Embedded Files
Development of a discrete event simulator for predictive maintenance planning
Development of a discrete event simulator for predictive maintenance planning
In my diploma thesis, a laboratory production line based on the processes of a dairy company was preventive maintenance schedule of a dairy production line at the Industry 5.0 laboratory of the University of Pannonia, for which I developed a simulation model based on SimPy (Python). In the first chapter, I will describe the literature review, followed by the methodology and then the content of my thesis, which I have divided into four parts. In the first, I will define the parameters that will allow the Ipar 5.0 laboratory to be used as a dairy simulator. This includes the selection of the products to be produced, the determination of the number of products to be produced, the packaging dimensions and the parameterisation of the working time of the machines. In the second part, I will describe in detail how the fail-safe SimPy simulator of the Ipar 5.0 laboratory works, what I had to take into account for which machines and how I ensured that the programme followed the process correctly. In the third part, I describe the data that is important for modelling revenues and costs, as well as the parameters related to breakdowns and maintenance, and where I got them from or how I calculated them. Finally, in the fourth section, I describe how I determined the maintenance date to avoid a particular failure, and I use a Monte Carlo method to determine a maintenance schedule using this greedy approach that takes into account the costs associated with maintenance and failure.
Workspace design and monitoring framework to evaluate the collaboration between human-machine
Workspace design and monitoring framework to evaluate the collaboration between human-machine
Maintenance represents a significant part of industrial companies. The vast majority of maintenance costs arise from inappropriate maintenance strategies. Choosing the right strategy and optimizing maintenance can significantly reduce both maintenance and production costs. However, achieving this goal is out of reach for most companies because they lack the appropriate technological systems. Currently, the cost of acquiring an adequate level of technology is so high that it is not worth the investment for smaller companies. This issue can be addressed through Brownfield Industry 4.0 developments or so-called retrofitting solutions, which provide a cost-effective way to integrate "not-smart" tools, machines, and equipment into the industrial network. This integration enables real-time monitoring of production parameters, allowing for the optimization of maintenance processes. For this purpose, a digital twin representation of the entire process would be the best option. However, a true digital twin is currently not feasible in a cost-effective manner. Therefore, we will propose a Digital Shadow for maintenance optimization. The aim of our research is to build a data acquisition unit (DAQ) that can provide a solution for mapping digital and analog signals on the shop floor in near real-time. The collected data will be used to create a Digital Shadow of the system, allowing us to examine the system's status in near real-time and dynamically manage the selection of the maintenance strategy based on its current condition.
Workspace design and monitoring framework to evaluate the collaboration between human-machine
Workspace design and monitoring framework to evaluate the collaboration between human-machine
Industrial development is marked by a pronounced shift towards human-machine collaboration, a trend exemplified by the beginning of Industry 5.0. This evolving industrial paradigm prioritizes human-centered solutions, emphasizing resilience and sustainability. A vital part of this collaboration is the need for robots to possess more complex cognitive abilities, facilitating safe co-working environments with humans. Technological advances have enabled diverse methods for monitoring human behavior. Analyzing these behavioral patterns enhances efficiency and collaboration.
This study introduces a setting where monitoring the operator and the robot is possible. It is achieved through a camera, an indoor positioning system, and wearable sensors. The collaboration can be represented by indicators (utilization of participants, level of stress based on physiological measurements of the person).
Targeted measurements were conducted, and the results underwent comprehensive analysis. The objective is to bridge the gap between human activities and collaborative robots and design their service to human needs. The developed methodology is presented in a complex demonstration game and a design of experiments is shown.
Development of a discrete event simulation model for human-machine collaborative work
Development of a discrete event simulation model for human-machine collaborative work
The advent of collaborative robots has made it technically feasible for humans and machines to work together in a shared workspace. A fundamental spirit of the fifth industrial revolution is to develop human-centred solutions and processes. This is necessary because full automation has its limits, both economically and technologically.
The aim of my research is to develop a discrete event simulation-based toolkit to efficiently and purposefully investigate the efficiency and feasibility of these collaborative processes. In the developed simulation model, I use key indicators to investigate collaborative and decoupled manufacturing between humans and machines. The simulation will make it possible to investigate more cost-effective improvements without interrupting real production. My work will show how to reduce the resource requirements of manufacturing and maximise the utilisation of robotic arms and workers. For a specific automotive supplier, I will prepare a case study for a specific workflow, where I will evaluate the workload of a given collaborative robot and operators.
