PCR仪器的IGWO-BP神经网络PID控制

The IGWO-BP Neural Network PID Control of PCR Instrument

聚合酶链式反应(polymerase chain reaction,PCR)仪温度控制存在时滞性、非线性、惯性大、时变等特点,通过传统的PID温度控制效果较差,远不能满足要求,而BP神经网络PID控制通过自适应学习较传统PID控制而言控制能力更强,其在聚合酶链式反应仪温度控制方面的应用基本还是空白,但BP神经网络的初始权重是随机选定的且沿着正方向不断进行调整,易导致多次训练结果不一致。针对这些问题,提出了一种基于改进灰狼算法(improved grey wolf optimal,IGWO)的BP神经网络自调整PID参数的控制方法,利用改进灰狼算法良好的全局搜索能力将其寻找的最优位置作为BP神经网络的初始权值。最后仿真结果表明,该算法很好地满足了PCR仪温度控制的要求,与传统方法相比,在同一升降温速率下,温度控制系统超调量为0.2%,大幅度减小且无稳态误差,对于聚合酶链式反应仪温度调节具有更佳的效果。

Polymerase chain reaction instrument temperature control has the characteristics of time delay,nonlinear,large inertia and time-varying,through the traditional PID temperature control effect is poorer and falls far short of what is the satisfied requirements.The BP neural network based PID control has better control ability than traditional PID control through adaptive learning,and its application in the temperature control of polymerase chain reaction instrument is basically blank,but the initial weights of BP neural network is randomly selected and along the positive direction of constant adjustment.The weights of the BP neural network in the process of iteration,and easy to fall into local optimum,lead to inconsistent results training for many times.To solve these problems,we proposed an algorithm based on improved grey wolf optimal location as initial weights of BP neural network,and using BP neural network to adjust PID parameter.Finally,the simulation results show that IGWO--BP neural network’s convergence speed and accuracy are improved,achieve the desired effect,and the algorithm is well meet the requirements of the PCR temperature control.Compared with the traditional method,under the same temperature rise and fall rate,the overshoot of the temperature control system is 0.2%,which greatly decreases,and no steady—state error,it has a better effect on the temperature regulation of the polymerase chain reaction instrument.