| With the continuing advance of modern industry, the scale of production graduallybecomes larger. for the control system, the function and structure get increasingly com-plex, the component units may be distributed over a vast geographical area, and the num-ber of the closed control loops climb sharply, such that the large-scale complex industrycontrol system emerges as required. For these large systems, Centralized Control makesthe information exchange difcult and less reliable in the whole systems. On the otherhand, distributed control systems based on wired communication networks still face somedevelopment bottlenecks such as high wiring, installation and maintenance costs, com-plex integration, poor scalability, etc. The traditional control methods have been alreadyunable to meet the demand of modularity, real-time control, ubiquitous detection, faultdiagnosis and safety management for the large-scale complex industry control systems.After entering the21st century, with the progress of sensing, microelectronics, communi-cation and information technology, low-cost, low-power, high-performance, miniaturizedembedded intelligent terminal equipments appear. A large number of intelligent terminaldevices connected via a low-cost, low-power, self-organization, and easy-deployed wirelesscommunications network can share and exchange information to achieve the deep inte-gration of the physical world and the information word, and then derives out the wirelessnetwork control system (WNCS) which is a new front research direction of cyber-physicalsystems (CPS) and has received extensive attention in the academic and engineer circles.Due to the involvement of the wireless network, the quality of performance (QoP) ofthe WNCN depends not only on the design of the controller, but also on the remote qualityof service (QoS) of wireless communication network. Thus, we will need to generatea co-design framework for WNCS based on the new theories and new technologies inrelated multidisciplinary felds and performance trade-ofs from wireless communications,network scheduling, network complexity, and some other factors that may afect theperformance of control. Combined with the latest research results at home and abroad，based on the emerging industrial wireless technology standards, this thesis investigatesthe modeling, stability analysis, distributed estimation fusion and robust control synthesisproblems for WNCN. The main work and contributions are summarized in the followingfour aspects.1) Considering the harsh industrial environment, under the two-tier hybrid networkarchitecture of industrial wireless standard WIA-PA, a new control-oriented distributed networked estimation and cooperative control algorithm is presented based on the inter-acting dual model (IDM) channel-aware technology and the federated multi-sources flterfusion theory.2) By the LMI method, a new fully distributed nonlinear robust optimal controlstrategy is developed for a class of nonlinear system modeled by standard neural networkmodel (SNNM) based on a multi-hop wireless neural control network (WNCN). Thewireless communication links in the WNCN are modeled by the fading channels. themesh structure of network topology and TDMA-based network scheduling mechanismused in WNCN are compatible with the Wireless HART.3) Considering an example of practical industrial application of neural network intel-ligent control. On the basis of deep study of the principles of multi-speed electronic let-of, an intelligent multi-speed electronic let-of system based on fuzzy immune single neuronadaptive PID (FI-SNAPID) algorithm for German Karl Mayer two-needle bar warp knit-ting machine. The detailed hardware and software design solutions are proposed. Thetest results of the prototype show the feasibility and correctness of the system design.4) We study the online distributed adaptive control problem of WNCN for the non-linear autoregressive moving averaging (NARMA) MIMO model using dynamic backpropagation (DBP) algorithm. |
PDF Full Text Request