AI-based active energy management

About the project

The main objectives of the project are to: 

1. a) Develop a digital module on top of existing MePIS Energy product that integrates real-time data on energy consumption, production plan (if available), contractual peak loads for different periods and market (pricing) data.
2. b) Implement advanced artificial intelligence (AI) and machine learning (ML) algorithms to forecast energy consumption and
3. c) Identify deviations in forecasted energy consumption, foremost events that may violate given business constraints such as maximum allowed peak power consumption in a certain time period.
4. d) Identify reasons for deviations (e.g. breakdown of energy consumption in that period).
5. e) Devise AI/ML-based optimal adjustment plan for energy consumers in that time period and if necessary, act preventively (e.g. fill stock, power down), while having minimal impact on core production processes and considering all prescribed constraints.

Challenges

Traditional approaches in energy management and consumption forecasting in manufacturing are not effectively addressing their complex energy environments. These environments turned from consumer to prosumer (where energy is both produced and consumed by the same entity) through installation of green energy capacity and battery storage systems. The environments have also become more complex due to modifications in manufacturing processes. The situation is becoming even more complex with upcoming dynamic energy tariffs and the concept of energy communities.  

However, reliable energy usage forecasting is essential due to the pressure of rising energy prices. There are insufficient dedicated energy management experts capable of analysing energy data and identifying problematic situations, so companies are instead looking to establish ML and AI-driven digital solutions. 

Solutions

This project will establish an automated solution for energy load management that focuses primarily on energy support systems (e.g. HVAC), but which can also be used for other consumer types. The product MePIS Energy will be extended with an AI and ML model built out of consumption data and production plans. This will then be used to optimise energy usage. 

Working with Weiler Abrasives the aim is to create a solution that will be fast to deploy, easy to use and have all the necessary functionalities for a manufacturing environment. 

The comprehensive solution is expected to reduce annual energy consumption by 5-8% and optimise peak load by 10%

Expected results

Implementation of the solution will be done in the Weiler factory in Zreče, Slovenia. Peak load optimisation will result in significant cost savings for network tariffs and overrun charges. Core energy management modules will contribute to lower energy consumption and CO2 emissions.  

The implementation of the full solution, which will include the new AI and ML-based module, in the production environment will also contribute to Metronik’s commercialisation efforts.