Condition monitoring and predictive maintenance of bearings applying IoT and cloud computing
Predictive maintenance of bearings is achieved through the analyzes of vibration signals. There are plenty of software available in the market that require a computer terminal for signal analyzing. However, under the umbrella of Industry 4.0 technologies -IoT, big data, machine learning, bearing monitoring and vibration signal analyzing should be done on IoT-Cloud based systems. Indeed, there are billions of bearings around the world waiting for being monitored, and if all the extracted data could be storage and subsequently analyzed in the same IoT Cloud-based system, the advantages will be countless. The massive real-time data can be used as an input to develop an interconnected system that performs a more accurate failure prediction. Neural network, machine learning and artificial intelligence are some of the tools that can be used to predict the health of a bearing in a more accurate way, and the system itself will learn and develop its own capabilities. Predictive maintenance will be more robust and reliable if an IoT Cloud-based system is used. Furthermore it will allow manufactures to run a more efficient maintenance plan. The aim of this project is to explore how predictive maintenance on mounted bearings can be achieved using Internet of Things and Cloud-based Systems.
Health analysis and condition monitoring of bearings
Rolling element bearings are widely used in rotating machines. Their condition drastically influences the performance of the whole machine or production line. Also in the construction and energy sectors such rolling element bearings play an important role. Therefore, the detection of bearing faults as well as wear monitoring have always received intense research interest. Practically, bearing faults evolve slowly and are hard to detect at an early stage, as wear processes in rolling element bearings are slow in nature. Most diagnostic techniques based on vibration analysis focus on the detection of major mechanical defects, rather than monitoring the actual bearing condition.
Vibration signal analysis is a common tool to detect bearing running condition. The effective vibration signal extracting method have a critical part in finding useful characteristic information for bearing condition monitoring and fault diagnosis. One parameter influencing the condition of rolling-element bearings is the radial clearance. The purpose of this project is to explore the relationship between bearing clearance and bearing vibration frequencies based on various methods.
Laser assisted low cost automation
Many small and medium-sized enterprises (SMEs) trust in the experience and skills of their long-term employees. Furthermore, a fully automated production is for many SMEs too capital-intensive and not flexible and versatile enough. In contrast, low-cost automation via intelligent assistance systems can create high-quality and ergonomic workplaces. At the same time, flexible and versatile production processes can be implemented.
The practice-oriented project concentrates on solutions for human-centric low-cost automation. Laser-based assistance systems are being used. They are cost-efficient and have a high degree of flexibility. They are interesting for applications in many industries due to their specific characteristics. The research work focuses on three dimensions:
- Human-machine: The acceptance of modern operating concepts through exemplary implementation and verification in case studies and interviews.
- Machine-Human: The ergonomically sensible presentation of digital content using laser assistance systems.
- Machine-Machine: The integration of assembly systems and/or process monitoring in production systems as well as the realization of cloud-based solutions for continuous process optimization and product quality.
The project is part of the Digital Factory of the Leuphana Digital Transformation Research Center.
diZI-FTS
The aim of this cooperation project with ek robotics GmbH is to develop a method for virtual commissioning of automated guided vehicle systems (AGV systems) with the help of artificial intelligence (AI).
One of the focal points is the creation of digital twins of the AGVs' working environments with relevant objects and ideal paths between such objects with which the vehicles interact. Specifications and parameters resulting from the respective vehicle properties and capabilities that are relevant for this purpose are modeled beforehand and are taken into account.
The interactions of the vehicles with their environment require a high positioning accuracy. In order to be able to map these high requirements in the digital twin and use them in the course of virtual commissioning, the twin of the operating environment is derived from a high-resolution 3D point cloud of the final operating location at the respective customer. All production and logistics processes of the real AGV system are then created and validated in a simulation application with the respective vehicle models in this virtual customer environment. Possible challenges and problems can thus be identified earlier and resolved before the actual commissioning. Likewise, hidden optimization potentials can be identified and implemented. This leads to increases in quality as well as efficiency in system integration.
The focus of research in this project is not only on the creation of suitable digital abstractions and twins, but also on the development of the processes required for their creation. The use of AI as a tool in the areas of 3D computer/machine vision and path recognition/creation in multi-agent systems plays a critical role in accelerating the creation of such twins, thereby moving them into the commercial realm and ultimately making them usable. However, since the working environments of AGVs often present special or even unique qualities, such as in the case of object detection, relying only on existing and widespread datasets for use in conjunction with AI is not practical or purposeful. Lack of data availability as a whole is not only problematic in this specific case but represents a fundamental problem for many practical applications of AI. Possibilities for automated and fast generation of synthetic training data as well as their use for 3D object detection and path planning in this specific case represent an applicable solution and therefore find special attention.
The project is supported by the European Regional Development Fund.
Current insights, documentation on the project and useful tools can be found here:
https://github.com/romankraemer