Study The Spectroscopy And Grain Traits Quickly Identify Cotton Variety Purity And Authenticity

Many reasons caused seed market chaos and seed quality problems, bring serious economic losses to the production of cotton. In order to improve the quality of cotton, maintain the order of seed market, reduce disputes and economic losses, safeguard the vital interests of farmers, the cotton seed purity identification is necessary. Cotton varieties purity and authenticity test results is the seed market supervision, quality control, and an important basis for variety protection. Survey:China each year there will be more than a dozen varieties of cotton。 However, the purity and authenticity of cotton varieties detection technology is still stuck in the field planting through morphological school identification stage, this method identified a longer period, with the rapid development of molecular techniques in recent years, the use of molecular markers for rapid identification of cotton seed purity has been applied in research institutes and other units, because of technical difficulties and the higher testing costs,It’s not been widely used in the production.In this study, spectroscopic techniques and varieties of cotton seed traits rapid identification of purity and authenticity. The main methods and approaches are:(1) Selection of77parts cottonseed samples, of which72samples were selected as calibration set, the remaining five samples as a validation set. Divided into two samples were collected for determination of the chemical composition and near-infrared spectral data, According to the data measured by chemical methods,established moisture, protein, crude fat NIR analysis mode respectively,In order to explore the feasibility of cotton seed moisture, protein and fat content by spectroscopy analysis and forecasting;(2) Choosed five different samples to validate the model, by comparing the difference value between predicted and measured by chemical, prooffing of the accuracy of the best model prediction;(3) Explore by principal component analysis and BP artificial network identification technology to build model of cotton varieties to distinguish the feasibility of cotton varieties;(4)Explore the feasibility of distinguishing characteristics of cotton seed traits through cotton hybrids. The results showed as follows:Choosed77cottonseed samples, using chemical methods determined the moisture, protein, crude fat content. Using72samples data to establish a near infrared spectroscopy calibration model..The models determination RSQ were0.998,0.992,0.992; the SEC were:0.081、0.169、0.235. This model can be used to predict cottonseed moisture, protein and crude fat content;(2) Using this model, five cotton varieties moisture, protein, crude fat content were predicted, and compared with chemical measurements, the two methods were compared, the results showed that the conduct five cotton samples tested, Moisture content of the absolute error is not more than0.19, the relative error is1.70%; Protein content of the absolute error is not more than0.62, the relative error is1.50%; Crude fat content of the maximum absolute error less than0.32, the maximum relative error of1.52%. The results showed that NIR calibration models can be quantitative analysis was carried out on the purity and authenticity of cotton varieties.(3) Validation set five cotton varieties were randomly selected25samples which collected125samples of spectra,the analysis suggests that the first seven principle components (PCs) can account for99.696%of the original spectral information, and it means that the seven PCs can explain most variation of original variables. In order to set up the model for discriminating varieties of paddy, the seven diagnostic PCs are applied as inputs of back propagation artificial neural network (BP-ANN), and the values of varieties of different paddy are applied as the outputs of BP-ANN. The BP-ANN is trained, the optimal three-layer BP-ANN model with7nodes in input layer,11nodes in hidden layer and1node in output layer would be obtained, and the transfer function of sigmoid is used in each layer. The results indicate that a100%recognition ration is achieved with the threshold predictive error±0.1. It proves that the model is very reliable and practicable.(4) The hybrid cotton and their parents, there are differences in thickness, length and weight, the effect of stripping bud can make the grain seeds become more short, in which the degree of influence are:valve head> lobe> flap tail, in color performance, stripping Lei performance between the different effects of different flap position, color differences:lobe> flap tail> valve head.The traits of grain can be detected cotton varieties purity and authenticity.In conclusion, the application of near infrared spectral analysis technology to establish calibration model, can achieve quantitative analysis of chemical components in different cotton varieties of cottonseed moisture, protein, crude fat content; Varieties of BP artificial neural network identification model, can achieve the qualitative analysis of different cotton varieties; Preliminary study based on grain traits, due to the influence of the strip’s effect on the behavior of hybrid grain, can through the grain traits to identify cotton varieties purity and authenticity. Thus, combine spectroscopy and grain traits, can be widely used in the production of rapid detection of cotton varieties purity and authenticity.