Study On Rapid Determination Of Moisture By Predictive Loss On Drying

Moisture content in substance is one of the important factors which determine its physical,chemical and biological properties of a substance.Loss on drying method has the advantages of high precision and wide application range.It is the standard method and arbitration basis for moisture determination of solid samples in many industries.It provides the most reliable results but is both labor intensive and time consuming.For a long time,the contradiction between the measurement accuracy and rapidity of the loss on drying method has always been the application limitations.Aiming at the technical bottleneck of the time-consuming and energy-consuming of the loss on drying moisture analyzer,scholars at home and abroad have improved the loss-in-dry method through two aspects.(1)Using the infrared and microwave heating method to increase the thermal efficiency of the drying oven.(2)The intelligent information processing method is applied to the data analysis and processing of the loss on drying method,and the moisture content of the substance is accurately calculated when the sample is not completely dried.Numerous research results show that the drying properties of the tested samples are inherent properties of the material.Even with advanced heating methods such as infrared heating or microwave heating,the dry characteristic curve of the tested sample shows periodic characteristics.On the other hand,the researches on the moisture content prediction fusion algorithm of loss on drying method are still in the exploration stage.There are many deficiencies in the extensiveness of the sample being tested,modeling mechanism,and the algorithm design.There is no mature technologies and products available.This dissertation supported by the National Natural Science Foundation of China," Energy saving and predictive grain moisture analyzer "(Grant No.61663039).Aiming at the application limitation of traditional loss on drying method,the adaptive moisture content predictive fusion method based on the loss-of-drying method is proposed.The main research contents include: 1)The predictability feature extraction of loss on drying process for different kinds of tested samples,2)The establishment of subtype prediction model(colloidal-like porous medium type and capillary-like porous medium type),3)Moisture content predictive fusion method is designed based on damping factor adaptive Levenberg-Marquardt(LM)algorithm,4)The application of predictive fusion method in loss on drying method moisture analyzer.As a kind of “no-screening” moisture determination method,the universality and diversity of the tested samples of loss on drying method is the biggest advantage of the method and the biggest difficulty in establishing the estimation fusion method.Aiming at this difficulty,the periodic characteristics of the drying characteristics of the water-containing samples are intensely researched based on the drying kinetics and thermodynamics theory.The drying characteristic curve and the drying rate curve are used as indicators to analyze the test sample,drying temperature,sample size,initial moisture content and initial mass on the drying characteristic curve and water loss rate of the sample.Based on the above findings,the typical samples of the loss on drying method are further refined into colloidal-like porous medium type and capillary-like porous medium type.Through theoretical research and experiments,it has been proved that the gradual stability of the falling rate period and the dehydration difference in the accelerating rate period are the main features of the predictability of the drying weight loss process of the two types.For the colloidal-like porous media type,the accelerating rate period is extremely short and difficult to capture.The dehydration center is located in the falling rate period.Through in-depth study of the irreversible transport mechanism and main driving force of moisture diffusion in wet samples during infrared drying process,Luikov theory and Fick's diffusion law are applied to the establishment of prediction model of drying weight loss method.The application range and boundary conditions of the theory are revised.A prediction model of the falling rate period is established based on Luikov theory and Fick's diffusion law.The goodness of fit of the model is evaluated and analyzed.A regression equation between the estimated model parameters and the drying conditions is established.For the capillary-like porous medium type sample,the drying time is short,and the dehydration center is located in the accelerated drying stage.The shortcomings of the existing drying weight loss prediction model are deeply concluded in this dissertation.Based on the numerical analysis characteristics of the water loss curve of the whole process of drying,a whole drying process prediction model is presented with an improved power exponential form.The goodness of fit of the model is verified by the measured data.Subsequently,on the basis of verifying the correctness of the prediction model,a prediction fusion method based on LM algorithm for estimating the moisture content of the loss on drying method is established.Aiming at the strong dependence on iteration initial values and weak convergence caused by the fixed value update of the traditional LM algorithm,the trust region method search technique is applied to the adaptive update of damping factor,and the adaptive discriminant mechanism for estimating the start and end time is proposed.The serial iterative structure prediction algorithm based on the damping factor adaptive LM algorithm is established to estimate the moisture content of the sample when the sample is not fully dried.The global convergence of the moisture content prediction fusion algorithm is proved by the optimization theory.The execution efficiency and accuracy of the algorithm are verified by experiments and simulations.The comparison accuracy of the adaptive LM algorithm is verified by comparison with the reference method.The algorithm's noise immunity is investigated and analyzed.Finally,the design of predictive fusion algorithm is combined with the design of moisture analyzer for loss on drying method,and the design of the predictive moisture analyzer is implemented on the embedded system platform with the structure of DSP + MCU.Finally,with the reference to Chinese National Standard "JJG 658-2010 drying moisture analyzer",the mass measurement performance index of the predictive moisture analyzer including error of indication,the repeatability error and error of measurement results of moisture under normal working conditions were tested.The uncertainty model of test results is established based upon the sources of test errorsThe test results showed that all the indexes of the predictive moisture analyzer meet the design requirements,including the error of indications in all of the scale is not greater than 0.005 g,and the repeatability is not greater than 0.008 g,the determination of the moisture content in the normal mode is not greater than 0.5%,the determination of the moisture content in the estimation mode is less than 0.5%.All these indexes meet the requirement for the high accuracy grade(2nd class)moisture analyzer specified by Chinese National Standard "JJG 658-2010 drying moisture analyzer".