Fast Detection Of Organic Phosphorus Pesticide Residues In Vegetable Plants Based On Colorimetric Spectroscopy

Fast detection of organophosphorus pesticides is important to ensure the quality safety of agricultural products. The research on typical organophosphorus pesticides detection in vegetable plants was carried out using colorimetric spectroscopy technology to provid a theoretical basis and technical support for the development of rapid pesticide detection instruments. The contents of the disertation were briefly summarized as follows: [1] Parameter optimization of colorimetric reagents for typical organophosphorus pesticides based on colorimetric methodOpalladium chloride colorimetric parameters were further improved and optimized through lots of experiments, acetic acid and sodium chloride was used respectively instead of traditional concentrated hydrochloric acid as chloride palladium solvent. Organophosphorus pesticides containing sulfur such as Dimethoate, Chlorpyrifos, Omethoate and Acephate were selected as the colorimetric reaction research models, and the absorption spectra of the pesticides showed that the range of 0.05～0.5 mg/kg was the minimum concentration of the pesticides that could be distinguished in the spectra. The results met or basically met the pesticide residues detecting requirements of part fruits and vegetables in the national standard GB2763-2014 regulations. The colorimetric reaction time was 2 min at room temperature, which satisfied the requirement of fast, safe and easy detection. [2] Prediction model of organophosphorus containing sulfurThe single linear regression models at maximum absorption wavelength of 4 kinds of pesticides extracted in colorimetric absorption spectra were established, but the predictive accuracy was generally low because the colorimetric spectra were influenced by many factors, and it was difficult to eliminate completely the experimental errors. Principle component analysis (PCA) method, partial least squares method (PLS), BP/RBF neural network method were selected to establish the pesticide prediction model. The comparing results showed PLS model was better than others. An example of dimethoate, the optimum model was obtained in the region of 400～600 nm, when principal component number was 4, the correlation coefficient of calibration and validation were 0.9941 and 0.9933 respectively, and the RMSEP of calibration and validation were 2.7703 and 2.2148 respectively, which indicated the feasibility of detecting trace pesticide residues with visible light waveband. [3] The multiple prediction models of concentration sections for improving the adaption abilityIt is difficult to determine the right amount of colorimetric reagent for the target pesticide because the pesticide content cannot be predicted in real samples. The colorimetric spectra was changed greatly when the addition of colorimetric reagent was excessive or insufficient, and the prediction model became more complicated. Therefore, the method of multi-model of concentration sections was studied and after comparison of PLS, PCA+BP/RBF neural network, SVM+PSO, SVM, PLS method were selected to establish the qualitative and quantitative prediction model, respectively. The results showed that the prediction models for large concentration range could improve the adaption ability. [4] Research on classification model of pesticide species based on PSO-SVM methodClassification model of pesticide species was established according to different colorimetric spectral features of pesticides. PCA+BP/RBF neural network, SVM+CV, and SVM+PSO method were studied, and the PSO+SVM method achieved best results. In the classification model of four kinds of pesticides, when optimized parameters c=2.6857, g=0.01, accuracy of training set and validation set reached 100% and 95% respectively. A new feasible method of fast classifying different kinds of pesticides was provided. [5] Effect analysis and elimination method of vegetable substrate on detection of organophosphorus pesticidesThe difference of colorimetric spectra between organophosphorus pesticide extracted in vegetable samples and the pure samples was studied and compared in absorbance and spectral line trend. The conclusion after an analysis was drawn that different vegetables could influence the colorimetric spectra of different organophosphorus pesticides in some wavebands, and the discriminative power of the spectra decreased significantly in the influenced wavebands. Therefore, the pesticide prediction model under vegetables substrates was built by differential spectrophotometry, which improved prediction accuracy and proved the validity of the model.The colorimetric reagent of was optimized based on lots of experiments. The prediction model of sulfur-containing organophosphorus pesticide residuces was established to improve the adaption ability. The effect of common vegetables on organophosphorus pesticide detection was analysed and elimination method was summeried. The feasible prediction model was obtained ultimately.