Intelligent energy management using data mining techniques at Bosch Car Multimedia Portugal facilities

Abstract

The fusion of emerged technologies such as Artificial Intelligence, cloud computing, big data, and the Internet of Things in manufacturing has pioneered this industry to meet the fourth stage of the industrial revolution (industry 4.0). One major approach to keeping this sector sustainable and productive is intelligent energy demand planning. Monitoring and controlling the consumption of energy under industry 4.0, directly results in minimizing the cost of operation and maximizing efficiency. To advance the research on the adoption of industry 4.0, this study examines CRISP-DM methodology to project data mining approach over data from 2020 to 2021 which was collected from industrial sensors to predict/forecast future electrical consumption at Bosch car multimedia facilities located at Braga, Portugal. Moreover, the influence of indicators such as humidity and temperature on electrical energy consumption was investigated. This study employed five promising regression algorithms and FaceBook prophet (FB prophet) to apply over data belonging to two HVAC (heating, ventilation, and air conditioning) sensors (E333, 3260). Results indicate Random Forest (RF) algorithms as a potential regression approach for prediction and the outcome of FB prophet to forecast the demand of future usage of electrical energy associated with HVAC presented. Based on that, it was concluded that predicting the usage of electrical energy for both data points requires time series techniques. Where "timestamp" was identified as the most effective feature to predict consume of electrical energy by regression technique (RF). The result of this study was integrated with Intelligent Industrial Management System (IIMS) at Bosch Portugal.