Wheat Gradation And Quality Tester Based On Multi-sensor Data Fusion

Now it is hard to judge the wheat gradation and acquire accurate measuring result by our eyes,which brings big error, is subjective and is bad operable and repeatable. Different people can reach a totally different result, which make difficult to grade the wheat reasonably, so it can not meet the requirement of objectivity and accuracy of wheat gradation when the wheat is purchased.According to the wheat gradation standard by the State Bureau of Quality and Technical Supervision in 2008, wheat quality assessment is mainly connection with wheat moisture content and volume-weight. The system designs a wheat gradation and quality tester based on multi-sensor measurement, to meet the actual demand in the wheat gradation and realize its automation and intellectualization. We regard high-speed single-chip computer C8051f020 as main controller, adapt sevaral sensors, combine advanced information fusion technology, complete the data fusion on the wheat moisture detection, realize the accurate gradation of wheat.The system adapts fixed measuring glass filled with wheat, adapts resistance strain sensor as precision weight sensor, adapts the differential bridge by compensate for temperature to amplify the electrocircuit, and adapts specified 24-bit AD transducer CS5550 to obation accurate volume-weight value of wheat.The system adapts the technology of capacitance sensor to detect moisture content, regards fixed measuring glass as cylindrical capacitance moisture sensor. Capacitive value of wheat has the relation of temperature, tense degree of wheat.On the basis, so we adapt BP neural network information fusion algorithm, realize data fusion about capacitance, temperature, and volume-weight, and achieve accurate test of wheat moisture.Wheat moisture fusion based on BP neural network conducts in two phases.The first step is forward propagation, which needs to import the measured moisture, volume-weight and temperature value to hidden layer and output layer neural cell and calculate.The second step is back-propagation process, which needs to compare the former result with the actual value and obtain a difference. The system will modify weights and thresholds according to the corresponding formula until the difference reachs the permissible error range, and finally obtain the weights and thresholds.Error of system measurement remains within 0.2% compared with national standard method. The instrument achieves the rapid detection on parameters of GB about the wheat moisture content, volume-weight, provides a relatively objective test method, and promotes the development of wheat measurement.