数据信息驱动的非高斯随机分布系统抗干扰控制

项目名称： 数据信息驱动的非高斯随机分布系统抗干扰控制

项目编号： No.61473249

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 裔扬

作者单位： 扬州大学

项目金额： 80万元

中文摘要： 项目研究了复杂非高斯随机分布系统数据驱动控制以及相关抗干扰算法的分析和设计问题。包含随机量测噪声、外部干扰、非线性和不确定动态的随机过程广泛存在于工业控制、航空航天等实际系统中，其中含有非高斯变量的随机分布系统往往因为系统工作环境复杂以及缺乏完备的模型信息，已经成为控制界的理论难点和瓶颈。项目提出工程上实用的输出分布泛函和统计信息集合驱动的随机分布控制理论和估计方法，建立新的基于数据信息驱动的反馈控制研究框架。在此基础上设计基于凸优化理论的多目标随机分布复合抗干扰控制方法、基于干扰观测器的模型自由迭代学习抗干扰跟踪控制方法以及同时具有干扰抑制和抵消性能的抗干扰控制和滤波方法。项目解决了模型信息不完备情况下的系统建模问题和多源干扰环境下的高精度控制难题，也为基于数据的控制和决策问题提供新的研究思路。同时将所提理论方法应用到纸张质量控制系统以及火焰燃烧过程中，完成理论算法的有效性和实用性验证。

中文关键词： 随机分布系统；非高斯过程；抗干扰控制；干扰抑制和抵消；数据驱动

英文摘要： This project studies the data-driven control and the analysis and design problem of anti-disturbance algorithm for complex non-Gaussian stochastic distribution systems. The complex stochastic processes with measurable random noises, exogenous disturbances, nonlinear terms and uncertain dynamics exist in many practical applications, such as industrial applications, aeronautics and astronautics, etc. Among these stochastic processes, the non-Gaussian stochastic distribution system has become a theoretical difficulty and bottleneck in the field of automatic control because of more complex external environment and lacking of necessary model information. This project proposes the stochastic distribution control theory and estimation method based on output distribution functional or statistical information sets that is easy to implement in engineering, and constructs a new data-information-driven-based feedback control research framework. Under the theory framework, this project develops the multi-objective stochastic distribution composite anti-disturbance control method with convex optimization theory, proposes the model-free iterative learning anti-disturbance tracking control algorithm based on disturbance observer design and studies the anti-disturbance control and filter method with disturbance attenuation and rejection performance. It is noted that this project not only can successfully solve the modeling problem with incomplete model information and the control problem with high precision for multiple disturbances, but also provide a new research idea for the data-based control and decision problem. Moreover, these proposed methods are applied into paper quality control systems and flame processes in order to test the efficiency and practicality of these theoretical algorithms.

英文关键词： stochastic distribution systems;non-Gaussian processes;anti-disturbance control;disturbance attenuation and rejection;data driven