model of the sub-range according to ANN. The experiment and the result of data processing show that this method improves the model's applicability, and evidently enhances prediction precision compared to traditional PLS.