Risk-based maintenance schedule based on Markov-chains
Risk-based maintenance schedule based on Markov-chains
The equipment and tools used in manufacturing have a lifetime, during which they need to undergo a number of maintenance operations. When we want to optimise the number of maintenance operations, we have to take into account which costs more: frequent maintenance or the cost of repairing the breakdown and the downtime it causes. In complex manufacturing systems, especially in the automotive industry, this scheduling task challenges traditional optimization algorithms.
Optimising the number of maintenance operations is therefore a key productivity driver to avoid downtime caused by unexpected failures, minimise downtime caused by maintenance and maintain acceptable product quality. The problem is that we do not know when a part or all of the machine will fail.
In my work, I work with machine event log data where only machine events are recorded, no diagnostic data. I use this information to build a Markov model of machine manufacturing and changeover processes. My goal is to create a model that can optimize the scheduling of preventive maintenance. I will use a Markov chain to handle the risk probabilities, the optimization and a population-based algorithm to solve the optimization.
Collaborative work scheduling between humans and robots
Collaborative work scheduling between humans and robots
Industrial development is increasingly focusing on human-machine collaboration. As Industry 5.0 is being introduced, more human-centered solutions are presented. The key elements of Industry 5.0 are resilience and sustainability, and it is also more human-centric, concentrating on collaboration between people and machines. Collaborative work requires that the robots have more complex cognitive abilities. These abilities allow humans and robots to work together safely near each other.
However, safety is only one part of human-robot collaboration. Scheduling the tasks of the operator and robot creates an environment where collaboration is safely achievable. Sharing activities and defining and scheduling these are complex tasks requiring monitoring and recognizing the operator’s actions. Thanks to technological advancements, several methods exist to monitor human behavior, such as cameras, indoor positioning systems, and wearable sensors. The patterns can be analyzed, and the result can be used the enhance efficiency and collaboration.
This work presents a workspace where monitoring the operator and robot is achievable using a camera, an indoor positioning system, and wearable sensors. Targeted measurements were made, and the results were analyzed. The developed methodology and optimization tools are presented in a complex demonstration game, and finally, will be detailed on how they can be applied in a real industrial environment.
Estimation of the operator comfort level and the layout information based on sensor fusion techniques
Estimation of the operator comfort level and the layout information based on sensor fusion techniques
The most significant problem for manufacturing companies today is human resource shortages. The problem with industrial workplaces is that the operators work long hours on uncomfortable shop floors (noisy, bad condition of air quality) that are not designed individually for the operators. We can get information about these parameters by measuring the environmental data on the shop floor. These aspects have an impact on production performance. The so-called comfort level of the operator can be determined by fusing different sensors. After all, the more comfortable the environment is, the more human worker can be productive, and the company can have a more comfortable workspace.
Our research aims to develop a framework that uses sensor fusion techniques to create a detailed layout. A mobile robot is designed to evaluate the comfort level of the shop floor zones. The mobile robot has different sensors. Thus we developed a so-called mobile measuring unit. We use temperature, noise, and humidity sensors to measure the environmental parameters to derive the comfort level. Besides, the indoor positioning system tracks all resources (including humans) during production. We will use 3D LiDAR in addition to the indoor positioning system to achieve more accurate positioning. Also, the robot provides a visual image (camera) of the inspected area, which will fuse with the LiDAR point cloud to provide a more accurate visual monitoring of the production. The developed mobile measure unit with the sensor-fusion algorithm is tested in the Operator 4.0 laboratory at the University of Pannonia.
Workspace design for human-machine collaboration monitoring
Workspace design for human-machine collaboration monitoring
The human-machine collaboration is coming to the forefront of industrial development. As the concept of Industry 5.0 is being introduced, more human-centered solutions are proposed, as opposed to the more machine-centric solutions in Industry 4.0. The main focus of Industry 5.0 is flexibility, sustainability, and the mentioned human-machine col- laboration. To create a more human-centric workplace, the workers’ behavior should be monitored. Technological advances have allowed new methods to monitor human behavior in the past years. For example, people’s motions can be tracked using the help of camera systems or indoor positioning systems. The patterns can be analyzed, and the result can be used the improve efficiency and collaboration.
However, this has several shortcomings. The goal is to prepare the collaborative robot based on human activities so that they serve people.
This work presents a workspace where the operator and robot can be observed using a camera and an indoor positioning system during their collaboration. The developed monitoring space and tools are presented in a complex demonstration game, and finally, it will be detailed on how they can be applied in a real industrial environment.
Page updated
Google Sites
Report abuse