The SelSus project adopts a systematic approach, supported by a well-defined work plan. The work plan comprises nine carefully defined work packages. In order to guarantee fully committed teams towards comprising goals, the number of individual work packages is kept clearly constrained.
In order to hear every member's point of view on their work package, they all answered a few questions. Read more below:
The aim of WP1 is to ensure the coherence, technical compatibility and industrial validity of the various SelSus developments across all the other technical and demonstration work packages. To do this, WP1 has captured the detailed industrial requirements, has defined and will continue to refine the overall SelSus system architecture, is monitoring the business motivation, and will benchmark and validate the final results of the SelSus demonstration activities.
The biggest challenge is to ensure a continued good match between the technical and commercial requirements and the different technical innovations created by the project’s multi-disciplinary team.
WP1 was instrumental to create a common architecture at the beginning of the project and mapped all the interactions between the technical developments. Towards the end of the project the focus is shifting now to capturing the technological decisions made in the project and benchmark the performance during the demonstration activities. As such, this work package is at the very center of the project ensuring and documenting the integration of results.
While WP1 is instrumental for the success of the SelSus project, the approach is based on well-established requirements engineering and system architecture definition and management approaches. The resulting architecture naturally reflects the core innovations of the project and has been deliberately designed to be fully compatible with the emerging specifications for Industrie 4.0 (I4.0) and Cyber-Physical Systems. The SelSus architecture is in fact the first to report an I4.0 compliant approach for the diagnostics, prognostics and maintenance management of complex, one-of-a-kind production systems.
The experience from previous projects such as Transparency, IDEAS, EUPASS, and EXPRES were invaluable to inform the SelSus architecture based on previous experience and best practice examples. It also helped consider the wider context of SelSus within the smart manufacturing components and system domain. Especially the diverse experience from different previous projects and industrial practices helped create a more mature architecture.
So far, the detailed industrial requirements and constraints for the proposed SelSus system have been captured and analysed. These have been translated into the initial SelSus system architecture covering both the smart component/device internal and production system level. This architecture defines the required functional components and their interactions from the low level data capture, interpretation, and reasoning on device level to the higher level maintenance scheduling and decision making.
To create and demonstrate the feasibility of the proposed plug-and-produce cyber-physical production system concept for improved diagnostics, prognostics and maintenance decision making and encourage industrial uptake.
The project is about increasing an OEE (Overall Equipment efficiency) in the factory by trying to avoid unexpected maintenance activities and unexpected falls of components on different levels. On the component level this goal can be reached by implementing intelligent self-healing and self-monitoring functionalities. Device or component is capable to analyze the process it is involved in and identify the origin of existing or potential failure, which could hamper the production. We have similar goal on system level (e.g. production line) where a problem may be caused not only by the component but also by the relation of involved components.
The aim is the development of intelligent self-healing and self-aware devices, which are capable of monitoring the process they are involved in, analyzing critical conditions, and dynamically predicting potential failures. Such devices will be capable to automatically recover or switch to safe mode preventing total failures. We work on innovative solutions to integrate such devices into a larger system with intelligent self-healing capabilities.
There are a number of different challenges. Within my work-package there are different components produced by different suppliers who implement self-healing and self-monitoring features with very different behavior. It is challenging to find out general mechanism/algorithm for self-healing and self-monitoring functionalities and reach synergy between different organizations and their developments. In order to be able to integrate various devices into a self-healing system general architectural guidelines are needed.
We are predominantly working on component level, on shop-floor. The information that those components provide is the basis for the algorithm of condition monitoring on system level. Intelligent components form basis of the whole system. Components provide the data, the KPI for the decision support.
Innovative approach is required indeed. We established common component architecture, which is termed as SelComp. Different components share common SelComp functionalities and are implemented similar to the innovative paradigm of the “Smart components”. Also, we implemented a number of sophisticated algorithms, e.g. for identification of collection of data sources from different equipment, analysis of these data, and finding correlations. These algorithms help to identify the origins of system process problems. Furthermore we implemented decision support module based on component network, which is, to my knowledge, something that is not implemented so far. These are the major innovations: common architecture, data mining tools, and network on the component level.
Well, we did a lot for intelligent components system on shop floor level. The project “XPRESS” is the basic project which gave us important input and where we discussed really basic architecture. We always try to extent the existing equipment with new sophisticated functions, such that also old equipment can be enriched with new intelligent functionalities. This is basic evolution principle which we followed also in IRamp3 and ReBorn projects. Intelligent equipment always need communication algorithms to the system level, you need self-description capabilities of components. These are the solutions that we developed within other projects.
One of the first things that we have developed was the common architecture, and this is really a big step towards ability of coherent integration of components into system. These components use the same description, communication mechanisms, and architecture. Presently we are targeting to the unification of architecture prototype where the developments of our SelSus partners will be included. It will be presented at 36 months meeting and finally refined.
I can provide impact of SelSus on example of our company. We can reach three main goals:
* Development of a decision support tool, which helps our internal service department to identify a problem at the customer side at very early stage and to help them with maintenance and repair.
