Algorithm Theory Research On Safety Monitoring And Early Warning For The Running Gears Of Railway Vehicles

As peoples’ demand of safety and comfort of railway vehicles operating become higher and higher, technology of railway vehicle running status monitoring and security early warning constantly development and be taken seriously. Real-time, fast vehicle state recognition method is the research trend of the future.Based on a review of previous research on the vibration monitoring, fault diagnosis, and safety warning for rail cars, this paper summarized the widely used fault warning methods and the characteristic vectors involved. Then two types of self-learning algorithm for real-time safety monitoring and warning was creatively developed through investigating the vibration signals from the running gear of rail car, Algorithms will be able to use some algorithms that can be used in traditional mechanical fault diagnosis areas but difficult to play in the environment of track irregularity, the fuzzy clustering algorithm and neural network algorithm. Combined with self-learning algorithm, these new algorithms can be used to overcome the difficulty of using parameter limits in the same standard to examine the mechanical performance of a running gear in complex environments. Later, an algorithm-centered information platform was established. This platform is capable of monitoring vibration signals from the running gears of rail cars and warning of abnormal vibrations promptly. Moreover, this platform enables identification of more details on faults in later data analysis, providing useful guidance on car maintenance for safer driving.Simulation and analysis were conducted to verify the feasibility of the proposed algorithm. First, the vertical vibration characteristics of rail cars were investigated, analyzing the variations in the characteristic vibration parameters of the rail car under different road conditions, working conditions and failure conditions. Later, vibration tests were performed in the vibration test platform to collect data on the characteristic parameter variations under the simulated track spectrum. The collected data was then analyzed to comprehensively examine the feasibility and effectiveness of the proposed algorithm.The aforementioned test platform, a vertical vibration simulation platform applicable to rail cars, is a universal mechanical vibration platform widely used in research on vibration characteristics. This platform is comprised of data acquisition equipment, acceleration sensor, power amplifier, signal conditioner, NI data acquisition card, and electromagnetic vibrator.After the hardware system(i.e. the test platform) was created, the warning system’s software responsible for data acquisition, processing, analysis, display, and management was designed, primarily in the C# programming language. The software system consists of the following major modules: Primary Display Interface, Signal Generator, Self-learning Warning System, Neural-network Fault Diagnosis, and Historical Data Management. The core warning algorithm exists in Self-learning Warning System and Neural-network Fault Diagnosis. In addition, the software system also consists of other key modules, including Track Spectrum Generator and Data Acquisition and Processing.In this study, a creative self-learning safety warning algorithm was designed using knowledge about fuzzy mathematics and artificial neural networks. This algorithm was then successfully implemented in the C# programming language, achieving the development of the visual safety monitoring information management platform. Later, a lot of tests and debugging were carried out in the vibration test platform in order to improve the safety warning algorithm. The results demonstrated that this algorithm is feasible, effective, and more sensitive and accurate than previous warning algorithms.Finally, make a summary of the algorithm and the experimental results in this paper and have an outlook for further research.