基于遗传小波神经网络的多传感器信息融合技术的研究

Study on genetic wavelet neural network based multi-sensor information fusion technique

依据小波函数的非线性逼近能力和神经网络的自学习特性,提出一种小波神经网络。为使小波神经网络具有更高的学习精度和更快的收敛速度。利用遗传算法对小波神经网络权阈值的优化,设计了遗传小波神经网络。将该网络用于多传感器信息融合设计了遗传小波神经网络多传感器信息融合系统。压力传感器数据融合系统的仿真表明该方法能有效的提高传感器的输出准确度,消除非目标参量对传感器输出结果的影响,此系统还可用于其他多传感器信息融合系统,具有实际应用价值。系统设计实现简单,适合工程应用。

Based on the non-linear approximation ability of wavelet and the serf-learning characteristic of neural network, a wavelet neural network is presented. In order to obtain higher learning accuracy and faster convergence speed, genetic algorithm is introduced to optimize the parameters, the genetic wavelet neural network is put forward. A multi-sensor information fusion system is designed based on the genetic wavelet neural network. The simulation of pressure sensor information fusion system shows that this system effectively improves the output accuracy of the sensor and successfully eliminates the impact of non-object parameters on sensor output. This system is also practicable for other types of multi-sensor systems and has its practical value. The system is simple and suitable for engineering use.