Study On Near Infrared Spectroscopy Detection Method Of Honey Quality

As a natural product,honey has a high nutritional value.With the continuous improvement of people s quality of life,the market demand for health care products increases year by year.As a best-selling commodity in the health care products market,honey has a serious imbalance bet,ween supply and demand in the sales market,resulting in uneven quality of the honey market and even the phenomenon of blending.There are currently no instruments on the market that can quickly detect the quality of honey.Therefore,this topic studies the rapid detection method of honey quality,analyzes the prediction accuracy of honey quality in different spectral intervals,and proposes a small portable honey quality near-infrared spectroscopy detection scheme.The paper mainly completed the following research work:First,the principle of near-infrared spectroscopy absorption is introduced to determine the position of the near-infrared absorption peak of carbohydrates and water in honey.According to the physical properties of honey,the collection method of near-infrared spectrum is determined.The commonly used three chemical model identification methods were compared,and the partial least squares regression method was selected to establish a honey quality detection model.Secondly,the near-infrared spectra of pure honey,water,fructose saturated solution and glucose saturated solution were analyzed and compared.71 sets of different types of mixed honey and 6 sets of pure honey samples were prepared.The sample spectrum was divided into six groups of characteristic bands,and different pretreatment methods were used to establish a model for detecting glucose,water,sucrose and fructose in honey.The modeling results of the six characteristic bands were analyzed,Based on the two evaluation indexes of the detection model,the prediction accuracy of fructose and moisture was found to be high,and the cross-validation coefficient(R)could reach 98.34%and 99.42%,respectively.Verify the calibration standard deviation(RMSECV)are 0.782 and 1.01 respectively.The reliability of the six test models was verified.20 sets of honey samples were prepared,and the spectral information of the test samples was subsituted into the model to obtain the predicted values of the four chemical contents of the samples.The predicted value and the true value of the four components of the sample were analyzed.It was found that when the doping ratio of the tested honey was the same as the doping ratio of the modeled honey sample,the relative error between the predicted and actual values of the four components of the sample was small.According to the model prediction results and the design requirements,the 4850-5120cm-1,5700-6200cm-1,and 6500-7100cm-1 bands were selected for combined modeling.The established model prediction accuracy is higher than the six bands independently modeled.The modeling intervals are 4880-5063cm-1,5800-6070cm-1 and 6762-7000cm-1.The four-component predictive evaluation index R is above 95%,and the RMSECV is below 1.84.Therefore,in the bands of 4880-5063cm-1,5800-6070cm-1 and 6762-7000cm-1,a selection interval is provided for the wavelength of the light source of the near-infrared detection system.Finally,according to the three-band combined modeling results combined with thecharacteristic absorption wavelength of the four components of honey,the four source wavelengths of honey quality detection were determined to be 1450nm,1687nm,1869nm and 2010nm,respectively.The honey-quality near-infrared spectroscopy detection system was designed.The optical simulation of the optical path of the detection system and the signal acquisition circuit design were completed,and the feasibility of the honey quality detection scheme was determined.