Research On Eggs Dynamic Weighing Algorithm And Embedded Software

With the continuous improvement of living standards of Chinese residents,people’s requirements for the quality of poultry and agricultural egg products are also higher and higher.Therefore,the demand for egg grading equipment is becoming higher and higher.The egg weight is an important index to measure the egg quality.In this paper,based on the egg dynamic weighing machine and the dynamic weighing signal conditioning module developed by the research group,an embedded software system for egg dynamic weighing is developed.Based on the selection of digital filter and the weight prediction based on the signal eigenvalue,the influence of different FIR filter and IIR filter on the filtering effect of the original signal of egg dynamic weighing is studied.At the same time,the impact of using different filtering signal eigenvalues to extract symmetrical weight was studied,and the edge computing of neural network using the second generation of neural compute stick for egg dynamic weighing was initially carried out.(1)The influence of different filter types on the filtering effect of the original signal is studied.The main conclusions are:1)The noise frequency of the original signal of egg dynamic weighing is more than 20 Hz,and the filter order of 20 Hz and below is the main one when the filter order is selected;2)When the weighing speed of FIR filter is 5 / s,it is easy to cause the program running error,which leads to the phenomenon of losing value.That is to say,under the double influence of high filter order and high speed,the running time of the filter program will affect the correct logic operation of the program.3)It is more suitable to adopt different digital filters according to different weighing speeds.When the weighing speed is less than 3.5pieces/second,the Butterworth filter with a cutoff frequency of 20 Hz and 7 steps is used.When the weighing speed is higher than 3.5 pieces / second,the 80 order Kaiser window filter with beta= 4.55 and the cutoff frequency of 5Hz is used.(2)The influence of different methods of filtering signal eigenvalue extraction algorithm on symmetrical weight effect is studied.The main conclusions are:1)Because of the lag of the weighing sensor,and the weighing result is on the high side in high-speed continuous weighing;2)The arithmetic means,arithmetic means optimization and median mean are used as the algorithm of filtering signal eigenvalue.When the weighing speed is 3 / s,the error range of the three algorithms is within ± 1g.When the weighing speed is 4 / s,it is better to use the arithmetic mean to optimize the mean value of the same median,and the error range is about ± 1g.However,when the weighing speed is 5 / s,the test results of the three algorithms have significant errors.(3)The application of neural compute stick,and edge computing in dynamic weighing is explored.A simple communication master computer is designed,and the result of filtering signal processing with the BP neural network is explored.To a certain extent,the BP neural network built in this paper does not improve the weighing accuracy in extracting the characteristic value of the filtered signal.