فصلنامه کهربا
سال سیزدهم شماره 44 (تابستان 1403)

Artificial intelligence and advanced technologies, including sensors, data processing, wireless networks and complex algorithms, have clearly brought about major changes in the automotive industry. One of the areas where these technologies have had a great impact is the control systems of smart cars. Improving the performance of these systems is very important because these systems directly affect the safety and efficiency of the vehicle. One of the ways to improve the performance of smart car control systems is to improve the algorithms and software used in these systems. Increasing the speed of data processing, improving the decisions made by the system, and increasing the accuracy in diagnosing and predicting different road conditions can improve the overall performance of smart car control systems.Also, the use of advanced sensors and more communication between cars and road infrastructure can also help improve the performance of these systems. These measures not only improve the safety and efficiency of cars, but can also help reduce accidents and urban traffic.

Web page recommender systems are used to provide suggestions related to the content of websites and web pages to users in order to reduce information overhead. By analyzing users' opinions and feedback, these systems can identify user behavior patterns and preferences and provide suitable suggestions based on them.In this paper, the combination of social influence maximization problem and collaborative filter based on deep auto-encoder network is used to estimate the ranking of users in the network of similar users based on users' opinions. The proposed method finds influential users by finding the network of similar users based on users' opinions based on the genetic algorithm. Influential users form the group of similar users in terms of ratings as the input of the deep autoencoder network, whose output is the rating estimation of new users. Finally, web visit recommendations are made for new users based on the collaborative filter resulting from the prediction of the favorite items of influential users. The proposed method with MAE, RMSE, Hit Rate and Accuracy values equal to 0.0024, 0.0025, 95.31% and 95.08, respectively, has obtained better results than other previous methods.

One of the most important and widely used fields of medical engineering is working on the heart and its signals. Automated heart sound signal quality assessment is an essential step for reliable heart sound signal analysis. An inevitable processing step for this purpose is heart sound segmentation, which is still a challenging task from a technical point of view. One of the fields of research in the fields related to the heart is the processing and classification of heart sound (usually to detect heart valve diseases) where various types of features are extracted from the sound signal of the heart and then used by classifiers. In this study, it is intended to extract a limited number of features that are calculated with the help of STFT from the sound signal of the heart and to classify the sound of the heart into two categories, "healthy" and "diseased", using the SVM classifier.

In this research and article, a local substrate removal step has been used to make a small panel with metal oxide-semiconductor technology complementary to photovoltaics with backlight suitable for high voltage and low power applications. Local substrate removal can be considered an advanced multi-chip module technology by creating a 50 µm physical gap between on-chip photovoltaic cells for electrical isolation, while requiring time-consuming pick and place steps. The proof-of-concept photovoltaic module achieved maximum opencircuit voltage and microamp-scale short-circuit current in a small form factor (3.13 V/mm2). The fabrication of this mini-photovoltaic module is based on standard microelectronics manufacturing/packaging processes, thereby ensuring easy integration with other microelectronics for self-powered systems.

This paper presents a new reliable cloud data storage system that uses a hybrid GWO-HHO metaheuristic algorithm. Existing algorithms often do not consider the general and specific constraints related to the allocation of components and VMs for data storage. These constraints include instances, interactions between components, allocation rules, and device capacity. This hybrid meta-heuristic algorithm optimizes the allocation of components and effectively solves the problems of cloud data storage in components. The performance of the proposed system was validated and compared with other models. Comparative analysis and statistical evaluation were performed to evaluate the effectiveness of the system. To evaluate the performance of the advanced system, it was compared with other existing approaches. The purpose of this comparison is to show the superiority of the proposed approach in terms of its performance and efficiency. The proposed method provides better performance in the cloud data storage model; Therefore, it can be proved that it can reduce the time complexity and increase the performance. Compared to other approaches, the proposed model shows a lower requirement for active servers. This shows that the proposed model is more efficient in using server resources, which in turn leads to energy saving in the cloud data storage process. By requiring fewer active servers, the proposed model improves system reliability performance. It minimizes the energy consumption associated with server operations and leads to increased efficiency and cost-effectiveness in cloud data storage. Also, the results of the proposed method clearly show that the recommended model requires less processing time for efficient cloud data storage. Decreasing the makepan value indicates the effectiveness of the model in optimizing the storage process and minimizing the time required for processing. As a result, the proposed model is more efficient and able to achieve efficient cloud data storage.

In this article, we have presented a simple but accurate model of metallic carbon nanotube. By using curve fitting, we have added coefficients in the simpler model, and the presence of these coefficients caused our proposed model to have an error of nearly 6 percent compared to the more complex model.

This study proposes a data-driven method for estimating the state of charge (SOC) of lithium-Ion batteries to overcome the limitations of conventional model-based approaches that require accurate parameters. The performance of the proposed model is analyzed by estimating the SOC prediction accuracy in environments with different operating temperatures. As a result, the proposed method shows high estimation accuracy (mean absolute error less than 4%) over a wide range of driving temperatures (0 to 25 degrees Celsius) with only limited driving data.