Anomaly detection is the process of detecting unusual or unforeseen patterns or events in data. Many factors, such as malfunctioning hardware, malevolent activities, or modifications to the data’s underlying distribution, might cause anomalies. One of the key factors in anomaly detection is balancing the trade-off between sensitivity and specificity. Balancing these trade-offs requires careful tuning of the anomaly detection algorithm and consideration of the specific domain and application. Deep learning techniques’ applications, such as LSTMs (long short-term memory algorithms), which are autoencoders for detecting an anomaly, have garnered increasing attention in recent years. The main goal of this work was to develop an anomaly detection solution for an electrical machine using an LSTM-autoencoder deep learning model. The work focused on detecting anomalies in an electrical motor’s variation vibrations in three axes: axial (X), radial (Y), and tangential (Z), which are indicative of potential faults or failures. The presented model is a combination of the two architectures; LSTM layers were added to the autoencoder in order to leverage the LSTM capacity for handling large amounts of temporal data. To prove the LSTM efficiency, we will create a regular autoencoder model using the Python programming language and the TensorFlow machine learning framework, and compare its performance with our main LSTM-based autoencoder model. The two models will be trained on the same database, and evaluated on three primary points: training time, loss function, and MSE anomalies. Based on the obtained results, it is clear that the LSTM-autoencoder shows significantly smaller loss values and MSE anomalies compared to the regular autoencoder. On the other hand, the regular autoencoder performs better than the LSTM, comparing the training time. It appears then, that the LSTM-autoencoder presents a superior performance although it was slower than the standard autoencoder due to the complexity of the added LSTM layers.

As unmanned aerial vehicle (UAV) technology has evolved, these systems are being increasingly utilized across diverse industries. However, controlling UAVs faces significant problems owing to several environmental circumstances and obstacles, making path planning a critical initial step for UAV operation. This paper offers an overview of UAV path planning research founded on intelligent algorithms, which are divided into three categories: computational intelligence (CI), machine learning (ML), and hybrid methods. Each category has been analyzed in depth to show its strengths, limits, and where it may be applied to UAV-related problems. The methodology includes a comparative analysis based on multiple performance metrics such as path length, flight time, collision avoidance, complexity, and environmental adaptability. Furthermore, the research covers the latest publications that deal with solving essential challenges of UAV path planning by using new hybrid algorithms and enhanced optimization methods. The results indicate that although each strategy offers specific strengths suited to particular scenarios, hybrid strategies are more likely to deliver greater flexibility and robustness, particularly in uncertain, and dynamic environments. These findings are significant for guiding future research in adaptive path planning and for supporting practical UAV applications such as autonomous delivery, aerial surveillance, disaster response, and environmental monitoring.

Path planning is a crucial step in robotic navigation to satisfy: tasks safety, efficiency requirements and adapt to the complexity of environments. Path planning problem is particularly critical for Unmanned Aerial Vehicles (UAV), being increasingly involved within important tasks in diverse military and civil fields such as: inspection, search and rescue and communication, taking advantage of their high flexibility, maneuverability and cost-effective solutions. This continuous growth made the solution of UAV path planning problem an interesting research topic in recent years. In this scope, machine learning algorithms were a promising tool due to their continuous data-driven selfimprovement to adapt with the high dynamicity of environments where conventional programming fails. This paper provides a review on recent developments in machine learning-based UAV path planning issued from credible databases like: IEEE, Elsevier, Springer Links and MDPI. The main contribution of this paper is to delve through these recent works providing a taxonomy of algorithms into the fundamental paradigms: supervised, unsupervised and reinforcement, evaluating their efficiency and limitations under distinct scenarios. Despite the relative generalization of deep reinforcement learning to different environments, this study highlighted some active challenges about computational cost and real-time applications that remain open.