The Prius-a power-split hybrid electric vehicle from Toyota-has become synonymous with the word "Hybrid." As of October 2010, two million of these vehicles had been sold worldwide, including one million vehicles purchased in the United States. In 2004, the second generation of the vehicle, the Prius MY04, enhanced the performance of the components with advanced technologies, such as a new magnetic array in the rotors. However, the third generation of the vehicle, the Prius MY10, features a remarkable change of the configurationan additional reduction gear has been added between the motor and the output of the transmission [1]. In addition, a change in the energy management strategy has been found by analyzing the results of a number of tests performed at Argonne National Laboratory's Advanced Powertrain Research Facility (ARRF). Whereas changes in the configuration, such as the reduction gear, are possibly noticeable, it is not easy to determine the effect of the energy management strategy because the supervisory control algorithm is, generally, not published. Further, it is almost impossible to analyze the algorithm without testing results obtained from a well-designed testing process. On the basis of extensive experience in designing the controllers of power-split hybrid electric vehicles in Autonomie, we could identify the supervisory control algorithm by analyzing the testing results obtained from the APRF. A vehicle model and a control model for the Prius MY10 have been developed to reproduce the real-world behaviors, and the simulation results are compared with the testing results. In the simulation, the developed vehicle model achieves fuel consumption that is close to the testing value, within 5%, and the operation of the engine model was similar to that of the real-world engine.

No abstract

No abstract

No abstract

The Prius, a power-split hybrid electric vehicle developed by Toyota, has been the top-selling vehicle in the United States hybrid electric vehicle market for the last decade. The transmission system of the vehicle is a frequent theme of study for hybrid electric vehicles. However, the control concept of the vehicle is not well known, since analyzing control behaviors requires well-designed facilities to obtain testing results and well-defined processes to analyze the obtained results. Argonne National Laboratory has these resources and capabilities. In addition, Argonne has produced a reliable simulation tool, Autonomie, by which a vehicle model for the 2010 Prius is developed on the basis of the analyzed results, and it is validated with the results of testing. The developed model demonstrates that results of vehicle performance from simulation are close to those of from real-world tests-within 5%. The main focus of this study is to provide information about the supervisory control for the 2010 Prius, so that researchers can reproduce the real-world behavior of the vehicle through simulations. The analyzed control ideas based on the testing results will be very helpful in terms of understanding the control behavior of the vehicle, and the information resulting from this study is useful to develop the controller for the vehicle at a simulation level.

No abstract

Traffic forecasting approaches are critical to developing adaptive strategies for mobility. Traffic patterns have complex spatial and temporal dependencies that make accurate forecasting on large highway networks a challenging task. Recently, diffusion convolutional recurrent neural networks (DCRNNs) have achieved state-of-the-art results in traffic forecasting by capturing the spatiotemporal dynamics of the traffic. Despite the promising results, however, applying DCRNNs for large highway networks still remains elusive because of computational and memory bottlenecks. This paper presents an approach for implementing a DCRNN for a large highway network that overcomes these limitations. This approach uses a graph-partitioning method to decompose a large highway network into smaller networks and trains them independently. The efficacy of the graph-partitioning-based DCRNN approach to model the traffic on a large California highway network with 11,160 sensor locations is demonstrated. An overlapping-nodes approach for the graph-partitioning-based DCRNN is developed to include sensor locations from partitions that are geographically close to a given partition. Furthermore, it is demonstrated that the DCRNN model can be used to forecast the speed and flow simultaneously and that the forecasted values preserve fundamental traffic flow dynamics. This approach to developing DCRNN models that represent large highway networks can be a potential core capability in advanced highway traffic monitoring systems, where a trained DCRNN model forecasting traffic at all sensor locations can be used to adjust traffic management strategies proactively based on anticipated future conditions.

No abstract