No abstract

No abstract

The Fuel_LF‐RTO strategy is real‐time optimization (RTO) strategy proposed here to find the optimal values of fueling for the polymer electrolyte membrane fuel cell hybrid power sources under unknown load profile, which is the case of fuel cell vehicle. The proposed optimization strategy is based on global extremum seeking (GES) algorithm and load‐following (LF) control for air and fuel flows. The results show the performance obtained with Fuel_LF‐RTO strategy in comparison with the Static Feed‐Forward strategy. The performance was estimated for constant and variable load. The FC system efficiency and the fuel consumption efficiency for maximum load of 8 kW can increase with up to 1.88% and 11.26 W lpm−1 in comparison with the sFF RTO strategy. Also, the fuel economy is 27.36 L during the 8 kW/12 s constant load cycle, which means an economy of 136.8 lpm. This performance is highlighted for constant load in range 2 to 8 kW, which represents 0.33% and 1.25% from nominal power of the 6 kW FC stack used in this study. Also, the performance was estimate for variable load considering the fuel economy, which can be up to 21.86 l during the 6.25 kW/12 s pulsed load cycle.

No abstract

In this paper, we present an analysis of important aspects that arise during the development of neural network applications. Our aim is to determine if the choice of library can impact the system’s overall performance, either during training or design, and to extract a set of criteria that could be used to highlight the advantages and disadvantages of each library under consideration. To do so, we first extracted the previously mentioned aspects by comparing two of the most popular neural network libraries—PyTorch and TensorFlow—and then we performed an analysis on the obtained results, with the intent of determining if our initial hypothesis was correct. In the end, the results of the analysis are gathered, and an overall picture of what tasks are better suited for what library is presented.