三电平海上风电柔性直流输电变流器的PID神经网络滑模控制

针对传统滑模变结构控制在参数摄动时会产生颤振的不足,融合比例–积分–微分神经网络(proportional-integral-derivative neural-network,PIDNN)和滑模变结构控制的优点提出一种新的在线解决方法。建立三电平电压型柔性直流输电变流器数学模型,构造以瞬时有功、无功功率误差为滑模面的滑模变结构控制器,并用PIDNN对选定的价值函数在线训练以取得全局最优解,实时对滑模趋近律参数优化选取,结合李亚普诺夫函数对控制系统的全局稳定性进行分析。对所提控制方案采用Matlab仿真验证,结果表明该方案可使控制系统全局稳定,对参数摄动有很强的鲁棒性,最大限度地减小颤振,易于数字实现。

PID Neural Network Sliding-mode Controller for Three-level Offshore Wind Power VSC-HVDC Converter

In this paper,an online real time method that allowed for the merger of the good features of proportional-integral-derivative(PID) neural network(PIDNN) and sliding-mode variable-structure control(SMVSC) design was presented.It can solve the chatting problem when the system input parameters exists disturbances.The mathematical model of the three-level voltage source converters(VSC) in offshore wind farms high-voltage direct-current(HVDC) transmission was developed.A new SMVSC controller based on instantaneous active and reactive power error was proposed.The PIDNN can online train the network with the chosen energy function to get the global optimal solution,which gets the optimal parameter of the sliding-mode control(SMC) in real time.In addition,analytical methods based on Lyapunov function were proposed to guarantee the convergence of tracking error.Matlab simulation results show the proposed method can make system global stability,have stronger robust under system disturbance,and can be easily applied to digital control systems.