Neural network-based cyperus esculentus oil content near-infrared analysis model and characteristic wavelength extraction method

The invention relates to a near-infrared model establishment method and a characteristic wavelength extraction method for predicting the oil content of cyperus esculentus, and belongs to the field of near-infrared spectrum analysis. According to the method, the MLP neural network is combined with PLS cross validation, the characteristic wavelength related to the oil content in the near infrared spectrum of the cyperus esculentus is extracted, the screened characteristic near infrared information related to the oil content is used for fitting with the oil content, the accuracy of the obtained near infrared model is better, the predictive capacity is greatly improved, and meanwhile the number of the characteristic wavelength is smaller. By utilizing the established near-infrared analysis model, the oil content of the cyperus esculentus can be predicted only by measuring the near-infrared spectrum information of the cyperus esculentus, and rapid, nondestructive and accurate measurement is realized. 本发明涉及一种预测油莎豆含