基于分散协同通信的安全生产机能维持控制研究

Actuator fault exits in networked process industry, such as shaft furnace and thermal power plant, which could cause serious safety problem if no proper action is taken in time.In this project, an adaptively functional maintenance based safe control technology is proposed for a twin-tank process fault occurs to trade off the safe operation and small loss. It is worth to maintaining at least some of function while keeping safety of the fault system. But,it is difficult to realize it because a control system with input constraint needs to be generated by analyzing the relationship among fault information, functional maintaince level and input constraint level and its robust control be solved. Thus, operator based robust right comprime factorization approach is presented to design fault observer firstly such that the main fault information is arrived. Secondly,support vector regression based functional maintenance model is built to get the functional maintenance based control system with input constraint. A shape variable function is considered as kernel function to design the support vector regression for modeling funcitonal maintenance evaluation by using the fault information. Meaningwhile, a fourth-order even moments method parameter estimation is discussed to estimate model parameters, such as shape parameter ,to get optimal model, and a fast support vector regression training is proposed by using edge detection.Then,operator based sliding mode variable structure right coprime control is proposed to realize the adaptively functional maintenance based control scheme with input constraint. A sliding mode hyperplane is designed to reconstruct the controllers and its realization is guaranteed by isomorphism. Also, the definition of contranction map theorem is extended to deal with the coupling effects of the MIMO nonlinear systems. Finally, cooperative communication based distributed control system is set up to ensure fault localization.

在竖炉、火力发电厂等流程工业中，执行器故障常导致产品质量下降甚至引发事故。针对故障系统安全与损失之间的矛盾，本课题拟进行基于分散协同通信的安全生产机能维持控制方法研究，把安全控制问题归结为生产机能适度维持的输入受限系统的鲁棒控制问题并解决。首先，应用基于演算子理论的右互质分解方法，设计抗有界未知扰动的故障观测器来实现故障状态的在线监测。其次，拟建立基于支持向量回归的生产机能维持度模型，生成机能适度维持的输入受限系统。在建模中，拟设计形状可变核函数，利用四阶矩估计法实现参数在线估计，并用边缘检测法改进模型在线训练速度。接着，拟提出基于演算子理论的滑模变结构右互质分解方法来实现输入受限系统的鲁棒稳定控制。设计平滑切换超平面使重构控制器在超平面内平稳滑动；使用压缩映射原理弱化多输入多输出之间的耦合影响；利用同构理论保证控制器的可实现性。最后，构建故障分散协同控制系统保证故障局部化和控制协同化。

鉴于执行器故障常导致产品质量下降甚至引发事故，针对故障系统安全与损失之间的矛盾，本项目研究了基于分散协同通信的安全生产机能维持控制方法。本项目首先建立了基于分散协同通信的串联双水槽执行机构故障控制系统。接着，应用基于演算子理论的右互质分解方法，设计抗有界未知扰动的故障观测器来实现故障状态的在线监测。其次，建立基于支持向量回归的生产机能维持度模型，生成机能适度维持的输入受限系统，把安全控制问题归结为生产机能适度维持的输入受限系统的鲁棒控制问题并解决。在建模中，设计形状可变广义高斯概率密度函数作为核函数，并利用四阶矩估计法实现参数在线估计。然后，针对得到的输入受限系统，提出基于演算子理论的滑模变结构右互质分解方法来实现其鲁棒稳定控制，其中设计了平滑切换超平面使重构控制器在超平面内平稳滑动。最后，考虑网络系统的跟踪控制问题和故障状态下的安全控制问题，针对网络控制系统的跟踪性问题提出了基于CGT的自适应模型输出跟踪控制方法；对于故障状态下的安全控制问题，提出把执行器故障问题转化为输入受限的约束优化问题进行求解。本项目的研究对故障系统在维持适当性能下的安全控制具有重要应用价值。