| Getting rid of cracked eggs is an important procedure for guaranteeing quality of processing eggs and reducing cost. Detecting eggs by hand is very difficult for it's dirty working conditions and laboring intension, and the efficiency is very low. So finding a scientific, applicable and fast detecting method for removing cracked eggs is very urgent and valuable.In this task, first we analysed the egg's phonation mechanism when it is knocked by a small rubber bead. A intact egg and a cracked egg don't phonate similarly because of different structures. Established the detecting system based on DSP, and then knocked on eggshells. The analog sound signals were converted to diginal signals after they had been amplified and filtered. We adopted the threshold circle to check the signals' section start, which successfully resolve the contradiction between DSP interior memorizer is small and the digital signal processing requires a lot of memory resources. We used DSP serial port to receive the digital signal converted by AD convertor, which simplified the layout of hardware in this system.Second, we maked spectrum chart for the sound signal by this system, after this we found out remarkable frequency factors that are affected to distinguish cracked eggs and intact eggs. We also analysed the cepstrum of the sound signal, and found out the remarkable factors in this field. Introduction of cepstrum remarkablly advanced the veracity. We made a lot of experiments to improve the detecting tactic. Considering the location, knocking on the bigger side, middle and smaller side respectively, we checked the certain location of this eggshell immediately, and then used this thrice results to allege this egg's status. We did thrice Stat. analysis respectively according to these influence factors. We made use of step discriminance to optimize the combination of factors, and then used the "Bayes" discriminance to establish the optimum distinguishing functions of intact eggshells and cracked eggshells respectively.At last, designed the realtime detecting system. We used assemble language to exploit the software of the system, which can stably execute realtime sampling and analyse of the sound signal, can judge whether the eggshell is cracked or not, and can control the sampling frequency.The system has been tested by experimentation, and the results showed that the detecting veracity is not sensitive to the knocking intension, which can deemphasize the precision when design knocking device. The system worked in high efficiency, could detect 36 eggs per second at most. The system could detect intact eggs at the accuracy of 92.8 percent and detect cracked eggs at the accuracy of 82.7 percent. |
