基于虚拟仪器和神经网络的禽蛋检测系统设计

文针对现有禽蛋无损检测方法系统集成度不高，操作界面易用性差，测量误差大、试验效率低等不足,研制了基于Lab Windows/CVI的多传感器融合的禽蛋无损检测虚拟仪器品质分级系统。该系统以Lab Windows/CVI软件为设计操作平台，以MATLAB神经网络的分析算法为分析模块核心，充分结合计算机视觉检测和声学检测方法的优点，在无损检测的信息采集和处理层面实现了多传感器融合。系统通过系统静态分级测试,系统动态分级测试和系统连续动态测试等3种测试，测试结果表明该系统界面易用性好，在快速运行（27个样品/min）的情况下具有较好的准确度 (系统连续动态测试准确率大于90%)，和漏检率（最高速情况下小于3%）。

Design of a non-destructive detection system for eggs based on virtual instrument and neural network

The system-integration degree, system working efficiency and user interface of egg non-destructive detection system cannot meet the requirement of fast and accurate detection. A non-destructive detection system for eggs based on virtual instrument and artificial neural network was designed and evaluated. This system used Lab Windows/CVI as the system design platform and the neural network as the core-analysis module. This approach fully uses the advantages of both computer vision and acoustics non-destructive detection. This system realizes the amalgamation of multi-sensors in data collecting and processing.. Three kinds of tests were conducted--the system static grading test, the system dynamic grading test and the system continuous dynamic test. The test results indicate this system was easy to use and provided a good graphical user interface. The system has good veracity (system continuous dynamic test result above 90%) and low un-detecting rate at high speed of the gearing machine (27 samples/min).