Guoqiang Cai

Approach of Train Group Operation Conflict Prevention Reasoning Based on Fuzzy Time Petri Nets

Avoiding conflicting between large numbers of trains is a key problem to train scheduling services in busy rail network. Verifying train group operation conflict within and between multi-station satisfying time constraints is the basis of the construction and coordination on train schedules and plans. While, for many rail networks, the conflicts of proposed train times, lines, and platforms are checked and resolved difficultly. This paper proposes an approach to construct a sprouting graph of train traffic control system based on fuzzy time Petri nets considering safety conditions called STPNS. The sprouting graph is used to check the consistency of temporal knowledge and perform safety reasoning. The temporal knowledge reasoning method efficiently describes time constraint and safety events happening simultaneously. Applications show the approach has good structure properties and good means in safety analysis model analysis and validation.

A novel fault diagnosis expert system knowledge acquisition method of metro vehicle equipments

This paper is to propose a fault knowledge acquisition method for metro vehicles fault diagnosis at present, the lack of metro vehicles fault diagnosis knowledge and the difficulty of acquiring the fault diagnosis knowledge hinder the development of fault diagnosis for metro vehicles. On the basis of profounding analysis of the mechanism for the forming of metro vehicle faults, this method is to design a new fault diagnosis models and simulations. Then simulating results are to comprehensively analyzed and transformed into fault diagnosis knowledge available for the expert system. Its proved by application that this method improves the reliability of fault information acquisition, which attaches great importance in fault knowledge acquisition for complicated equipment, especially for equipment of metro vehicles.

Train ticket predictive analysis based on decision tree induction

According to the requirements of decision analyses and the limitations of current prediction methods in China railway train ticket system (CRTTS), a novel method TTDTPA, which is based on decision tree induction, is presented. TTDTPA extracts a kind of instructive rules that collect the advantages both prediction and statistic, therefore it is suitable for supporting multi-level requirements of the decision-makers for predictive analysis in CRTTS. In this paper, firstly, the authors describe TTDTPA, and then through the detailed analysis on the application about passenger seat type, we prove the efficiency and the validity of TTDTPA in CRTTS. Finally this paper concludes the predictive analysis problems that is faced in CRTTS.

A Multi-grade evaluation model for traffic level of service

Regional traffic level of service (LOS) evaluation is a sticking point in urban traffic research. The objective of this study is to develop a model for estimating LOS of regional traffic. Firstly, we presented one method applied in the micro-layer traffic. Secondly, one macro-layer method was constructed, it based on the result of micro-layer. We combine these two layers into a single Multi-grade evaluation model for traffic level of service. The multi-grade evaluation model owns two layers including the micro-layer, which is used to distinguish the segment traffic status, the macro-layer, which is to estimate regional traffic status. The proposed model is applied to several city areas, and the results show its validity.

Research on Dimension Tree Network and its application in Jing-Hu Line Safety Monitoring System

Safety detect information fusion method is the key problem of railway safety monitoring and early-alarm system. This paper proposes a kind of multi dimension information fusion model named dimension tree network. This model solves crucial miscellaneous information fusion problem in three factors. First, it adopts tree type structure to improve scalability and reconstruction Secondly, it can be used as a kind of specification of data interface, which is used to solve large-scale and real-time numerous kind information integrating problem. Finally, node assemble operation is helpful to formalize information storage and manage structure. DTN is applied in Jing-Hu Line Driving Safety Monitoring System and achieve better effect.

Application of Quantitative Analysis Method in Traffic Volume Prediction

Predictive analysis about China railway parcel is an urgent task for railway intelligent transportation system (RITS). This paper proposes a novel parcel traffic volume predictive method named FCP (fuzzy cell prediction), which provides an effective means to build strategic models and enhances the predictive ability of interactive and hierarchical analysis. The key components of FCP are traffic data process mode which map variable to infinite spaces. The variable multiple correlation problem are solved. Applications to maximize transport capacity suggest that FCP is a contributive and powerful tool for the RITS.

Further Research and Application of Train Interconnecting Network System

This paper introduces the buildup process of train interconnecting network base on CAN bus and the process that realizes multiple locomotives network’s dynamically buildup. The paper first introduces the functions and then described the inauguration and regular process of the interconnecting network, analyzes its inauguration arithmetic and describes the process of using object Petri net (OPN). Finally, this paper checks the feasibility and stability designed above and gives further applications expectation.

Research on relative error rate analysis method between the Earth surface safety monitoring devices of vehicle operation estate

Establishing an accurate and reliable network for the Earth surface safety monitoring devices of vehicle operation estate is the key factor to the railway intelligent traffic system in order to avoid incidents caused by super load of wagons. It is critical to seek an efficient analysis method about relative error rate analysis between devices so that designing technique of device can be checked. This paper presents a novel relative error rate analysis method DVM (distribution-oriented, verify and match). DVM includes two parts: distribution-model-oriented sample, verify and match of multi-system. Distribution-model-oriented sample can overcome defect in current sample methods. The maximum entropy theory is applied and distribution model of relative error rate is concluded. A verifiable match algorithm of multi-system realizes verify and match among the detective information, automatic identify system, and QB system. Application shows that distribution model of relative error rate can be obtained accurately according to DVM method.