基于小波-卡尔曼滤波的二维离散随机系统鲁棒H∞控制

In the traditional robust controller design, there is no use of the noise characteristics, nor weaken or even eliminate the output signal noise, the.criterion is conservative. In this project, two-dimensional random Robust H-infinity controller is designed based on wavelet-kalman filter. First, Using the variational technology and optimized design ideas, we translate the wavelet-kalman filter design problem in the background of robust H-infinity controllers into a multi-variable optimization problem. And using an alternating split iterative algorithm, design two-dimensional wavelet-kalman filter with the low computational complexity and greater stability margin. Then, random robust H-infinity output.feedback controller design method is presented with constraints of wavelet-kalman filter output characteristics. Based on entropy index function, dual objective coordination optimize mechanism between filters and controllers are established to achieve optimal control of the controller. Finally, in view of the hot-rolled laminar cooling control system with multiplicative noise (quantization error) and additive noise (measurement noise) is uncertain complex two-dimensional (strip length and thickness, length and width of the strip) dynamic process, we.intends to apply research results to laminar cooling coiling temperature control system. Design heat rolling laminar cooling controller to improve the uniformity and hit rate coiling temperature.

传统的鲁棒控制器设计没有利用噪声特性，也没有减弱/消除输出信号中的噪声，导致得到的设计判据有一定的保守性。因此，拟研究基于小波卡尔曼滤波的二维离散随机鲁棒H∞控制器设计方法，降低设计判据的保守性。首先，从变分技术和优化的设计思路出发，将以鲁棒H∞控制为背景的小波卡尔曼滤波器设计转化为多变量优化问题，利用多尺度分析和分裂迭代算法求解，设计具有较低复杂度和较大稳定裕度的小波卡尔曼滤波器；其次，以小波卡尔曼滤波器输出特性为约束，提出带约束的随机鲁棒H∞控制器设计方法；再次，基于熵型指标函数建立滤波器和控制器的双目标协调优化机制，进一步优化控制器性能，降低控制器保守性，实现系统最优控制；最后，鉴于热轧层流冷却控制系统是含乘性噪声和加性噪声的不确定复杂二维系统，拟将研究成果应用到该系统，给出复杂网络环境下的热轧层流冷却控制器设计方法，提高卷取温度的分布均匀性和命中率。