Study On The Acoustical Determination Method Of Wheat Moisture Content

Grain moisture has important implications for food transport, storage, processing, andwhich is an important indicator for evaluating the food quality too. In this paper, we researchthe relationship between the acoustic characteristics of wheat kernel and its moisture content.We collected the sound signals what is produced by the wheat kernel hits a metal target, thenanalyze and extract the feature parameters of the wheat moisture content, construct theacoustic determination model of the wheat moisture content. This lay a theoretical basis forresearch and development of wheat moisture content of acoustic measuring instruments.In this thesis, we use the acoustic feature of the wheat kernel to determine its moisturecontent. According to the acoustic detection principle, we design the wheat kernel audiosignal acquisition device, and then gather the wheat grain sound signals. Using thehomemade wheat automatic feeding device, let the wheat kernel continuous, signal, natural tofall and hit the metal target. Using microphone to receive the sound signals what is generatedby the wheat kernel hit the metal target, and then recuperates the sound signals. Using theA/D converter of data acquisition card to dispose the sound signal, and save them to thecomputer. Using signal processing techniques and MATLAB software compile the relatedprocedures to deal with the wheat sound signal. Analysis the sound signal feature in timedomain and frequency domain, and then extract the better characteristic parameters withwheat moisture content. In this paper, we research the different and same varieties of wheatmoisture content, and extracted13acoustic feature parameters with different varieties ofwheat moisture content in the time and frequency domain; extracted4acoustic featureparameters with same kinds of wheat moisture content in the time domain. Select theextracting feature parameters what has a better correlation with the corresponding wheatmoisture content, research and analysis the relationship between them by the linear techniqueand the BP network algorithm. Establish relevant the acoustic prediction model of the wheatmoisture content, then analyze and forecast the effect of the prediction model. Finally, makethe feature of WTAF1and WTAF2as the acoustic prediction model of the different varieties of wheat moisture content. The maximum relative error of the model predicted results is-0.00692%and the average relative error is-0.0067%. Make the BP neural networkprediction model which is based on the feature parameters DCTF5as the acousticdetermination model of the same kind wheat moisture content. The experiment results showthat the error of these two forecast model is small. This indicates that it is feasible fordetermining wheat moisture content by using the acoustic method.