论文部分内容阅读
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。
Aiming at the shortcomings of traditional sliding mode variable structure control in the parameter perturbation, such as flutter, proportional-integral-derivative neural-network (PIDNN) and sliding mode variable structure control, New online solution. The mathematic model of three-level voltage-controlled flexible HVDC converter is established, and a sliding mode variable structure controller with instantaneous active and reactive power sliding mode surfaces is constructed. The selected value function is trained on-line with PIDNN to obtain the global optimum Optimal solution, and real-time optimization of the parameters of the sliding mode approach law, the global stability of the control system is analyzed with Lyapunov function. The proposed control scheme is verified by Matlab simulation. The results show that the proposed scheme can make the control system globally stable, robust to parameter perturbations, minimize chatter and make it easy to realize digitally.