基于RBF神经网络的自动包装机温度控制算法研究

目的针对传统热封工艺中温度调节PID算法参数过度依赖人工经验的缺点,提出一种RBF神经网络与PID算法相结合的具有参数自适应的热封温度控制算法。方法使用控制系统的输出误差作为代价函数,采用最小均方误差(LMS)调整权值与偏置参数,并通过中心自组织算法实现径向基函数中心和中心宽度的动态调节,在Matlab软件中的Simulink子系统中建立仿真模型进行算法验证,并与传统PID控制算法进行比较。结果仿真结果表明,径向基神经网络与传统PID算法的结合使得系统输出响应在动态性能和静态性能方面均优于传统PID,在系统上升时间、调节时间等方面均优于增量式数字PID。结论将RBF神经网络PID算法应用于自动包装机,避免了传统热封工艺中PID控制算法参数不能适应于复杂变换控制环境的问题,神经网络PID算法的自适应性强,实现了热封温度变化下PID参数的自动调整,在一定程度上提升了生产效率和包装设备的智能化水平。

Temperature Control of Automatic Packaging Machine Based on RBF Neural Network

The work aims to propose a heat seal temperature control algorithm with parameter adaption based on RBF neural network and PID algorithm for the excessive dependence of parameters of PID control algorithm on artificial experience in traditional heat sealing process. The output error of the temperature control system was used as the cost function, the minimum mean square error (LMS) was applied to adjust the weight and bias, and the dynamic adjustment for the radial basis function center and the center width was realized by the center self-organization algorithm. The simulation model was set up in the Simulink sub-system of the MATLAB software to verify the algorithm and then compared with the traditional PID control algorithm. The simulation results showed that the combination of RBF neural network and traditional PID algorithm made the output response of the system better than the traditional PID in both dynamic performance and static performance and also better than the incremental digital PID in rise time and adjustment time. The RBF neural network PID algorithm is applied to the automatic packing machine to avoid the problem that PID control algorithm parameters cannot adapt to the complex transformation control environment in the traditional heat sealing process. The neural network PID algorithm has strong adaptability and realizes the automatic adjustment of PID parameters under the change of heat seal temperature, so it can improve the production efficiency and the packaging equipment intelligent level to some certain.