HyperLink


HyperLink



                                                           

 

 

Research Results

1) Energy Management System of Electrical Scooter.

    Because of its nonlinear discharge characteristics, the residual electric energy of a battery remains to be an open problem. As a result, the reliability of electric scooters or electric vehicles is lacking. To alleviate this problem and enhance the capabilities of present electric scooters or vehicles, we propose a state-of-charge learning system that can provide more accurate information about the state-of-charge or residual capacity when a battery discharges under dynamic conditions. The proposed system is implemented by learning controllers, fuzzy neural networks and cerebellar model articulation controller networks, which can estimate and predict nonlinear characteristics of the energy consumption of a battery. With this learning system, not only could it give an estimate of how much residual battery power is available, but it also could provide users with more useful information such as an estimated traveling distance at a given speed, and the maximum allowable speed to guarantee safety arrival at the destination.

2) Embedded System based State of Charge System for Electrical Scooter.

   We design the EMS (Energy Management System) in order to evaluate battery of electric scooters. Because of the nonlinear battery discharge characteristics and different driving behaviors of riders, it is not easy to predict how much electric power remaining in the battery. As a result, the reliability of the electric scooter is lacking. We propose an intelligent control system SOC (state-of-charge) that predict the remaining power of the electric scooter precisely and combine with current driven hardware-Linux embedded system. 

3) Energy Management System for EVT-4000 Electrical Scooter.

   The proposed energy management system is designed for EVT-4000 electrical scooter. In order to monitor the electric energy consumption in EVT-4000 electrical scooter, we further design the interface circuits and user interface in the proposed system. Because of its nonlinear discharge characteristics, the estimating of state of charge and state of health for a battery in EVT-4000 is still an open problem. To solve this problem, the proposed embedded system based energy management system can be used to estimate state-of-charge and state-of-health when a battery discharges under dynamic conditions. The proposed system combining with the existing control hardware will be a prototype of energy management system for EVT-4000 electrical scooter.

4) Neuromuscular-Like Control.

  The muscle-like control theory is designed to achieve compliance control of motor. The servo control system is a simple model of automatic door which is set up with servo motor and ball screw. For a good servo control system, the closed-loop system is the best to use. Since we attempt to use torque motor with encoder to achieve the closed-loop control. And employing position feedback error of encoder, we can achieve the different control target by muscle-like controller. Besides we also apply muscle-like controller on compliance control. And muscle-like controller can improve transient response situation on tracking. On force control, muscle-like controller will immediately produce a compensate value of torque to stop motor when the system has a on unexpected force disturbance (e.g., automatic door damage user). Most especially, the motor can still finish action when disturbance is disappeared. It is necessary to be mentioned the expensive force sensor does not need to achieve compliance force control by muscle-like control theory. It shows muscle-like control theory is very suit to perform an intelligent safety system.    .

5) A Hybrid Adaptive Scheme of Fuzzy-Neural Iterative Learning Controller for Nonlinear Systems.

   To deal with an iterative learning control (ILC) problem of unknown nonlinear systems with initial state errors and state dependent input gain, a hybrid adaptive scheme of fuzzy neural iterative learning controller is presented in this work. A new time-varying boundary layer with variable initial layer width is first introduced to design an error function in order to overcome the uncertainty from initial state errors.   The hybrid adaptive learning controller is then constructed based on the above error function. This hybrid learning controller combines the features of an indirect and a direct adaptive learning control such that the tradeoff between plant knowledge and control knowledge can be realized by adjusting a weighting factor. As in general the nonlinearities of the dynamic system are unknown, a fuzzy neural network (FNN) is applied for the compensation of nonlinearities and certainty equivalent controller. The linguistic descriptions about the plant and linguistic control rules are appended into the indirect and direct component, respectively. Since the optimal parameters of the FNN for a minimum approximation error are generally unavailable, some stable adaptive algorithms are derived along the iteration axis to search for suitable parameter values. It is shown that the proposed learning system will guarantee the boundedness of all the internal signals in time domain and ensure the convergence of learning error along iteration domain, respectively. Moreover, the norm of tracking error vector at each time instant will asymptotically converge to a tunable residual set as iteration goes to infinity even the initial state errors are varying and large.

Publications and Patents 

  • Publications:
    • D.T. Lee, S.J. Shiah, C.M. Lee and Y.C. Wang, "State of Charge Estimation for Electric Scooters by Using Learning Mechanisms," To Appear in IEEE Trans. Vehicular Technology.
    • D.T. Lee, S.J. Shiah, C.M. Lee, and C.H. Wu, “Intelligent Control of Electric Scooters,” presented in 5th IASTED international conference, Tsukuba, Japan , 2002.
    • Y.C. Wang, C.J. Chien, and D.T. Lee,“A hybrid adaptive scheme of fuzzy-neural iterative learning controller for nonlinear dynamic systems,”International Journal of Fuzzy Systems, (EI), Vol. 7, No. 4, pp. 147-157, 2005.
    • Y. C. Wang, C. J. Chien and D. T. Lee "An Adaptive PID-type Iterative Learning Controller for Unknown Nonlinear Systems," The Eighth International Conference on Control, Automation, Robotics and Vision (ICARCV2004), Kunming, China 2004.
  • Patents:
    • An Intelligent System for Regulating the Speed of an Electric Vehicle
      • Inventors: D.T.Lee & C. H. Wu
      • Country: R.O.C
      • Date of Patent: 2002/06/~2019/01/
      • Abstract: An Intelligent System for Regulating the Speed of an Electric Vehicle
    • An Intelligent System for Regulating the Speed of an Electric Motorcycle
      • Inventors: D.T.Lee & C. H. Wu
      • US
      • Date of Patent: 2001/11/~2019/01/
      • Abstract: An Intelligent System for Regulating the Speed of an Electric Motorcycle
    • Nonlinear Muscle-Like Compliant Controller
      • Inventors: D.T.Lee & C. H. Wu
      • US
      • Date of Patent: 2001/06/~2019/01/
      • Abstract: Nonlinear Muscle-Like Compliant Controller, (with C-H Wu)
    • 李德財, 吳紀皞,“自動調節電動交通工具速率的智慧型系統”,中華民國專利第158191號.

Download Area 

       ·         Hot Science from Taiwan- National Geographic Channel Oct. 21,2002 (MPEG)          features the electric motorcycle project as of 2002 in Taiwan . 

·         Neuro-Muscular-like Controller Technology (PUMA560 Video Demo) (MPEG) is a video demo of a neuromuscular-like controller (NMC) technology, which has applications to rehabilitation robots and to suspension and vibration control for vehicles and aircraft. See compliant control without force sensor for a more detailed description of this demo.

 

 

HyperLink HyperLink HyperLink HyperLink  








HyperLink