陶瓷窑炉普遍具有纯滞后、大惯性、非线性、时变复杂等特点,其精确数学模型往往很难获取。针对这类系统,本文采用RBF神经网络建立被控对象模型,避免了常规控制算法建立对象精确数学模型的困难。应用动态矩阵预测算法实现对被控系统的预测控制。该控制方法具有很好的动、静态性能和强鲁棒性。以陶瓷窑炉温度为对象,与PID控制进行了比较;仿真结果证明了所提控制方法的有效性。 Ceramic kilns generally have the characteristics of pure hysteresis, high inertia, nonlinearity and time-varying complexity. The accurate mathematical models are often difficult to obtain. For such systems, RBF neural network is used to establish the controlled object model, which avoids the difficulty of the conventional control algorithm to establish the accurate mathematical model of the object. Applying the Dynamic Matrix Prediction Algorithm to Predict the Controlled System. The control method has good dynamic and static performance and strong robustness. The temperature of ceramic kiln was compared with PID control. The simulation results proved the effectiveness of the proposed control method.