基于DCQGA的PID参数自整定

Self-tuning of PID Parameters Based on DCQGA

量子遗传算法相对于普通遗传算法具有更好的多目标寻优能力,在PID参数优化中具有一定优势,但是标准量子遗传算法容易陷入局部极值,并且在进化后期容易出现早熟收敛的现象,提出了一种新的方法——双链量子遗传算法(DCQGA)来整定PID控制器的参数。该方法直接用实数来编码基因位,用量子旋转门和量子非门进行染色体的更新和变异,用最优解和当前解量子位概率幅组成的行列式来确定量子门旋转方向,用结合目标函数梯度信息的自适应策略调整旋转角大小。同时采用误差绝对值时间积分性能指标作为PID参数选择的最小目标函数进行Matlab仿真,通过与普通遗传算法和标准量子遗传算法整定效果的对比,结果验证了DCQGA具有更快的响应速度和更好的多目标寻优能力。

Quantum genetic algorithm has better multi-objective optimization ability than general genetic algorithm. It has some advantages in PID parameters optimization. However, standard quantum genetic algorithm is easy to fall into local extremum and prone to premature convergence in late evolutionary stage. A new method of double chain quantum genetic algorithm （DCQGA） is proposed to tune parameters of the PID controller. Real number is directly used to encode genes position in the method. The updating and aberrance of chromosomes are realized with quantum rotary gates and negater. The rotation direction of the quantum gates is determined by the determinant of the probability of the optimal solution and the current solution qubit. The size of the rotation angle is adjusted by the adaptive strategy combined with the gradient information of the objective function. At the same time, performance index of the error absolute value time integral is used as the minimum objective function of PID parameter selection to carry on the Matlab simulation. Through comparison with the general genetic algorithm and standard quantum genetic algorithm, the result verifies DCQGA has faster response speed and better multi-objective optimization ability.