Research And Application Of Fault Diagnostic Technique On Synchronous Brake System In Heavy-haul Combined Train

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. With distributed, networked, intelligent, modular design features and the fault characteristics of strong coupling, concurrent chains, the synchronous braking system as the basic braking system of the heavy-haul combined train provides a good interdisciplinary application platform for distributed artificial intelligence and fault diagnosis technology.Based on DaQin line 20,000 tons heavy haul combined train and the distributed fault diagnosis technologies, the fault diagnosis, dynamic alliance strategies of the diagnosis task and real-time task scheduling algorithm of the heavy haul combined train synchronized braking system are researched in this paper. The major innovation research achievements include:(1) Based on the analysis of operation principle and fault characteristics of the heavy haul combined train synchronized braking system, a MAS based design scheme with hierarchical structure, which focuses on three issues of the multi-attribute and multi-parameter diagnostic object, such as concurrent fault chain propagation feature, diagnosis task decoupling control, and real-time requirements, is proposed for the diagnosis system of the heavy haul combined train synchronized braking system.(2) For the strong coupling problem of the synchronized braking system, a task allocation strategy based on multi-parameter fuzzy match is proposed. Firstly, according to dynamic changes and superposed operation of the operation mode of synchronized braking system, a double-layer task decomposition strategy based on system functions are established under multiple operation modes. In order to realize the task allocation strategy based on multi-parameter fuzzy match, a multi-dimensional needs vector is proposed to quantify the multi-attribute constraints characteristic of diagnostic task, a dynamic coupling membership matrix between the corresponding tasks of diagnosis agent is build, and a fuzzy statistics membership function is designed to optimize the ability-type matching indicator of task allocation.(3) A reinforcement dynamic alliance strategy based on trust and commitment is proposed for the efficiency optimization and stability problem of the diagnosis task with staged changes in fault diagnosis. First the degree of trust and commitment is used to describe the ability and expected reward of the diagnosis agent in the dynamic alliance; then, an adaptive mechanism based on the feed-forward neural network is employed to realize the dynamic match of the ability-type matching indicator between tasks and agents. And the degree of trust and commitment of the diagnosis agent is updated circularly to improve the work efficiency and stability of the dynamic alliance.(4) To meet the dynamic coupling and strong real-time requirement of the diagnosis task of the synchronized braking system, a dynamic task schedule mechanism based on weight compensation and dependence shifting is proposed. A piecewise continuous dynamic weight compensation algorithm based on discrete weight classifier is proposed to ensure the schedulability of tasks. A coupling intensity map matrix and local oriented function are designed in the dependence shifting dynamic executable controller to reduce the influence of dynamic coupling to system response frequency between tasks. And a local optimal function in a cycle is designed to maximize the optimal function which is targeted at the labor opportunity contributive value and ensure the average occupancy rate of system sources.(5) The air pipe leakage is a major hidden danger to the safety of heavy-haul combined train. In order to solve the pipe problem of the synchronous braking system, a multi-stage leakage propagation air reserve model for brake pipe based on the least square is proposed. Firstly, an adaptive step unstable PCA process monitoring algorithm is proposed to solve the non-stationary and progressive fault of pressure sensors and flow sensors in synchronous braking system. Then, a multi-stage leakage propagation air reserve model is build and the cross-sectional area parameters of limiting valve in the model are optimized dynamically by the least squares parameter estimation algorithm. Finally, in order to realize the fault detection of the pipe leakage for the whole train, the leakage fault of brake pipe of the freight car is detected vehicle by vehicle based on the standard flow detection algorithm in accordance with the composition mode of the heavy haul combined train.The fault diagnosis system proposed in this thesis has been successfully applied in distributed fault diagnosis of the 20,000-ton heavy-haul combined train synchronous braking system on DaQin line, the interactive relationship between system hardware and software models is designed and the on-line operation process of the fault diagnosis system is described in detail. After more than one year of on-line operation, the operation situation indicates that the distributed fault diagnosis system and its relevant algorithm have safe and reliable design, scientific and efficient diagnosis progress, which verifies the system's reliability and stability.