Are you Yingze Yang?

Claim your profile

Publications (6)0 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper, the fault detection and isolation problems (FDI) of heavy-haul trains equipped with electronically controlled pneumatic (ECP) brake system are studied. A longitudinal dynamical model and the fault modes of trains are considered. A fault detection estimator bases on the nonlinear observer is designed to generate residual for detecting fault. After the fault is detected, a group of fault isolation estimators with different residuals and thresholds can be designed, each estimator corresponds to one possible actuator fault of trains. And in each fault isolation estimator, the unknown actuator fault of trains can be estimate simultaneously by introduce a learning algorithm. A simulation analysis is proposed to show the validity of the designed FDI scheme, by using the parameters from the heavy-haul trains system running on Datong-Qinhuangdao railway in China.
    2014 26th Chinese Control And Decision Conference (CCDC); 05/2014
  • [Show abstract] [Hide abstract]
    ABSTRACT: Due to various operation modes, complex working environment, and various components with strong coupling in CCBII braking system, multiple incipient faults may happen simultaneously. It is difficult to carry out the multiple incipient faults prognosis for the CCBII braking system using the traditional faults prognosis techniques. In this article, a model-based fault prognosis approach is used for CCBII braking system with multiple incipient faults. Firstly, the models of the working modes are established, the latest concept of Augmented Global Analytical Redundancy Relations (AGARRs) is used to consider parametric and nonparametric incipient fault prediction. Before the faults are detected, mode tracking is realized by comparing Mode-change Signature Matrix (MCSM) with the coherence vector. Once the mode tracking fails, the fault hypothesis set, including suspected faults and suspected mode change, is established correspondingly. The degradation behavior of incipient fault is unknown in advanced and the unknown degradation characteristics of each incipient fault can be predicted with some matching dynamic models. Furthermore, the degradation models and incipient faults are identified using a mixed differential evolution (MDE) algorithm. The real-time online faults prognosis of CCBII braking system is realized. Finally, simulation tests validate the efficiency of the proposed scheme.
    2014 26th Chinese Control And Decision Conference (CCDC); 05/2014
  • [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents a robust adaptive fault estimation design method to solve the problem of actuator fault estimation for CCBII Braking System with disturbances and model uncertainties. The proposed robust adaptive fault estimator which is based on an adaptive observer can improve the speed of fault estimation. By transforming the system model into a special coordinate basis (SCB) form, we can structure the desired observer. This transformation can achieve a decoupling between the fault estimates and the disturbances. As a result, the conventional adaptive observer is improved to enhance the speed of fault estimation through this decoupling. Simulation experiments of CCBII braking system are given to validate the effectiveness of the fault detection method.
    2014 26th Chinese Control And Decision Conference (CCDC); 05/2014
  • Yingze Yang, Weihua Gui, Zhiwu Huang
    [Show abstract] [Hide abstract]
    ABSTRACT: For a certain real-time task scheduling problems with characteristics of uncertain task sets, the executing time dynamic changing and the task's mutual coupling strength dynamic changing, a multi-tasking dynamic scheduling algorithm is proposed based on dynamic weight value compensation and adaptive task jumping, which solved the loads storming caused by the dynamic task scheduling as well as the too long system response time caused by the changes of the tasks' coupling coefficient. The scheduling diagram and function of the modules are given firstly. Then, the focuses are the implementation of the execution controller and the weight compensation method based on the research of task weight value, the weight compensation and weight classification. The reliability and efficiency of the algorithm is verified by comparing tests, and has been applied to the combining heavy-duty train's braking condition monitoring and early warning system.
    Networking and Digital Society, International Conference on. 01/2009; 2:229-232.
  • Yingze Yang, Weihua Gui, Zhiwu Huang
    [Show abstract] [Hide abstract]
    ABSTRACT: In view of design characters of CCBII brake, a state monitor system based on multi-hierarchy evaluation architecture and fuzzy evaluation method is given. Block diagram of brake state monitor system is given, and design features and functions of each part are described, design methods of factor set, safety degree set and itpsilas criterion in multi-hierarchy evaluation system are discussed. Key points in multi-hierarchy fuzzy evaluation method are deduced essentially, such as diagnosis matrix based on weight set, subordinate degree matrix and multi-hierarchy fuzzy evaluation matrix based on weighted average model. Finally the whole CCBII Brake performance evaluation results are given. The application result shows that the system fits the design features of CCBII brake very well and is available for improve the operational efficiency in depot maintenance.
    CSIE 2009, 2009 WRI World Congress on Computer Science and Information Engineering, March 31 - April 2, 2009, Los Angeles, California, USA, 7 Volumes; 01/2009
  • [Show abstract] [Hide abstract]
    ABSTRACT: To achieve braking control of locomotive brake control system (LBCS) accurately and steadily under high nonlinearity various time delay condition, a locomotive brake control method based on T-S fuzzy modeling predictive control (MPC) is proposed. Firstly, the paper uses fuzzy clustering method (FCM) to initial parameters, and uses back-propagation algorithm to rectify rectified its premise parameters by learning off-line. The consequent parameters of the fuzzy rules are self-learning online by recursive least square method with the forgetting factor. By introducing the conception of quality satisfying degree to rectify forgetting factor, the paper can
    Intelligent Information Technology Applications, 2007 Workshop on. 01/2008; 3:602-607.