针对工业过程的非线性,本文首先提出了一种操作轨迹非线性模型的辨识方法:根据调度变量的操作轨迹,选取若干个典型工作点;在各个典型工作点,辨识各自的线性模型;根据测试数据以及过渡数据,得到全局插值非线性模型。在此基础上,本文进一步提出一种基于操作轨迹模型的非线性预测控制算法,并采用多步线性化方法进行问题求解。由于仅需要在典型工作点上进行测试,降低了全局建模的辨识成本,而且控制品质好,仿真结果表明了该算法的有效性。 Aiming at the nonlinearity of industrial process, this paper first proposed a method to identify the nonlinear model of operating trajectory. According to the operating trajectory of scheduling variables, several typical working points were selected. At each typical operating point, the linear models were identified. Data and transition data to obtain a global nonlinear interpolation model. On this basis, this paper further proposed a non-linear predictive control algorithm based on the operating trajectory model, and used multi-step linearization method to solve the problem. Because it only needs to be tested on a typical operating point, the identification cost of global modeling is reduced and the quality of control is good. The simulation results show the effectiveness of the algorithm.