Search

In view of that Sync brake system of combined-train has a complex configuration and distributed control characteristic, an on-line diagnosis system based on multi-agent technique is designed. The diagnosis system is a two-level multi-agent system, in which diagnosis tasks are divided into simple tasks and accomplished by agents with collaboration. Based on the design of the diagnosis system, the key problems such as the mechanism of task decomposition as well as agent collaboration are resolved. The application results show the validity of this diagnosis method.

Multifunction vehicle bus (MVB) is a standard protocol defined in train communication network criterion, and used for transmitting information among various equipments inside the vehicle. In order to insure compatibility and inter-operability among the MVB equipment, it requires that the data transmitted in the bus need to be collected, and correctness and reliability of data transmission of the MVB network should be analyzed. Base on the characters of periodically broadcasting and master slave frame response. According to IEC61375 standard, this paper proposes a method to implement an analyzer for MVB device protocol. In the method, the transmitted data on the bus are sampled using error-tolerance decode algorithm, and then the compatibility and dependability of data transmission among devices are analyzed via protocol analysis application. The experiment results showed that the analyzer can exactly validate the design of the equipment on the MVB bus.

Speed identification is crucial in speed sensorless induction motor drives. Because the voltage model involves pure integrator, which will cause initial value and drift problems. These problems and resistance varies effect the precision of speed identification. This paper presents an improved method, which uses high-pass filter to avoid these problems and adapt the cut-off frequency on-line to improve dynamic performance by using fuzzy controller; and then eliminate the influence of stator resistance varies by using fuzzy controller to estimate and adapt stator resistance on-line. Simulation results indicate that the method has the characteristic of good dynamic and static performance and high precision speed identification.

By analyzing the existence of unstable regions of the present adaptive observer using pole-placement technique in regenerating mode at low speeds, conditions that ensure the adaptive observer stability at full speed regions are got. An improved speed adaptive law and a novel method to choose the gain of the observer are proposed based on observer. By using the Lyapunov stability theory, speed identification adaptive law is got. The gain of the observer can be obtained by solving two bilinear matrix inequalities. The simulation model of speed sensorless induction motor using direct torque control based on observer is built in the Matlab/Simulink. Simulation results show that the adaptive observer has good steady and dynamic performances and good robustness.