ABSTRACT

Due to the non-availability of model parameters, the model-base control of a nonlinear and infinite-dimensional underground coal gasification (UCG) process is a challenging task. In this paper, a robust neuro-adaptive sliding mode control (NASMC) is designed for the UCG process to maintain a desired heating value level. The unknown model parameters used in NASMC are estimated using the feed-forward neural network. Moreover, the controller also requires time derivatives of some model parameters, which are estimated by uniform robust exact differentiator. As the relative degree of the output with respect to the input is zero, therefore, to apply NASMC, the relative degree is increased to one. This approach maintains the desired heating value and provides insensitivity to input disturbance and model uncertainties. A comparison is also made between NASMC and an already designed conventional SMC. The simulation results show that NASMC exhibits better performance as compared to the conservative SMC design.

摘要

由于模型参数不可用，非线性和无限维地下煤气化（UCG）过程的基于模型的控制是一项具有挑战性的任务。在本文中，为 UCG 过程设计了一种鲁棒的神经自适应滑模控制 (NASMC)，以保持所需的热值水平。NASMC 中使用的未知模型参数是使用前馈神经网络估计的。此外，控制器还需要一些模型参数的时间导数，这些参数是通过统一鲁棒精确微分器估计的。由于输出相对于输入的相对度数为零，因此，为了应用 NASMC，相对度数增加到一。这种方法保持了所需的热值，并且对输入干扰和模型不确定性不敏感。还对 NASMC 和已经设计好的传统 SMC 进行了比较。仿真结果表明，与保守的 SMC 设计相比，NASMC 表现出更好的性能。