基于计算智能的高压直流输电系统自适应控制

HVDC系统的常规控制普遍采用PI控制策略,该策略是基于固定增益的稳定运行状态进行控制器设计的.由于HVDC系统的响应随运行状态的变化而改变,使得PI控制器的性能很难实现最优.本文基于计算智能理论(Computational Intelligence,CI),将粗糙集理论与粒度计算技术相结合,提出了一种新型的高压直流输电自适应控制器,该控制器可根据HVDC系统运行参数的动态变化对增益大小进行调节,从而使系统的控制性能得到优化,仿真试验结果证明了这一结论.     

基于自适应模糊神经网络的故障诊断方法

对模糊逻辑和神经网络的优缺点进行分析,提出一种基于自适应模糊神经网络的故障诊断方法。它把模糊逻辑和神经网络有机结合起来,生成自适应模糊神经推理系统(ANFIS);文中给出了该系统的原理及实现算法,最后使用M atlab Fuzzy工具箱对柴油发动机喷油泵柱塞故障数据进行建模,结果表明该方法是可行的,并且有较高的诊断效率。     

Intelligent VIV control of 2DOF sprung cylinder in laminar shear-thinning and shear-thickening cross-flow based on self-tuning fuzzy PID algorithm

A self-tuning fuzzy PID (ST-FPID) control scheme is implemented within a joint interactive (Matlab/Simulink/Fluent) co-simulation framework for effective two degrees of freedom (2DOF) vortex-induced vibration (VIV) control of an elastically-mounted circular cylinder in laminar cross-flow of incompressible non-Newtonian power-law fluids based on the control action of a single transverse force actuator. The model-free controller, which systematically tunes the control parameters online in real time based on given rules, is well-known to be highly advantageous over the previously employed conventional PID controllers. It is particularly capable of handling the intricate non-linear dynamic effects inherent in the complex fluid rheology of non-Newtonian flow past the cylinder in presence of unmodeled system dynamics, high parametric uncertainties, diverse operational conditions, and time-varying external disturbances and control signals. Extensive numerical simulations reveal that the complex shear-thinning and shear-thickening behaviors of fluid viscosity can substantially influence the cylinder dynamic response, applied hydrodynamic forces, and flow structure. In particular, effectiveness and high performance of the adopted ST-FPID control strategy in substantial suppression of the high amplitude coupled 2DOF VIV of the elastically-mounted cylinder at selected critical reduced velocities in the lock-in region are established for a wide range of power-law index parameters (e.g., up to 83% reduction in RMS value of cylinder cross-flow displacement and up to 35% reduction in RMS value of cylinder in-line displacement for n=1and U* = 5 at Re = 100). Also, the vigorous action of the error-driven ST-FPID controller in forcing the high strength vortex shedding patterns of the uncontrolled cylinder out of the lock-in condition into the classical von Kármán vortex street of 2S-type mode of moderately weaker strengths is verified.