Research On Methods For Simultaneous Detection Of Cu, Zn And Pb In Edible Oils Using Electroanalytical Chemistry And Chemometrics

Heavy metals in edible oils including toxic and essential metals are an importantdetection index, and its content has an important influence on the quality. Nowadays,the reports on simultaneous detection of multi-component heavy metals byelectrochemical method in edible oil are few, however, due to its advantages of simpleequipment, low detection limit, high sensitivity, and low detection cost, the technologywas employed to study on detection method of multi-component heavy metals inedible oil with the aid of chemometrics. At the same time, a new detection method ofmulti-component heavy metals was also provided.Differential pulse stripping voltammetry was applied to detect the concentrationof Cu, Pb and Zn in their mixture solution samples. The main works are as follows inthis paper:1. The best condition for electrochemical measurements was obtained by the in-depth experimentation, such as acetic acid-sodium acetate buffer supporting solution,whose concentrations and pH respectively is0.1mol/L and4.5, and detectionparameters of deposition time, deposition potential and scan increments are60s,-1.2Vand10mV/s, respectively.2. In order to improve the signal-to-noise ratio, the stripping signal wasrespectively dealt with smoothing, smoothing-derivation, kalman filtering and three-layer wavelet packet analysis. To some extent, these methods could reduce detectionerrors, while the results based on smooth derivation and wavelet packet were better.What’s more, the smooth-derivation formula was perfected in the denoising.3. To build a reliable and robust detection model, the regression predictionmodels of Cu, Zn and Pb corresponding the four denoising methods were respectivelyestablished by PCR, MLR, PLS, BP-ANN, LS-SVM. Comparing their prediction values of forecasting samples, the detecting accuracy of PCR, MLR, BP-ANN and LS-SVM models based on smoothing-derivation was higher; the PLS model based onwavelet packet analysis had also a better result.4. The above five models were used to detect the content of Cu, Zn and Pb inedible oil samples, the test results showed the prediction results of BP-ANN regressionmodel based on smoothing-derivation are better than other methods, and its results hadno significant difference with flame atomic absorption spectrophotometer.The paper demonstrates that the electrochemical technology integratedappropriate signal denosing and modeling method can be competent for thesimultaneous determination of Cu, Pb and Zn in edible oil.