论文部分内容阅读
铝电解过程是一个非线性、多耦合、时变和大时滞过程,受强电场、强磁场、强热场交互干扰,形成了复杂多变的槽况特征,故障种类繁多,发生频繁,有效地故障预报和诊断,对电解系列平稳供电,节约电能、提高铝的产量和质量有重要意义。根据铝电解过程故障特点,提出了基于主成分分析的集成神经网络铝电解多故障诊断方法,建立分层故障诊断模型结构,包括子神经网络层和决策融合神经网络层,子神经网络模块采用了改进型的Elman神经网络,强化信息的记忆功能,并通过主成分分析优化了神经网络结构;决策融合神经网络通过各子网络传递的相关信息,进一步验证对子神经网络诊断结果和复合故障进行综合决策。仿真结果表明,具有良好的诊断效果,验证了该故障诊断方法的可行性和有效性。
Aluminum electrolysis process is a nonlinear, multi-coupled, time-varying and large-time delay process. Due to the interaction of strong electric field, strong magnetic field and strong thermal field, the complex and changing characteristics of the tank are formed. The variety of faults occurs frequently and effectively Fault prediction and diagnosis of the electrolytic series of smooth power supply, save energy, improve the output and quality of aluminum is of great significance. According to the characteristics of the aluminum electrolysis process fault, this paper proposes a multi-fault diagnosis method based on principal component analysis (ICA) of integrated neural network for aluminum electrolysis, builds a layered fault diagnosis model structure, including sub-neural network layer and decision fusion neural network layer, Improved Elman neural network to strengthen the memory function of information, and optimize the neural network structure through principal component analysis; decision-making fusion neural network through the relevant information passed through each sub-network to further verify the diagnostic results of sub-neural network and composite fault synthesis decision making. The simulation results show that the method has good diagnostic results and the feasibility and effectiveness of the fault diagnosis method are verified.