基于DFNN的智能配电异构无线网络准入控制算法

为提高智能配电通信业务的服务质量,根据智能配电网对通信技术的要求,提出一种基于动态模糊神经网络(DFNN)的智能配电异构无线网络准入控制算法。在智能配电网络的异构准入控制模型中构建神经网络系统,以网络的接入阻塞率差作为系统参数强化学习的目标,对网络的负载均衡具有较好的动态适应性。神经网络系统在输入层较多时容易产生太多规则而影响决策结果,而DFNN通过计算当前系统规则的完备性,动态添加规则,并通过计算所有规则的重要性,动态删除规则,使得系统的规则有效而不冗余。仿真结果表明,该方法较多接入选择算法(MLB)明显降低了网络的接入阻塞率,相对于模糊神经网络算法(FNN)而言简化了系统结构,突出了规则的重要性,具有较低的接入阻塞率和更好的均衡效果。

Access Control Strategy of Heterogeneous Wireless Networks Based on DFNN for Smart Distribution Grid

According to the requirements of smart distribution grid (SDG), an access control strategy based on dynamic fuzzy neural network (DFNN) in heterogeneous wireless networks for SDG is proposed to improve the quality of service (QoS) of the communication business of SDG. A neural network system is built in heterogeneous access control model of SDG with the objective of taking the equal blocking probability of access networks as system parameter to enforce learning process, which has better dynamic adaptability for load balancing of the fuzzy neural network. Due to such problem that neural network generate too many rules that may influence the result in input layer, rules are added automatically through calculating completeness of current system rules by DFNN, and rules also can be deleted dynamically by computing the importance of all rules, which make system rules work effectively without redundancy. The simulation results show that proposed DFNN obviously decrease the blocking probability of access networks by comparing with that of MLB algorithm. At the same time, DFNN simplify the structure of neural network and enforce the importance of rules with FNN algorithm, which has lower access blocking ratio and better balance effect.