Growing battery use in energy storage and automotive industries demands advanced Battery Management Systems (BMSs) to estimate key parameters like the State of Charge (SoC) which are not directly measurable using standard sensors. Consequently, various model-based and data-driven approaches have been developed for their estimation. Among these, the latter are often favored due to their high accuracy, low energy consumption, and ease of implementation on the cloud or Internet of Things (IoT) devices. This research focuses on creating small, efficient data-driven SoC estimation models for integration into IoT devices, specifically the Infineon Cypress CY8CPROTO-062S3-4343W. The development process involved training a compact Convolutional Neural Network (CNN) and an Artificial Neural Network (ANN) offline using a comprehensive dataset obtained from five different batteries. Before deployment on the target device, model quantization was performed using Infineon’s ModusToolBox Machine Learning (MTB-ML) configurator 2.0 software. The tests show satisfactory results for both chosen models with a good accuracy achieved, especially in the early stages of the battery lifecycle. In terms of the computational burden, the ANN has a clear advantage over the more complex CNN model.

This paper describes a complete approach to filtering Partial Discharge (PD) pulses from interference in High Voltage electrical equipment using supervised Machine Learning (ML) techniques. The PD signals are registered in Ultra High Frequency radiation band with a multisensor acquisition system composed of 4 antennae. The proposed methodology focuses on the implementation ML algorithms and proposes a novel in this field approach to the onset detection of incoming signals. The goal was to achieve high accuracy of filtering with reasonably low compilation times of the ML classifier. That would allow to use the model on edge sensor devices. In the paper, different models and training variants of the ML framework are tested. The presented results are based on a robust measurement campaign performed in laboratories of GEIRI Europe. The methodology is validated through tests on 3 separate test scenarios. Each represents a different complexity of the problem with an increasing number of active sources. The results show high potential for utilization of the ANN and other classifiers for PD filtering problems as the accuracy achieves the desired threshold of 80% for most of the tested variants. The methodology is a step forward toward a fully online PD and interference filter.

We analyzed the earthquake density of the Geysers geothermal field (California) as a function of time and space over a decade. We grouped parts of the volume of the geothermal area sharing similar earthquake rates over time; in this way, we found three concentric spatial domains centered on the principal exploitation area and labelled as A, B, C, moving from the inner-to the outermost domain, and characterized by peculiar time-history of the earthquake rates and different stress conditions. The earthquake density decreases moving from domain A to C, and different slopes of the earthquake frequency-magnitude distribution appear for the domains A-B and domain C. Stress field propagates via a diffusive mechanism up to about 3.5 km from the center of the geothermal area outwards, and a mean hydraulic diffusivity of about 0.05 m ² / s is estimated; at larger distances a poroelastic stress transfer dominates. Our approach can identify spatio-temporal patterns of physical mechanisms driving induced seismicity and can in principle be extended to other settings of man-induced earthquakes. Moreover, it potentially allows a differentiated assessment of the seismic risk within each domain, as well as the identification of domains with no or minimal induced seismicity.