This work presents a Linear Quadratic Gaussian (LQG) controller in cascade loop applied to a Two-Switch Forward Converter (2SFC). This arrange is suitable for low power battery chargers and provides both performance for reference tracking and disturbance rejection. Once the weighting matrices of the LQG have several parameters to be selected, they are here adjusted by Particle Swarm Optimization (PSO), in a strategy called LQG-PSO. This allows working intuitively with complex characteristics, such as noise, uncertainties and saturation, just by inserting them into the simulation used for PSO. Since this algorithm is a metaheuristic, this work shows a convergence analysis of PSO for this LQG-problem. The obtained solutions from different trials resulted in LQG gains contained in a well-defined region, suggesting that optimality was reached. Simulation results have also shown good dynamic behavior with LQG-PSO strategy, which has outperformed a cascade PI controller for different scenarios. Experimental results comparing the LQG-PSO and PI controllers applied to the 2SFC confirmed that LQG-PSO provided superior performance.

中文翻译：

---

由应用于 DC-DC 转换器的 PSO 调谐的级联回路中的 LQG 控制器

这项工作提出了应用于双开关正向转换器 (2SFC) 的级联回路中的线性二次高斯 (LQG) 控制器。这种安排适用于低功率电池充电器，并提供参考跟踪和干扰抑制的性能。一旦 LQG 的权重矩阵有几个参数可供选择，它们就由粒子群优化 (PSO) 以一种称为 LQG-PSO 的策略进行调整。这允许直观地处理复杂的特征，例如噪声、不确定性和饱和度，只需将它们插入到用于 PSO 的模拟中即可。由于该算法是元启发式算法，因此这项工作展示了 PSO 对这个 LQG 问题的收敛性分析。从不同试验中获得的解决方案导致 LQG 增益包含在明确定义的区域中，表明达到了最优。仿真结果还显示了 LQG-PSO 策略的良好动态行为，在不同场景下其性能优于级联 PI 控制器。比较应用于 2SFC 的 LQG-PSO 和 PI 控制器的实验结果证实 LQG-PSO 提供了卓越的性能。