Case Study AI-Based Energy Management

Xentara and ShiraTech-Knowtion developed an AI-powered energy management system to reduce energy costs and CO₂ emissions in manufacturing. By integrating real-time data from sensors, weather forecasts, and production systems, the solution enables predictive energy use, peak shaving, and optimized equipment maintenance. Implemented with ONNX machine learning and Xentara's data platform, the system achieved up to 30% energy savings, supported ISO compliance, and advanced sustainability goals in line with Industry 4.0.

AI-Powered Energy Forecasting
AI-Powered Energy Forecasting

Reducing Consumption Costs and CO2 Footprint

“By introducing the energy management system provided by Xentara and ShiraTech-Knowtion, we managed to not only lower utility costs but also make significant progress on our sustainability goals.”

VP of Production, German Machine Building SMP

The Project

Effective energy management is essential for numerous manufacturing firms to minimize energy expenses and comply with increasing sustainability and regulatory demands. ShiraTech-Knowtion was commissioned to develop a prototype of an AI-enhanced energy management system. The project's main objective was to create a thorough solution for forecasting and assessing energy use in a production setting, emphasizing temperature control and air conditioning usage. This entailed the collection, storage, analysis, and visualization of energy consumption and environmental data to enhance energy management and notify stakeholders of potential energy surges.

The system facilitates effective strategies like intelligent peak shaving and load shifting, and enhances equipment maintenance by identifying trends. AI-driven predictions enhance solar energy utilization and diminish CO2 emissions, presenting considerable savings opportunities, especially in energy-intensive operations. Intelligent data analysis and the utilization of suitable sensors can identify irregularities in manufacturing processes and execute predictive measures.

Real-time data integration is crucial for establishing optimal circumstances for AI utilization. The plethora of sensors and data sources necessitated the integration of disparate formats and protocols. Consequently, ShiraTech-Knowtion resolved to implement Xentara as a data aggregator, preprocessor, and pipeline for the machine learning model.

Data Acquisition and Consolidation

A crucial element of the strategy is the accurate and thorough gathering of energy consumption data. This is accomplished by integrating existing energy meters and systems, together with supplementary sensors, such as temperature sensors. Incorporating shop floor data, including production area utilization metrics, facilitates more precise labelling of energy data, contextualizing energy consumption within actual production settings. Weather data, especially temperature forecasts, is essential for effectively predicting energy consumption. This project involved the importation of meteorological data from the German Weather Information Service (Deutscher Wetterdienst, abbreviated as DWD).

The Xentara platform functions as a central hub for communication and convergence. For the demonstration project, the Xentara team dedicated months on gathering data from multiple sources, integrating it into the semantic data model, and standardizing the diverse data formats. The data was subsequently put into a Machine Learning model, produced in ONNX (Open Neural Network Exchange) format, and delivered to ShiraTech-Knowtion.

AI-Driven Analysis and Assessment

The fundamental component of the solution is ShiraTech-Knowtion’s AI-driven data analysis and assessment. It facilitates the forecasting of peak loads and the recognition of additional opportunities for optimizing energy consumption, hence diminishing CO2 emissions. Identified signals and recommendations are transmitted to pertinent departments to provide focused controls and actualize savings possibilities. The automatic and timely identification of correlations facilitates predictive measures beyond energy management. Anticipated consumption surges are forecasted, prompting immediate notifications to pertinent stakeholders, while historical data is prepared for visualization and analysis across many specialized dashboards.

The implementation of advanced multisensor technology, exemplified by the iCOMOX™ sensor box as a versatile retrofit solution, facilitates both accurate energy optimization and seamless integration into predictive maintenance.

Software Components Provided

• Xentara Core for Data Management • ONNX Machine Learning Skill • Websocket Connector (Server & Client) • Multiple I/O and Protocol Connectors (EtherCAT, Modbus, OPC UA etc.)

Deliverable Work Packages

• Interfacing with Different Energy Meters • Integrating iCOMOX Sensor Box • Configuring Data Model • Collecting and Importing Training Data • ONNX Model Integration • Local Edge Implementation

Results and Benefits According to Pilot Customer

Up to 30% energy cost savings • Enhanced ecological impact of manufacturing procedures Swift adaptation to alterations • Dependable forecasting and mitigation of peak loads (peak shaving) Effective facilitation of energy management systems in accordance with ISO 50001 (Proactive monitoring using AI) • Compatible with existing management systems such as ISO 9001 and ISO 14001 • Seamless transition into predictive maintenance • Introduction to a comprehensive, real-time data management system for Industry 4.0 / Manufacturing X applications.

 


Share this article

If you found this article helpful, please share it with your friends!