Research On Traffic Dispatching System Human Error Prediction For The Subway Emergency

ABSTRACT:In recent years, with the development of the subway train operation control system technology, the degree of automation of traffic dispatching system is higher and higher, which reduces the probability of accident caused by the traffic dispatcher human error. But at the same time, the problem of accident or consequence deterioration leading by the traffic dispatcher human error in emergency is increasing. Traffic dispatching system is an important guarantee of the subway safety, while the traffic dispatcher is the system’s ultimate protection and recovery mechanism under the emergency. How to effectively avoid and reduce the subway traffic dispatching system human error under emergency is an urgently security problem for the world subway companies. Therefore, relying on the National High Technology Research and Development Program863of ’Integrated Optimizing Control Technology for Train Operation’, this thesis analyzes the subway traffic dispatching system human error causing, and proposes two key research aspects of human error behavior risk prediction and human error forcing factor prediction. Thereby, the scenario based emergency human error behavior identification technique, prediction indexes grading technique and human error risk assessment technique are constructed. At the same time, the human error forcing factor risk prediction technique based on the causal relations between factors is buit, and the normalized human error data collecting and management method based on the proposed human error prediction technique is also built to provide the theory method and technical support for subway traffic dispatching system human error prevention and reduction. Research on the problem of traffic dispatching system human error prediction has important theoretical and actual application value.Firstly, a subway traffic dispatcher human error mechanism model is constructed based on the detail analysis of the emergency processing task and the human interaction features of the subway traffic dispatching system, which describes traffic dispatcher human error mechanism in two aspects of human error behavior and human error forcing factors. Then, a human error mode analysis model and a human error forcing factors identification model are built based on the hierarchical task analysis method and the m-SHEL model respectively, and the related traffic dispatcher human error mode classification framework and the human error forcing factors optional library are also developed by modifying the multi-resource theory and Jae W. Kim’s human error forcing factors set respectively. Furthermore, the created models and classification framework’s validation are tested and verified by grey correlation analysis, data mining and intra-rate reliability of98emergency human error analysis reports of1997-2011from a domestic subway company.In order to solve the problem that the subway traffic system human error probability is difficult to calculate, estimate and predict both in direct and indirect way because of the lack of human error data, a human error behavior risk prediction model is proposed by referring to the failure mode, effects and criticality analysis technique, which estimates the human behavior risk based on the three aspects of human error behavior possibility, restorability and consequences seriousness. Thereby, the human error behavior possibility grading standard is determined according to the traffic dispatching system human error mode classification framework, while the human error behavior restorability grading standard is built by considering the human error barrier character, and the human error consequences seriousness grading standard is established with the subway accident management regulations. Also, the measurement figure of criticality matrix and risk assessment matrix are constructed for human error behavior risk calculating and grading.Considering the requirement of human error behavior risk prediction model’s quantification, the human error behavior possibility and restorability quantification grading standard are built referring to the failure mode probability grading standard and traffic dispatching system human error barrier’s error probability reduction degree. As a support, the difficulty of quantification human error data collection is also solved by introducing Beta Distribution into human error data collection and estimation method. And an ATS experiment simulator based on the traffic dispatcher’s typical tasks is designed and developed to collect the qualification human error data by the eye tracking technique. The experimental result verifies the rationality of the data collection method and provids the idea for the subway traffic dispatching system quantification data collection.Based on the established human behavior risk prediction technique, the emergency scenario human error risk prediction method is studied. In this study, the emergency human error scenario generation technique is created according to the modularized characteristics of traffic dispatcher emergency processing task, and the emergency scenario human error behavior consequences seriousness, restoration, possibility, risk degree calculation rules and key human error behavior identification technique are analyzed and identified on the basis of the human error scenario building rules. Additionally, an example of the contact rail power off emergency human error risk prediction is used to verify the practical applicability of the structured emergency scenario human error risk prediction method.In the aspect of human error forcing context risk prediction, the causal relations between human error forcing factors and the changes of the human error forcing context comprehensive and the overall effect caused by the causal relations are taken into full account. By the time, the human error forcing context graph model described by the fuzzy cognitive map is built. Consequently, the human error forcing context effect assessment technique is formed by introducing the fuzzy cognitive map’s research achievement in causality reasoning and factors weight calculating, and is used to complete the example analysis of a subway company’s traffic dispatching system human error forcing context risk prediction. In order to ensure the constructed graph model’s validity, evidence theory and inconsistency judgment method are utilized to synthesize and evaluate the experts judgment. Meanwhile, a small sample data screening technique is introdued to improve the graph model construction efficiency.Finally, the subway traffic dispatching system human error data requirements and uncertainty are analyzed in detail according to the structured human error prediction technique characteristics of traffic dispatching system. Then, a normalized human error data collecting method is established.