Over the last few years, human activity recognition (HAR) has drawn increasing interest from the scientific community. This attention is mainly attributable to the proliferation of wearable sensors and the expanding role of HAR in such fields as healthcare, sports, and human activity monitoring. Convolutional neural networks (CNN) are becoming a popular approach for addressing HAR problems. However, this method requires extensive training datasets to perform adequately on new data. This paper proposes a novel deep learning model pre-trained on scalograms generated using the continuous wavelet transform (CWT). Nine popular CNN architectures and different CWT configurations were considered to select the best performing combination, resulting in the training and evaluation of more than 300 deep learning models. On the source KU-HAR dataset, the selected model achieved classification accuracy and an F1 score of 97.48% and 97.52%, respectively, which outperformed contemporary state-of-the-art works where this dataset was employed. On the target UCI-HAPT dataset, the proposed model resulted in a maximum accuracy and F1-score increase of 0.21% and 0.33%, respectively, on the whole UCI-HAPT dataset and of 2.82% and 2.89%, respectively, on the UCI-HAPT subset. It was concluded that the usage of the proposed model, particularly with frozen layers, results in improved performance, faster training, and smoother gradient descent on small HAR datasets. However, the use of the pre-trained model on sufficiently large datasets may lead to negative transfer and accuracy degradation.

This article addresses the problem of finding the shortest route for Automated Guided Vehicles (AGVs) in a production environment with constrained battery state-of-charge (SoC) and time-dependent operating conditions. The route map is divided into a uniform grid containing stationary obstacles and two types of dynamic obstacles: human, for which AGV transportation is prohibited, and inanimate (moving objects), which impose a penalty function. A key contribution of the proposed methodology is the introduction of a battery residual charge matrix, which embeds cell-level energy feasibility directly into the grid-based environment representation by determining minimum admissible SoC constraints and accounting for transition-dependent energy costs. This matrix restricts the set of traversable cells under low-energy conditions, enabling energy-aware route feasibility evaluation during both initial planning and adaptive replanning. The proposed approach is based on the A* and D* Lite algorithms, providing shortest-path construction that explicitly integrates battery SoC into the spatio-temporal cost function. To avoid collisions in a multi-agent environment during routing, a simplified hybrid scheme with M* elements performs local coordination and adaptive trajectory replanning. The effectiveness of the proposed methodology was assessed using travel time, temporal complexity, and spatial complexity metrics. Simulation results on a $10\times 10$ grid showed that agents with sufficient battery completed routes of 8 and 11 cells with travel times of 7.2 to 10.7 conventional units. A critically low-energy agent was initially unable to move, but after adjusting the minimum SoC constraint, all agents completed their routes with travel times up to 11.4 conventional units, demonstrating the direct impact of energy constraints on system performance. Additional experiments with varying agent counts and SoC thresholds confirmed reliable balancing of route feasibility and energy constraints across configurations.

Understanding how artificial intelligence (AI) systems detect and correct anomalies in distributed industrial environments is crucial for building trustworthy autonomous systems. Automated Guided Vehicles (AGVs) operating in industrial settings generate a substantial amount of sensor data, providing an ideal testbed for developing explainable AI approaches to anomaly detection. Among the most important is the battery discharge, as it depends on the operation of the AGV. However, for a number of reasons, including sensor hardware instability, insecure communication during data transmission, unsynchronized timestamps, and uncalibrated sensors, data loss occurs. In federated learning, these issues are further compounded by decentralized data, the uniqueness of each client's data, and the model's inability to observe anomalies during training, among other factors. The publication provides a detailed analysis of classical anomaly detection methods for time series describing the battery discharge of three AGVs: IQR, Z-score, LOF, rolling mean, and SH-ESD. At the points where anomalies were detected, the values were filled in using interpolation methods: linear, nearest, spline (cubic), and polynomial. The advantages and disadvantages of each method are shown. Based on the analysis of detection and imputation methods for outliers in the battery discharge signal, an iterative anomaly detection method for the AGV sensor data was developed, assigning weights to each method.

The paper is devoted to the analysis of the spread of the COVID-19 pandemic in Ukraine based on finding the correlation between search terms in Google search engine and laboratory-confirmed cases. Statistics were obtained from open sources. The analysis was performed on matrices based on the Pearson correlation coefficient. To do this, we analyzed 25 typical search phrases, and after grouping them – 7 remained. The data were reduced to the same discreteness. Correlation matrices were calculated for each wave of the pandemic and for altogether. As a result, the correlation between search phrases and laboratory-confirmed cases was observed only in the second and third waves of the pandemic. Moreover, in the first wave, the preconditions for its occurrence were found; in the second - Pearson’s correlation coefficient was 0.74, and in the third wave, it decreased to 0.57. Other correlations that are specific to each pandemic wave are also analyzed. Additionally, it was proved that polynomials of the 6th degree most effectively restore lost data.

The incorporation of guided robotic platforms (GRPs) has become an essential component of Industry 4.0. A significant amount of data obtained from sensors can be used for intelligent control. Analyzing and selecting informative data, as well as removing redundant data for subsequent use, is a crucial task. The paper analyses and compares different industrial enterprise management types that utilize GRPs. A comparative analysis of procedures for optimizing the decision choice in general and situational management has been carried out. It has been proved that using traditional management for industrial enterprises that use GRPs is ineffective. The advantages and disadvantages of traditional and situational management of industrial enterprises are shown. Taking into account the shortcomings of each of them, the authors have developed an improved method of situational management that utilizes reliable artificial intelligence (AI). For this purpose, the stages of the modified general procedure for optimizing the choice of AI solutions have been described. A mathematical description of the process of evaluating alternatives has been made, which can be adapted to the use of AI models and their training. A model of the procedure for optimizing the choice of solutions based on situational control that considers the GRP’s battery charge has been developed. The model has been tested on a route map built at an industrial enterprise for the GRP Formica 1 manufactured by AIUT, Gliwice, Poland. The frame of the GRP Formica 1 data acquisition structure has been described. Data mining has been applied to the collected sensor data. The predictive model has been built based on an artificial neural network with deep learning and takes into account the residual charge of the battery, the mass transported by the GRP, and the presence of obstacles on its way. The mean absolute percentage error for the short-term forecast of the battery discharge of the GRP has decreased to 8 %. This proves the high efficiency of the developed method of situational control.