关键词: 农药检测, 红外光谱, 基于变异的紧凑遗传算法, 变量选择

Abstract:

In order to simplify the forecasting model of pesticide detection,and to improve the prediction accuracy of the detection model,this study uses the infrared spectroscopy technology and the mutation-based compact genetic algorithm (mCGA) to achieve the wavelength variable selection,and to a certain extent,reduce the non-information variables and disturbance variables.Wavelength variables are selected to build prediction model with different algorithms.The root mean square average of the prediction obtained is 0.198 by using mCGA,smaller than those by the compact genetic algorithm (CGA) and by the simple genetic algorithm (SGA) which are 0.241 and 0.289,respectively.The results show that the variable selection by means of mCGA can improve the convergence rate and accuracy of the model,and enable the detection of pesticide quickly and efficiently.

Key words: pesticide detection;infrared spectroscopy;mCGA;variable selection