改进BP神经网络在电力手持防误系统的应用

Application of Improved BP Neural Network in the Electric Hand-held Anti-misoperation System

采用C#编程语言,结合WINCE嵌入式技术、条形码扫描技术,开发设计了基于PDA(个人数字助理)平台的电力防误系统。传统条形码扫描是通过光束进行,对易损坏和脏污条形码无法正常扫描识读。开发设计缺损条码识别防误功能模块,在采集缺损条码图片后,进行图像预处理,然后结合神经网络进行识别。进一步地,给出一种新的、基于混沌理论的遗传算法对神经网络进行改进;改进后的神经网络在系统中应用提高了条码识别准确率、加快了识别速度。实际应用表明,该系统能够规范变电所人员的日常工作,减少误操作,提高工作效率。

The electric anti-misoperation system based on the platform of PDA （personal digital assistant） is exploited and designed, C# is the main programming language, and the WINCE embedded technology as well as the bar code scanning technology are combined in the exploitation. Traditional bar code scanning is realized by light beam, while the damaged and dirty bar codes can＇t be scanned and identified normally. The anti-misoperation function module of defective bar code recognition is developed and designed, that the images are preprocessed after the aqcuisition of defective bar code images with the identification using the neural network. Furthermore, a new genetic algorithm based on chaos theory is given to improve the neural network. In the system application, the bar code recognition based on the improved neural network is realized with increased accuracy and faster speed. And it is revealed by the practical application that the system can regulate the personnel daily work in the substation, and decrease misoperation, which improves the working efficiency.