Proceedings of the 4th World Congress on Intelligent Control and Automation (Cat. No.02EX527)

In this note, we propose lower and upper matrix bounds for the positive semidefinite solution of the unified coupled algebraic Riccati equation (UCARE). Using the results of this note, computational algorithms to solve the UCARE can be established.

A relay feedback identification auto-tuning of PID/PI controller is proposed in this paper, which is capable of controlling long time-delay processes. The real process is approximated via FOPDT or SOPDT models, whose parameters are specified through a modified relay feedback identification method. Applying the zero-pole cancellation principle, the PID/PI controller is tuned according to the specified amplitude and phase margins. Model identification and controller parameter tuning are done online without much influence on the normal operation. This algorithm also has many other advantages such as a fast setpoint response, strong robustness and good disturbance rejection ability.

We construct a simulation system for a live line maintenance robot using virtual reality. The simulation system uses the same input devices and controllers as a real robot system. It provides functions of operator training, human-machine intelligent control simulation, and robotic trajectory simulation. Those simulation data can activate the real robot. The simulation system framework and modeling methods of the virtual world are described in the paper. An experimental simulation result shows the validity of the proposed system.

In the field of industrial automation, fieldbus is changing the traditional DCS's important position and the architecture of distributed computer control system. However, it is not the time that fieldbus is popularized widely in China because of the factors of economy, technology and industrial automatic level. Therefore, this paper presents a mixed distributed computer control system that is composed of fieldbus and RS485 communication protocol. Through fieldbus and RS485, process parameters and status information of field devices are delivered to monitoring computer. In addition, the information integration between control system and management system is achieved using ODBC technology. This system has already been implemented on the product line in one large-scale tobacco group successfully. The performance results show that the system exerts the advantages of both Fieldbus and traditional data bus adequately, and has many advantages, such as openness, reliability and high ratio of capability-to-price.

Aimed at a multi-channel loading system, a fuzzy model based predictive control with multi-step linearization scheme is proposed. Using a Takagi-Sugeno fuzzy model as a predictive model of the loading system, a linearized fuzzy model with multi-step, the predictive control is applied to the loading system. The result of tests show that this scheme can effectively overcome the unknown influence of nonlinearities and turbulence, it has high control precision and robustness, and other unknown systems can use it for reference.

This paper presents a new structure of an intelligent monitoring system based on a fieldbus control system (FCS) for the marine engine room. An adaptive optimal predictor based on a diagonal recurrent neural network (DRNN) is proposed for trend analysis. The system structure, predictor model and learning algorithm are described in this paper. An application result in the marine engine room is illustrated.

In this paper, variable-structure recursive algorithms and a structure-adaptive self-tuning recursive predictor are given for time-varying structure systems. They are effectively applied for the prediction of oil productions.

The paper introduces a fuzzy controller based on an in-system programmable logic circuit. The fuzzy transformation of accurate input data, fuzzy control rules based on observing the control action and the user's experience, fuzzy decision of output can be carried out by the programmable device. Owing to using VHDL language to write fuzzy rules, the characteristic of this controller is that the fuzzy rules can be modified at any moment even if working. The paper gives the controller's working principle and the source code of VHDL.

A kind of classifier systems based on genetic algorithms (GAs) is proposed. In the process of constructing the classifier system, the method uses GAs to adjust weights of all object attributes in all categories and membership functions of all categories synthetically to make them become harmonious and consistent. Finally, an experiment is conducted to demonstrate that this system is effective and practical.

Một thuật toán di truyền lai, kết hợp một lộ trình tìm kiếm cục bộ dựa trên gradient vào thuật toán di truyền phân nhánh, được đề xuất để giải quyết các bài toán tối ưu hóa liên tục. Thuật toán tối ưu hóa này được áp dụng cho ba hàm phi tuyến, mỗi hàm có thể lên tới 100 biến và đa cực tiểu. Kết quả thử nghiệm cho thấy so với một thuật toán phân nhánh tiêu chuẩn, sự kết hợp giữa tìm kiếm dựa trên gradient và phân nhánh cải thiện độ chính xác tìm kiếm lên nhiều bậc và khả năng xác định cực tiểu toàn cục được cải thiện rõ rệt.