基于改进的多种群遗传算法的LQR汽车悬架主动控制研究

Research on Active Control of LQR Automobile Suspension Based on Improved Multi-Population Genetic Algorithm

摘要: 文章以1/4车辆二自由度的主动悬架作为研究对象，通过牛顿第二定律建立1/4车辆二自由度汽车主动悬架的动力学模型，之后构建线性最优二次型控制函数。针对加权矩阵Q和R两个参数在LQR控制函数中通过人为经验手动输入对于优化精确度的干扰，使得最后的优化结果是次优解的问题，本次实验提供了一个新的思路：基于自适应的多种群遗传算法的LQR汽车悬架控制。这个方法通过多种群遗传算法与自适应函数相结合，使得多种群遗传算法可以通过自适应函数进行参数自适应调节，从而提高控制器参数的收敛速度，确保实验结果的精确性。LQR控制器的控制参数通过遗传算法进行获得。仿真结果表明，该控制算法相比于传统的多种群遗传算法寻优结果更佳，并且对于车身加速度、悬架动行程、轮胎动位移都有着有效的优化，降低了在驾驶过程中由于路面的不平稳导致的车身的震动，使得汽车的平稳性、安全性以及舒适性有着较大的提高。

Abstract: The article takes a 1/4 vehicle two-degree-of-freedom active suspension as the research object. Based on Newton’s second law, a 1/4 vehicle two-degree-of-freedom active suspension dynamic model is established. Subsequently, a linear quadratic regulator (LQR) control function is constructed. To address the issue that manually inputting the weighting matrices Q and R in the LQR control function based on human experience interferes with optimization accuracy, resulting in suboptimal solutions, this experiment proposes a new approach: LQR vehicle suspension control based on an adaptive multi-population genetic algorithm. This method combines the multi-population genetic algorithm with an adaptive function, allowing the algorithm to adjust parameters adaptively through the adaptive function, thereby improving the convergence speed of the controller parameters and ensuring the accuracy of the experimental results. The LQR controller parameters are obtained through the genetic algorithm. Simulation results indicate that this control algorithm achieves better optimization results compared to traditional multi-population genetic algorithms. It effectively optimizes vehicle body acceleration, suspension dynamic stroke, and tire dynamic displacement, reducing vibrations caused by road unevenness during driving, and significantly improving vehicle stability, safety, and comfort.