一种复合NN智能诊断系统的设计及应用研究

Design and Applied Research on a Compound NN Intelligent Diagnosis System

NN智能诊断系统通过把计算机技术、NN技术、ES、GA与力学模型融合在一起,建立一种新型的智能诊断系统。将GA与改进的BP网络相结合,利用GA宏观寻优特点,用来克服BP网络的缺点,二者结合起来相得益彰。与传统的BP算法相比较,GA-BP网络训练速度和预测精度都得到了很大的提高。NN与ES互补,结合ES经验,预测使用寿命。应用MATLAB软件编制的具有自学习功能的智能化监测诊断软件,结合VC++6.0和MATLAB的各自优点和功能,实现了VC++6.0与MATLAB的混合应用,软件设计比较简单。根据系统的分析、设计、编程和仿真实验,系统运行效果很好。应用本系统对混凝土结构进行诊断,能给出结构受损程度,损伤位置及结合ES预测使用寿命,具有高度棒性、高精度、快速的特点。

NN intelligent diagnosis establishes a new type of intelligent diagnosis system through the mix of computer technology, NN technology, ES, GA and mechanical model. Using macro - GA -optimizing features, GA combines with improved BP networks to overcome the shortcomings of BP networks and the combination of the two complements each other. Compared to traditional BP algorithm, GA - BP network＇ s training speed and accuracy have been greatly improved. NN complements ES, together with ES experience, can forecast life length. Applying the intelligent monitoring diagnosis software which has a self - learning function, it achieves VC ＋ ＋ 6.0 and MATLAB mix - use when combined with their respective advantages and functions and the design of the software is relatively simple. According to system analysis, design, programming and simulation, the system is running well. Applying this system to diagnose the concrete structure, it can give the extent of structural damage, the location of injury. And when it combines with ES service life length, it has a quality of high robustness, precision and rapidness.