Research On Distributed Diagnosis With Overlapping Decomposition For Pipeline Leakages In Heavy-Haul Combined Trains

As representation of advanced level and important development direction of world freight transportation nowadays, heavy haul combined train is the main pattern of coal freight transportation and provides strong support for the rapid development of national economy in China. The pipeline leakages of braking system results in the abnormal change of braking pressure and threatens the safe operation of train. As the pipeline of heavy haul combined train is physically distributed, this thesis focuses on insuring the operation safe by researching on model of pipeline and distributed diagnosis of pipeline leakages. The main works are as follows:Firstly, according to the structure and characteristics of the braking system of the heavy-haul combined train, we divide the problem of pipelines leakage into two kinds:multi-pipeline leakage of locomotive and pipelines leakage of vehicles, for which the reasons for leakages and diagnosis method are analyzed separately. As a result, an general diagnosis architecture for pipeline leakages in heavy haul train is established. As for the vehicle braking system, a model-based fault diagnosis scheme is proposed. By introducing the air flow theory, mathematical model of the brake system under three different work modes is built based on the model of basic components.Secondly, to solve the problem that the vehicle braking system is widely distributed and the coupling is strong, the distributed fault diagnosis method with overlapping decomposition is proposed. The whole braking system is modeled as the interconnection of several subsystems. The subsystems are allowed to overlap, thus sharing some state information. For each subsystem, a Local Fault Detector is designed based on the measured local state of the subsystems as well as the transmitted variables of neighboring states that defines the subsystem interconnections. The local detection decision is made on the basis of the knowledge of the local subsystems dynamic model and of an adaptive approximation of the interconnection with neighboring subsystems. The use of a specially-designed consensus-based estimator is proposed in order to improve the detestability of faults affecting variables shared among different subsystems.Lastly, based on the overall structure of the system, the hardware platform and software architecture are outlined. The performance of the system operation is analyzed and the result shows the practicality of the system.