Concentration Measurement For Food Aqueous Solution Based On Ultrasonic Signal Transformation

Continuous process monitoring is a fundamental requirement for process control in food industry. Besides process parameters, such as temperature, pressure, liquid level, flow rate and so on, concentration measurements are also of special interest. More qualified information obtained form new or better sensors can significantly enhance the process quality and thereby product properties. Low-intensity ultrasonic sensor systems can contribute to this development, and this measuring technique has been widely used in the fields of non-destructive detection and medical image because of its non-destructive, non-invasive and rapid detection.In this research, low-intensity impulse ultrasound measuring technique was applied to determine the concentrations of solutions. Under different situations, statistical models were built for measuring concentrations with different digital signal processing methods. The specified contributions and results are listed as follows:(1) An ultrasonic liquid analyzer system was set up in our laboratory. This measuring system is composed of an ultrasonic pulser-receiver, transducers, a data acquisition card, measuring cells and software. The ultrasonic pulser-receiver, CTS-8077PR, is made by China Circuit Technology (Shantou) Corporation. In our research, impulses with amplitudes of-25V and widths of100ns were used. Transducers were also bought from the same corporation, with a center frequency of5MHz. Signal was detected using a16-bit data acquisition card PCI-9846H/512produced by ADLINK Technology Inc., Taiwan. It was triggered by rising edge of external digital impulse, with a sample rate of40MS/s. Measuring cells were made by ourselves. Different cells with different wall thickness and different materials were used depending on specific purposes. A friendly user interface was also developed to acquire and process data in LAB VIEW2010.(2) A concentration model of NaCl water solution was established with OMLR, PLS, iPLS and siPLS regression technologies.The temperature, sound velocity, frequency information, continuous wavelet coefficients and wavelet decomposition coefficients of ultrasound signal were utilized in models. The results show that the best model for NaCl solution concentration was developed by synergy interval partial least squares regression (siPLSR) together with frequency information. The determination coefficient of calibration Rcal is0.9999, the determination coefficient of validation Rcal2is0.9988, the root mean square error of cross calibration (RMSECV) is0.0526g/100g, and the root mean square error of validation(RMSEP) is0.0628g/100g, while the RMSEP of OMLR model was0.1015g/100g. The best model based on temperature, ultrasound velocity, frequency information and siPLSR is stabilization, accuracy, unmcomplicated, and is suitable for predicating the concentration on-line non-invasively and rapidly. At the same time, the concentration of citrus pectin solution was also measured. The RPD was14.2, which is obtained by the model developed by spectral information combined with PLS. It illustrates that this method is more efficient than OMLR model developed by sound velocity. (3) Regression models were developed for measuring solute concentrations in a sucrose-ethanol-water triple solution with ulthrsound velocity at2℃and30℃. The calibration determination coefficient Rcal2for measuring sucrose concentration was0.9793, validation coefficient Rval20.9475, RMSEP0.4544g/100g and RPD4.92. The calibration determination coefficient Rcal2for measuring ethanol concentration was0.9876, validation coefficient Rval20.9853, RMSEP0.1741g/100g and RPD8.29. Meanwhile, PLSR models were developed by using spectrum information of ultrasound signal. The calibration determination coefficient Rcal2for measuring sucrose concentration was0.9852, validation coefficient Rval20.9842, RMSEP0.4270g/100g and RPD5.25. The calibration determination coefficient Rcal2for measuring ethanol concentration was0.9952, validation coefficient Rval20.946, RMSEP0.1549g/100g and RPD9.32. So the models devepoled by PLSR were more accurate than models developed by sound velocity.(4), An ultrasonic system for measuring characteristic acoustic impedance of liquids was developed, using multiple reflected echoes caused by the difference impedance of the solid-liquid interface. The characteristic acoustic impedances of four materials, methanol, ethanol, pure water and glycerol, were measured. Results indicated that the larger the liquid’s characteristic acoustic impedance value, the more accurate the system. The impedance measurement error is less than0.2%for pure water and glycerol. The impedance of sucrose solution was also measured. Results indicated that the impedance was in linear with the concentration of solution, and the RMSEP of model was1.5g/100g. Besides, the impedance of CaCl2solution was also measured. Results indicated that the impedance was in linear with the concentration of solution, too, and the measurement error was no more than2.7g·L-1for validation data set. The two examples indicated that this measuring system has a nice accuracy and is suitable for measurements in practice. Because of no need of transparence of ultrasound in liquids, this measurement system could be used to measure the concentration of a high attenuation liquid out from a melt container or a pipeline wall.