Eduardo Coronel

This survey presents a comprehensive review of data-driven approaches for industrial asset maintenance, emphasizing the use of data mining and machine learning techniques, including deep learning, for condition-based and predictive maintenance. It examines 534 references from 1995 to 2023, along with three additional articles from 2024 on natural language processing and large language models in industrial maintenance. The study categorizes two main techniques, four specialized approaches, and 27 methodologies, resulting in over 100 variations of algorithms tailored to specific maintenance needs for industrial assets. It details the data types utilized in the industrial sector, with the most frequently mentioned being time series data, event timestamp data, and image data. The survey also highlights the most frequently referenced data mining algorithms, such as the proportional hazard model, expert systems, support vector machines, random forest, autoencoder, and convolutional neural networks. Additionally, the survey proposes four level classes of asset complexity and studies five asset types, including mechanical, electromechanical, electrical, electronic, and computing assets. The growing adoption of deep learning is highlighted alongside the continued relevance of traditional approaches such as shallow machine learning and rule-based and model-based techniques. Furthermore, the survey explores emerging trends in machine learning and related technologies, identifies future research directions, and underscores their critical role in advancing condition-based and predictive maintenance frameworks.

In the present work the problem of Optimal Placement of remote controlled switches, through the replacement of installed manual switches, in Electric Power Distribution Systems with radial configuration is approached, considering economic, technical, operational and social aspects. The approach used for the identification of strategic points is based on the implementation of meta-heuristic algorithms for optimization with multiple objectives. For the evaluation of each solution, an evaluating function based on the Monte Carlo method is implemented, which estimates the defined indices through the simulation of failures in the network, whose probabilities are obtained using the Reliability Block Diagram method. Three different multi-objective algorithms were implemented and their results compared using the hypervolume metric, having the Generalized Differential Evolution 3 algorithm the best performance.

In the present work the problem of Optimal Placement of remote controlled Switches in an Electric Power Distribution Systems is analyzed with a Multi-objective approach. The selected objectives consider economic, technical, operational and social aspects. Real data from Paraguayan East Region is used. For the evaluation of solutions, an evaluating function based on the Monte Carlo method is implemented, which estimates the defined indices through the simulation of failures in the network, whose probabilities are obtained using the Reliability Block Diagram method. Six different multi-objective algorithms were implemented and compared using ONVGR and Hypervolume metrics, having the Multi-objective Ant Colony Optimization algorithms the best performance.