Effect Of Flow On Dielectric Properties And Quality Prediction Of Raw Milk

The heterogeneity of milk affects the quality detection accuracy of dielectric spectroscopy.To address the problems of expensive milk composition analyzers due to high-pressure homogenization and unclear influence of milk flow rate in milking tubes on dielectric properties,a benchtop instrument-oriented method is proposed to assist in the measurement of dielectric spectra of non-homogeneous sample under the dynamics.For the first time,we investigate the effect of milk flow rate in milking pipes on its dielectric properties,and constructe a new method for predicting the main components of milk in milking pipes through model transfer.The main research and findings of this paper are as follows:(1)Exploration of a new method to improve the repeatability of dielectric properties-Feasibility analysis of replacing homogenization with sample flow dynamics measurement.The dielectric spectra of raw milk were analyzed at 23°C and 40°C under the conditions of static and oscillating,and the treatments of centrifugation and different homogenization pressure(30/5,60/5,90/5 MPa).The coefficient of variation was used as an indicator to analyze the repeatability of dielectric spectra of milk,and the variation of particle size distribution of milk samples with time was also compared.The results showed that the repeatability of the dielectric spectra of cow’s milk in the standing state was poor.Repeatability of dielectric spectra for oscillatory state measurements closed to 90/5 MPa homogeneous treatment.The mechanism could be explained by the results of centrifugal treatment,i.e.,oscillation disrupted the floating and aggregation of fat globule.Oscillatory state(23°C better than 40°C)in measurement are feasible instead of homogenization treatment.(2)Analysis of the dielectric properties of milk based on desktop instrument and its measurement repeatability with measurement in the flow state.The effects of milk measurement under flowing state(sonication,rotation and oscillation),standing state,and homogenization treatment on its dielectric spectra were compared and analyzed at 23°C.Significance analysis was performed on the coefficient of variation of the measurement of dielectric spectra at three specific frequencies 40.68,915 and 2450 MHz for different measurement conditions.The results showed that all three flowing measurements improved the values of dielectric parameters and measurement repeatability.of milk.The measurement repeatability of rotational and oscillatory measurements was close to that of the 90/5 MPa homogenization treatment.(3)Effect of milk flow rate on the repeatability of dielectric spectra.A pipeline flowing measurement equipment was built,the repeatability of the dielectric spectra in deionized water,tap water,raw milk and homogenized milk at different flow rates(0-3 m/s)were compared and analyzed.At three special frequencies 40.68,915 and 2450 MHz,significance analysis of the coefficient of variation of dielectric spectra at different flow rates of milk was performed.The results showed that the flow rate improved the repeatability of the dielectric spectra of non-homogeneous milk,and appropriate flow rate was helpful to prevent the floating and aggregation of fat globules.When the flow rate of milk in pipeline reached 0.1 m/s,the measurement repeatability of its dielectric spectra was no longer significantly improved.(4)Effect of flow rate on the dielectric properties of milk.The values of dielectric parameters were measured for different flow rates(0-3 m/s)of milk.Significance analysis was performed on the values of dielectric parameters at different flow rates at three special frequencies 40.68,915 and 2450 MHz.A modified analytical model of dielectric relaxation parameters was built based on Debye’s equation,and the relaxation intensity,relaxation time and conductivity of the milk at different flow rates were fitted.The results showed that theε’andε"of milk increased with the increase of flow rate.When the flow rate of milk reached 0.1 m/s,its dielectric parameters no longer improved significantly.As the flow rate increased,the relaxation intensity and conductivity gradually increased,while the relaxation frequency decreased.The modified Debye equation can be used to reveal the influence mechanism of flow rate on the dielectric properties of milk.(5)Quality prediction of milk in flow state based on benchtop instrument.At homogenization 30/5 MPa,ultrasonic frequency 37.5 k Hz,rotational speed 250 rpm and oscillation speed 350 rpm,the prediction models were developed for the main components of milk under flow state(ultrasound,rotation and oscillation),standing state and homogenization treatment,respectively.In the modeling process,the prediction accuracy of Savitzky-Golay(S-G)and standard normal variables transformation(SNV)preprocessing,full spectra and uninformative variable elimination(UVE)variable extraction methods,and partial least squares regression(PLS)and least squares support vector machine(LSSVM)models were compared and analyzed.The results showed that measurements in flowing state(rotation and oscillation)effectively improved the prediction accuracy of main milk components.The highest predicted RPD for milk components,i.e.fat,protein,lactose and water content,were 2.498,2.114,2.134 and 2.529,and the minimum RMSEP were 0.194%,0.119%,0.050%and 0.203%,respectively.The prediction effect is much better than the static measurement and homogenization treatment.(6)Research on quality prediction of flowing milk in pipeline based on model transfer.A new method for predicting the main components of flowing milk in pipeline was established,with the optimal models for the static,rotating and oscillating states of milk established by the benchtop instrument,combined with three model transfer methods,i.e.,slope/bias(S/B),direct standardization(DS),and piecewise direct standardization(PDS).The results showed that the r_p and RMSEP for fat content prediction based on the model developed in the rotating state using optimal PDS were 0.680 and 0.501%,respectively.The model developed in the static state of milk achieved optimal prediction of protein,lactose and moisture content of pipeline flowing milk using the model transfer method,with the r_pof 0.656,0.363 and 0.609,and the RMSEP of 0.400%,0.067%and 1.079%,respectively.