智能压路机压实模糊神经网络控制及仿真

Compaction fuzzy neural network control and simulation for intelligent road roller

将模糊系统与神经网络相结合,提出了一种由模糊化层、模糊推理层和清晰化层组成的模糊神经网络结构,并将其用于智能压路机压实控制.针对振动压路机的压实性能要求,采用钟型函数作为隶属度函数,通过计算规则重要度来提取模糊推理层规则群中比较重要的规则,运用补偿模糊神经网络的学习算法解决参数的自动调整问题.把工程实践中得出的模糊控制规则表作为训练模糊神经网络的样本,仿真结果表明该模糊神经网络控制器具有在误差限度范围内的泛化能力.

A kind of structure of fuzzy neural network is presented by connecting fuzzy system with artificial neural network in this paper, which consists of the fuzzification layer, fuzzy inference layer and clarification layer and is used for compaction control of intelligent road roller. For the requirement of compaction performance of vibratory roller, bell function is used as membership function; more important rules are obtained from the rules of fuzzy inference layer by computing important degree of rules; and self-adjusting of fuzzy neural network parameters are settled by leaming algorithms of compensated fuzzy neural network. Fuzzy control rules table that was obtained from engineering practice is taken as training sample of fuzzy neural network. The result of simulation shows that the fuzzy neural network controller has generalization ability in error limit.