机构地区: 中山大学
出 处: 《机械科学与技术》 2006年第10期1191-1193,1240,共4页
摘 要: 从提高Bayes决策精度的角度出发,结合线性回归理论,对机械系统损伤定位问题提出了一种新的集成神经网络决策融合算法,并应用于多个概率神经网络的集成。采用简单遗传算法优化了集成神经网络的参数,有效地提高了集成神经网络进行损伤位置识别的精度。用质量-弹簧系统模型作为算例,验证了本文理论和方法的有效性。 A novel decision fusion method for integrated neural networks (INN) is proposea for me purpose of enhancing the Bayes decision accuracy and in combinat on with the linear regression theory. The method is used to locate damage to mechanical systems and app ted to the integration of multiple probabilistic neural networks. The INN's parameters are optimized by using a simple genetic algorithm, which enhances the damage location identification accuracy effectively The theory and method proposed in the paper are verified by using a mass-spring system model as a calculation instance.
关 键 词: 集成神经网络 概率神经网络 决策融合 简单遗传算法
领 域: [自动化与计算机技术] [自动化与计算机技术] [建筑科学]