Research Of Railway Disaster Prevention And Mitigation Distributed Micro-computer Monitoring System Based On Multi-agent

With the rapid development of Chinese economy, China’s railway construction scale andtechnology has improved continuously, especially in recent years, the development of China’srailways rapidly, high-speed, heavy-duty and high-density become an important direction ofChina’s railway development, an east-west, north-south communication, Lunar combinedrailway network has formed a considerable scale and gradually improve. However, the mainartery of the railway as China’s national economy, but every year because of accidents andlosses caused by natural disasters directly or indirectly is very serious, coupled with recentyears, the natural environment has continued deteriorate, the frequency and degree of harmdisasters continue to strengthen, the damage caused by also increasing. In this context, thestudy of natural disasters along the railway line monitoring system to improve the ability ofthe railway disaster prevention and mitigation, so that the railway system can real-timeforecast, monitor and diagnosis natural disaster to ensure the safety of the railway becomesextremely important. However, the efficiency and accuracy of China’s railway system relieson computer monitoring system to get data of natural disasters to early warning, diagnosis andtreatment is still relatively low, has been unable to adapt to the new era of China’s naturalenvironment continues to deteriorate and requirements of railway to improve the speed. So,how to combine the causes and laws of natural disasters and use the advanced monitoring,diagnostic techniques and methods to provide a scientific basis for early warning, diagnosisand treatment of natural disasters along the railway line, which is of great practicalsignificance to protect the safe operation of the railway.Due to the complexity and variability of the natural environment, methods of monitoringand diagnostic based on a single inference mechanisms, can’t meet the real needs, but thenatural disaster monitoring system based on MAS technology can able to integrate differentmonitoring and diagnostic methods to complete tasks of disaster monitoring and diagnostic ofcomplex environment. Therefore, this paper with MAS technology, research and design adistributed computer monitoring system based on MAS railway disaster prevention andmitigation.The monitoring and diagnostic subsystem of this system composed by managementAgent, communications Agent, diagnosis Agent, data acquisition Agent, data analysis Agent,database Agent, rule base Agent and user Agent. Management Agent is responsible for thedecomposition of data, assign diagnostic tasks, and evaluation of diagnostic results;Communication Agent is responsible for managing communication and consultation betweenthe various monitoring and diagnostic systems between Agent and maintaining diagnosticsystem function repository, each new adding diagnostic system Agent only need consider the maintenance and communication with the communication Agent, thereby improving thescalability of the monitoring and diagnosis system; diagnostic Agent combines characteristicsof collected real-time data and historical data and the actual situation to use disaster type ofdisaster-oriented design strategy and composed by winds diagnosis Agent0, sleet diagnosisAgent1, mudslides diagnosis Agent2, seismic diagnosis Agent3and to assess the extent of thedisaster, this is the core part of this subsystem; rule base Agent provide rules for the type ofdisaster, data monitoring and diagnostics and maintain and manage the rule base, and provideusers to view and modify the rules of the function; data acquisition Agent and database Agentare responsible for obtaining and maintenance the real-time collected data; user Agent isresponsible for maintaining man-machine interface.For the complexity of natural environmental changes, this paper designs Agent-orientedfuzzy algorithm and neural network hybrid diagnostic strategies. Initial diagnosis using fuzzyalgorithm to obtain disaster type of a large membership types, accurate diagnosis using neuralnetworks to calculate the severity of the resulting disaster of initial diagnosis, through the useof hybrid diagnostic strategies can effectively improve the accuracy of diagnostic systems.Paper designed the monitor and diagnose system workflow: data acquisition Agentreceives sensor collected real-time data and then send them to the management Agent andproposed data monitoring and diagnostic requests, manage Agent receives the monitoring anddiagnosis requests and determine data type and initiate the appropriate data analysis Agent,when data analysis Agent obtain data, preliminary diagnosis and analyze collected data firstly,if the results of the data analysis and diagnosis beyond safe limits, alarms and enable thecorresponding position video monitoring equipment and reproduction of the scene inreal-time and accurate of the disaster, while the data upload diagnosis Agent, throughdiagnostic Agent make data disaster diagnosis, and then use neural network to accuratediagnosis type of a large disasters membership of preliminary diagnosis and finally getdiagnostic results and send back to the management Agent for evaluation of diagnosticresults.Finally, the paper by case sample data to simulate the natural environment along therailway, simulation results show that natural disasters Agent-oriented hybrid diagnosticstrategies and MAS-based railway disaster prevention monitoring system can be bettercompleted early warning, monitoring and diagnostic tasks.