Fault Diagnosis For DK-2Brake Based On Robust Adaptive Observer

Abstract:DK-2brake is a new generation of electronically controlled air brake, which is self-developed by our country independently. It is a key control equipment of the train braking system, and has been widely used in the heavy-haul combination trains running on Shenshuo and Shuohuang railways. The reliable operation and fault diagnosis of DK-2brake are hotspot issues in current researches. In our paper, the system function models of DK-2brake are established, and a robust adaptive observer based diagnosis method is adopted to overcome the influence of system disturbance and modeling uncertainty. It is realized to detect and isolation fault for the DK-2brake accurately. The main works are as follows:Firstly, the phase switches and overlay characteristics of its operating mode are analyzed on the basis of the air channel structure and the working principle of DK-2brake. Through the analysis of fault classification, fault features and diagnostic difficulties, and considering the influence of modeling uncertainty and disturbance to the diagnostic accuracy, a robust adaptive observer based fault diagnosis scheme is proposed.Secondly, to address the modeling difficulty brought by large number of components, complicated conditions and multiple work modes, the basic mathematical models of pneumatic module such as control valve, relay valve, high speed solenoid valve are established based on the flow equations, adiabatic process equations and ideal gas state equations. Then according to the air channel structure and coupling relationship between modules of DK-2brake, the mathematical models of automatic braking and independent braking function can be derived through mechanism analysis method, the accuracy of the system models are verified by contrasting the simulation results and actual experimental data.Thirdly, as the DK-2brake often runs in the severe environments suffer from frequent impacts and vibrations, the brake system can be easily influenced by external disturbance and model uncertainty. Therefore, a robust adaptive diagnosis observer is designed on the basis of the established system model, in which a special coordinate basis (SCB) transformation method is used to decouple external disturbance and generate the robust residuals, and then time-varying adaptive thresholds are designed based on the model uncertainty to realize the accurate detection. The fault detection probability can be ensured and the miss probability can be effectively reduced in the interference environment. For each state variable, an independent residual is constructed and compared with the adaptive threshold to get the corresponding indication vector. Then the indication vector is compared with the fault features table to realize the fault isolation. We use the Matlab simulation tool to verify the effectiveness of the proposed methods.Finally, according to the framework of diagnosis system, the hardware and software implementation of fault diagnosis system for DK-2brake are given by using a hierarchical component-based structure. The diagnosis system has been used for DK2brake installed on the new Shenhua eight-shaft electric locomotive. The practicability of this proposed diagnosis system is validated by the actual diagnosis results of train’s online running.