Manoranjan Kumar, Joel Cramsky
Mauro Caporuscio, Lars Håkansson, Diego Perez
Linnaeus University; Volvo Construction Equipment (VCE), Braås, Sweden
Volvo CE, the Knowledge Foundation (KK-stiftelsen)
14 Sept 2020–31 Aug 2025
Computer and information science (Department of Computer Science and Media Technology, Faculty of Technology)
More about the project
The vision of Volvo CE is to virtualize all products, systems and components, so that physical prototype verification can be avoided completely. Instead, Volvo CE aims to predict and verify their products’ performance in the virtual space. The company aspires to create a so-called digital twin for every machine (physical twin) that they sell. It is fed with data from its physical twin through sensors, edge-based algorithms and data analytics (IO-1).
Digital twins are expected to generate increased sales and affect site optimization, requirement engineering, product development, and after sales. For the latter, the focus is on predictive maintenance and customer support for optimized operation, including fleet management. Each digital twin will consider quality deviations from the nominal specification, and thereby define its own capacity.
Edge-based analytics is used to process sensor data for each physical twin and send it to the digital twin for cloud-based data analytics. There, algorithms frequently and automatically will calculate remaining component life and plan for maintenance on both machine and fleet level.
The following tasks are identified so far in achieving the goal and would be carried out during the research:
- Develop and verify virtual models of machine behaviour and performance, thereby develop the vehicle simulation model representing real machines. This development also includes a variety of driving cases in simulations. Further, verify the simulation model based on loads generated within different components in normal and extreme working conditions with real measurements.
- Predict customer usages based on artificial intelligence (AI) algorithms using simulation models. For example, which combinations of known test track simulations best suite a particular customer environment, like road roughness or topology.
- Customer services can be defined related to productivity, fuel efficiency, remaining useful life of machine components etc. Hence evaluate the existing data usages to identify what kind of customer service can be developed using existing data logs. Thereby use machine learning approaches that could be a combination of supervised and unsupervised learnings. If existing data logging does not support those customer services, then assist in log implementations for the future machines.
- Use the artificial intelligence (AI) algorithm in predicting the repair and maintenance schedule for every machine based on their usages and environments, by using new/existing logged data and component condition status from inspections for every machine.
- Implement a visualization tool, connecting machines' data and their coupling to machine learning approaches. Hence visualization will support in finding anomaly within populations of machines and customer support can be extended on real-time basis.
The project is part of the industry graduate school Data Intensive Applications (DIA) and the research in the Data Intensive Software Technologies and Applications (DISTA) and Linnaeus University Centre for Data Intensive Sciences and Applications (DISA) research groups.