Pilot project 2 – Digital twin with AI for sustainable and efficient energy management 

CHALLENGE! The objective is to construct a digital twin integrated with artificial intelligence that facilitates sustainable and efficient energy management by bridging the gap between energy producers and consumers. 

HOW? The proposed approach involves linking a simulation model with the actual physical world and leveraging AI to enable real-time energy management. 

WHY? The primary motivation behind this endeavor is to expedite and enhance the decision-making process when it comes to managing energy resources.

FINAL RESULT→ The desired outcome is the development and implementation of a functional digital twin equipped with advanced artificial intelligence capabilities to effectively manage energy consumption in the physical world. 

GOALS FOR INNO2MARE PROJECT: The INNO2MARE project aims to leverage digital twin technology, energy management, and artificial intelligence to promote innovation and sustainable development in the maritime industry. The project seeks to develop and implement digital twin solutions that can facilitate efficient and safe energy management in maritime operations, utilizing AI to optimize energy consumption and reduce carbon emissions. Additionally, the project aims to foster collaboration between stakeholders in the maritime industry to drive the development of new technologies and processes that can enhance the performance and sustainability of maritime operations. By achieving these goals, the INNO2MARE project will enable the maritime industry to meet the challenges of the future while promoting innovation and sustainability. 

Recent Developments and Progress

Experiments and actions on the pilot projects 2 until 31.12.2024

 1. Data Connection, Visualization, and Work Orders 

The connection to ISKRA’s database was successfully established, allowing for real-time data acquisition for energy usage across production lines and energy producers. Visualization of this data has been in progress, with the main focus on integrating energy data into the digital twin. Discussions focused on gathering and structuring the work orders for the production lines, ensuring the correct format and data structure needed for the digital twin were aligned. 

 The data format for work orders was reviewed and agreed upon. ISKRA presented the current work order report, which included the production line, date/time, and quantity of parts produced. It was agreed that ISKRA would extend this data to the other production lines and filter existing data to match the required format for the digital twin.

2. Anomalies in Energy Data 

Anomalies in the energy data, including missing values, abnormal spikes, and missing timestamps were addressed. DIGITEH presented the visualizations of the energy data and pointed out abnormal peaks in the data, which were attributed to missing timestamps. ISKRA explained that these missing data points were likely caused by counter shutdowns, where data was not recorded during these periods. 

A discussion followed regarding the implementation of a failsafe system at DIGITEH’s end to prevent the digital twin from receiving false data. It was agreed that such anomalies would be handled, and energy peaks caused by missing data would be accounted for. 

3. Advances in Digital Model 

Significant progress has been made in the development of the digital twin model. New algorithms were created to improve the simulation’s performance. Previously, the simulation always started with the first-time stamped data. Now, the algorithm checks for any new data, imports it into the model, and continues the simulation from the previous endpoint, enhancing the efficiency of the simulation. 

In parallel, the development of artificial intelligence (AI) and neural networks has started. Several neural networks were evaluated, and an LSTM (Long Short-Term Memory) network was selected for implementation in the simulation model. This addition will enable the digital twin to predict both energy usage and energy production, further enhancing the accuracy and capabilities of the model. 

Conclusions and Next Steps 

Considerable progress in visualizing energy data and addressing data anomalies was made. The work order data has been structured to meet the requirements of the digital twin, and predictive algorithms are being developed to enhance its capabilities. With the integration of AI, particularly the LSTM (Long Short-Term Memory) neural network, the digital twin is set to gain the ability to predict future energy usage and production, improving its performance. 

To further advance the project, the following steps are planned: 

  • DIGITEH will continue working on predictive algorithms based on the available data. 
  • Simulation algorithms will continue to be refined to improve their performance and adaptability. 
  • The LSTM neural network will be integrated into the digital twin, enabling predictive capabilities for energy usage and production. 
  • Management algorithms will be developed to facilitate energy usage and consumption over the whole factory. 

 

Figure 1: Visualisation of produced energy in Solar factory - Technomatix Plant Simulation.

Figure 1: Visualisation of produced energy in Solar factory – Technomatix Plant Simulation.

Figure 2: Visualisation of consumed energy in production line - Technomatix Plant Simulation.

Figure 2: Visualisation of consumed energy in production line – Technomatix Plant Simulation.

Figure 3: Block diagram - working of LSTM neural network.

Figure 3: Block diagram – working of LSTM neural network.

June 2023
N
June 2023

Progress 30.6.2023

December 2023
N
December 2023

Progress 31.12.2023

June 2024
N
June 2024

Progress 30.6.2024

December 2024
N
December 2024
June 2025
June 2025

Progress 30.6.2025

Lesson learnt and implementation strategy
Lesson learnt and implementation strategy

Lesson learnt and implementation strategy

Publications

A Robust Heuristics for the Online Job Shop Scheduling Problem

Hugo Zupan , Niko Herakovič , Janez Žerovnik ,  A Robust Heuristics for the Online Job Shop Scheduling Problem

Published in: MDPI , Algorithms 2024, 17, 568, https://doi.org/10.3390/a17120568

Abstract:

The job shop scheduling problem (JSSP) is a popular NP-hard problem in combinatorial optimization, due to its theoretical appeal and its importance in applications. In practical applications, the online version is much closer to the needs of smart manufacturing in Industry 4.0 and 5.0. Here, the online version of the job shop scheduling problem is solved by a heuristics that governs local queues at the machines. This enables a distributed implementation, i.e., a digital twin can be maintained by local processors which can result in high speed real time operation. The heuristics at the level of probabilistic rules for running the local queues is experimentally shown to provide the solutions of quality that is within acceptable approximation ratios to the best known solutions obtained by the best online algorithms. The probabilistic rule defines a model which is not unlike the spin glass models that are closely related to quantum computing. Major advances of the approach are the inherent parallelism and its robustness, promising natural and likely successful application to other variations of JSSP. Experimental results show that the heuristics, although designed for solving the online version, can provide near-optimal and often even optimal solutions for many benchmark instances of the offline version of JSSP. It is also demonstrated that the best solutions of the new heuristics clearly improve over the results obtained by heuristics based on standard dispatching rules. Of course, there is a trade-off between better computational time and the quality of the results in terms of makespan criteria.

Keywords: job shop scheduling problem; online algorithm; heuristics; simulation; digital twin; smart manufacturing

Skip to content