dSPACE的单神经元温度控制算法研究

Single Neuron Temperature Control Algorithm Based on dSPACE

针对电加热炉温度过程控制系统,利用基于dSPACE的电加热炉实时半实物仿真平台,将单神经元自适应PID算法应用于dSPACE温度控制系统中,并完成S函数的M EX C代码格式建模,在M atlab/Simulink中进行模块化设计,将实质性代码放在S函数源文件之外的C文件中,完成单神经元自适应PID算法的移植和维护,实现方便快速原型化设计。实验表明,在dSPACE实时系统中应用单神经元自适应PID算法后,既利用了神经网络充分逼近任意非线性函数的能力,又发挥了PID控制器结构简单、稳态精度高及动态响应速度快等优点,同时dSPACE平台大大缩短了研究周期,获得了良好的控制效果。

The working principle of temperature control system based on dSPACE real-time system and the principle of single neuron adaptive PID algorithm are introduced. Based on the dSPACE-based real-time hardware-in-the-loop simulation platform,a single-neuron adaptive PID algorithm is applied to the dSPACE temperature control system,and the MEXC code format of S-function is completed. Modularized design in Matlab/Simulink,the substantive code is placed in the C file outside of the S function source file,and the transplantation and maintenance of the single neuron adaptive PID algorithm are completed,which is convenient and rapid prototyping design. Experiments showthat single neuron adaptive PID algorithm in dSPACE real-time system not only can fully realize the ability of neural network to approximate any nonlinear function,but also has advantages of simple structure,high precision of steady state and fast dynamic response. And dSPACE platform greatly shortens the research cycle and obtains good control effect.