The proposed project aims at establishing the stationary PHEV test/emulation system. This system, based on the acquisition of requested instruments, will be used to support various research activities, including intelligent controls of permanent magnet motors; fault tolerant power converters; battery power and thermal management, and digital diagnosis of progressive damage in the battery. In addition, the project will provide a state-of-the-art PHEV design/testing facility in engineering education.
Intellectual Merit: The proposed project plans to incorporate control/diagnosis technology and evaluation/test training for plug-in hybrid electric vehicles in the research and education to meet the current and future demands of the U.S. automotive industry and the paradigm shift in U.S. energy dependency. The proposed approaches will investigate: 1. Performance improvement of the electric motor drive based on intelligent direct torque control; 2. Power converter fault detection/compensation schemes; 3. Battery thermal management and digital diagnosis of progressive damage in batteries.
Broader Impacts: The University of Michigan-Dearborn is uniquely housed in the midst of the Big 3 (GM, Ford, and Chrysler) and numerous small automotive companies. They need an institution in close proximity capable of supplying qualified personnel in the new technology. The proposed project and acquisition of the instruments will create opportunities to co-operate with local automotive companies. The outcomes of the proposed project will not only offer a useful and needed research and educational experience that is relevant to the automotive industry of the day, but they will also help to bolster the industry by providing more qualified students.
The main objectives of this project are to establish a plug-in hybrid electric vehicle (PHEV) test and emulation system which contains the PHEV evaluation and battery testing functionality. The key instruments, including a real time simulation equipment, data acquisition instruments, and a battery cycler/simulator, were acquired and located in a lab space on the University of Michigan-Dearborn. The lab includes the PHEV evaluation and battery testing functionality based on the battery cycler/simulator and dynamometers along with a control/monitor station. This lab is being used by engineering faculty members, students, visiting scholars, and visiting local automotive engineers for the research of PHEV and batteries. The system based on the acquisition of the instruments has been and will be used to support various research activities, including intelligent controls of the permanent magnet motors, fault tolerant power converters, battery power and thermal managements, and digital diagnosis of progressive damage in batteries. The PIs developed a PHEV model implemented in the real time simulator. The acquired real-time simulator and the data acquisition instruments were very helpful to investigate the converter fault detection and compensation control algorithm since the hardware in the loop functionality of the real time simulator and the fast data acquisition are essential to detect a fault and to come up with the compensation algorithm. The battery management and diagnosis research has been performed as well. The high efficient battery balancing circuit development has been performed and the research of the battery state of charge estimation for the battery management has been conducted as well. During the project period, the new fault tolerant power conversion and fault compensation control for a bi-directional DC-DC converter in electric vehicles were investigated. The new multi-mode, single-leg energy conversion system replaces two independent converters (a boost converter and a bi-directional DC/DC converter) utilizing two functions. Therefore, the converter can be employed in electric vehicles having fault tolerant characteristics if one switch of the boost converter fails. For the research of progressive damage diagnosis in batteries, a systematic approach to explore the material damage with an emphasis on the link from microscale to macroscale is implemented. Experiments are performed to validate the efficacy of our SRA (super representative volume) approach, compared to many analytical models in the field of damage mechanics and micromechanics. We found that the SRA method is also superior to traditional homogenization methods based on a statistical mean. Our numerical model and analysis method could play an important role in the quantification of material degradation. The machine learning framework for the optimization of energy management in a PHEV was also investigated. This framework includes machine learning algorithms for predicting roadway types, traffic congestion levels, and driving trends to use these predicted values in another algorithm that learns optimal energy settings. The neural network has been adopted and trained for the prediction of roadway types and traffic congestion levels. Its performance for a 9-s window was approximately 94% in accuracy for both training and test data. In addition, the proposed project and acquisition of the instrument created opportunities to co-operate with local automotive companies. The outcomes of the proposed project provided a useful and needed research and educational experience that is relevant to the automotive industry.
