---

The control strategy in hybrid electric vehicles (HEV) determines the energy management between an internal combustion (IC) engine and an electric machine to generate the power required to drive the vehicle. In this study, the parameters of an electric assist control strategy of HEV, which is used in ADVISOR software, are optimized by minimizing the fuel consumption while maintaining the engine emissions below the Euro3 standard. In this way, a parallel HEV has been simulated in two drive cycles and the optimal parameters of controller are obtained by the genetic algorithm (GA). The simulation results show that by optimizing the control parameters, the fuel consumption decreases while satisfying the Euro3 standards and other constraints in various drive cycles. Also, different values for optimal parameters have been extracted. This indicates that the optimal parameters of the controller are dependent on the drive cycle.

---

Since the introduction of the first silicon solar cell, there have been steady improvements in its performance parameters such as light trapping, solar absorption, cell efficiency and manufacturing costs. In thin silicon cells, some of the light photons that are not absorbed by the semiconductor are always lose in various ways. The diffraction grating causes the photons to travel a longer light path due to the collision with this structure, which increases the length of the light path of the photons and cell absorption, that thus improving cell efficiency. In each of the mentioned structures, optimal materials and geometric properties have been used to achieve maximum efficiency of silicon cells. Intelligent optimization methods have been used to find the optimal geometric parameters for the structure. In choosing search methods from the two algorithms particle swarm optimization and genetics and creating a combination of the both, the positive feature of both algorithms was used to achieve the best answer. This combination has produced very positive results, which thereby, 23.293 efficiencies and 35.41 mA/cm2 short circuit current were obtained.

---

Insufficient training data is one of the main challenges of utilizing deep Convolutional Neural Networks (CNNs) for Environmental Sound Classification (ESC). As a promising solution, Transfer Learning (TL) has addressed this issue by adapting a network pre-trained on a large-scale dataset to the target task. In this paper, we demonstrate that not all neurons/kernels of every layer in CNN networks are equally utilized to process the inputs of different classes, but there is a specific subgroups of neurons/kernels in every layer that play the key role in classification of every output class. Based on this observation and due to similarities that exist between feature spaces of some source and target classes, we propose to concentrate the fine-tuning process only on those neurons/kernels that do need changes and have the greatest impact on misclassifying target data. To identify these neurons/kernels, we pose a nested optimization problem for which we propose an effective evolutionary approach as solution.  Compared to the conventional fine-tuning approach, our proposed method achieves absolute improvements of about 1.9% and 2.3% in accuracy on ESC-50 and DCASE-17, respectively; remarkable improvements produced not by adding augmented data but with a more efficient utilization of knowledge stored in the pre-trained network. It is noteworthy that the computation time overhead of the proposed evolutionary method is rather small (about one third of the time required to train the model from scratch.