论文部分内容阅读
首先基于一种扩展原理和模糊算术得到一类前向模糊神经网络——折线模糊神经网络.当模糊神经网络的输入为一般模糊数,激励函数为单调连续型Sigmoidal函数时,分析网络的拓扑结构及相关性质.然后证明该折线模糊神经网络能作为模糊连续函数的通用逼近器,其等价条件是模糊函数的递增性.因此关于输入为一般模糊数的折线模糊网络是否为通用逼近器的问题得到解决,且折线模糊神经网络的应用范围将进一步扩大.
First of all, based on an extended principle and fuzzy arithmetic, a kind of forward fuzzy neural network-polyline fuzzy neural network is obtained.When the input of the fuzzy neural network is a general fuzzy number and the excitation function is a monotonically continuous Sigmoidal function, the topology of the network And then it is proved that the broken line fuzzy neural network can be used as a universal approximation of fuzzy continuous functions and the equivalent condition is the increment of fuzzy functions.Therefore whether the broken line fuzzy network input as a general fuzzy number is a universal approximation Be resolved, and the application of the line fuzzy neural network will be further expanded.