* Development of controlling and analytical tools for our customers, e.g. for quality or for various local process related problems. They can have convenient overview over their production line, optimize it, and prevent failures.
* Implementation of limping home mode, which allows usage of equipment in safe state even if some of the components are broken. This allows one to gain valuable time and continue the production until service will be possible and the equipment will be repaired or recovered.
The purpose of WP3 - Multi-modal data acquisition and analysis is to take advantage of Wireless Sensor Networks technology to acquire additional information from machines, and therefore create innovative and advanced sensing systems suitable for industrial purposes. This sensing system / platform is called Sensor SelComp.
Create innovative sensing platforms with embedded capabilities and functionalities to ease the acquisition and interpretation of data for condition-based monitoring, optimization and process reconfiguration purposes. Additionally, these sensing platforms should be external to the machine, easily integrated into existing equipment, with zero (negative) impact on the manufacturing process and capable to extend the equipment capabilities.
It enables new process measurement capabilities that are not initially available in the shop-floor equipment, and allows for on-the-fly reconfiguration of data processing used for diagnostics and decision making. It is also an enable for the implementation of self-health monitoring, self-diagnosis and renovation management.
Yes, the work was oriented by an approached grounded on agile methodologies leading to disruptive innovations. There are two major innovations: 1) Dynamically create a parser capable to interpret new integrated sensors and automatically acquire data from those sensors; 2) On-the-Fly reconfiguration of SelComp data processing.
Experience about the use of Wireless Sensor Networks in industry knowing what is relevant for the field, and a set of suitable skills that improved the used methodologies to develop better and more applicable solutions for the domain.
6. What did you reach until now in your work package / overall project? (interim result)
Successfully develop the Sensor SelComp equipped with innovative technologies that can be used in the project demonstrators such as Dynamic Modular Software Reconfiguration and Flexible Sensor Integration.
7. What is the goal of the project for you?
Develop solutions that are both scientifically relevant and suitable for industrial use. We want to contribute for the advance of scientific knowledge in the field but also to develop and demonstrate solution that the industrial partners can use to increase their competitive advantage.
1. What is the aim of your work package?
The aim of Work package 4 is to develop a set of models supporting diagnosis and degradation prediction including self-diagnosis models and algorithms for renovation and repair. A hierarchical approach based on the use of object-oriented Bayesian networks is taken in order to support the development of component based models for diagnosis and degradation prediction which can be integrated into a wider system level model. Initial models are constructed based on information found in the literature and expert knowledge while algorithms for continuous learning from operational data are being considered.
2. What are the challenges of your work package?
Main challenges are to collect the necessary information to construct reliable and robust models for diagnosis and degradation prediction and the test and evaluation of such models.
3. Which impact does your work package have on the overall project?
The diagnosis and degradation prediction models play a significant role in achieving health monitoring and capability management for self-sustainable manufacturing systems.
4. Did you work on your work package with an innovative approach? If yes, what was the innovation?
We used the framework of object-oriented Bayesian networks to capture and represent knowledge and experience on the systems being modelled. These models are excellent for capturing uncertainty in a problem domain.
5. How can you benefit from your previous projects for working on SelSus?
Yes, it is important to be able to interact and communicate efficiently with subject matter experts and technicians in order to have an efficient and effective model development process. HUGIN has benefitted from working closely with subject matter and technicians in previous projects.
6. What did you reach until now in your work package / overall project? (interim result)
We have successfully constructed and evaluated a number of diagnosis and degradation prediction models for components. We are in the process on integrating such component based models into wider system level models. We have developed a methodology to support the model construction and are considering algorithms for continuous self-learning from operational data.
7. What is the goal of the project for you?
The main goals of the project from HUGIN’s point of view is to be involved in the development of a number of use-cases demonstrating the capabilities of Bayesian networks that may after the end of the project be further developed into operational solutions, collaboration with potential users of HUGIN software, HUGIN software prototype enhancements tested on real applications and networking with potential partners of new projects.
The aim of work package WP 5 is the development of a decision support system to find an optimal strategy for maintenance and repair activities related to components and systems in a manufacturing environment. This optimization is accomplished by using several KPIs (key performance indicators) like the expected lifetime or the degradation of specific components. The decision making process is supported by Bayesian models - such as Bayesian networks and influence diagrams - as well as by a discrete event simulation model, which will both contribute to find an optimal remedy strategy to guarantee upmost yield expectations.
The major challenge is building up the models both for decision making and simulation in a way that they reflect the real manufacturing environment appropriately.
The main impact of the work package is reduced downtime and higher productivity of the manufacturing process by smart decision making concerning maintenance and repair activities.
As uncertain knowledge in the domain can also be covered by the deployment of influence diagrams for decision making, it is much preferred to the classical decision trees. In addition, the combination of the simulation system with Bayesian networks to update its probabilities is also an innovative approach.
Experience in the field of uncertain reasoning, especially with the Bayesian paradigm, as well as projects regarding reliability theory and degradation processes were beneficial for the SelSus project.
We already successfully built up decision models for several components, which will be refined and optimized during the rest of the project duration.
The overall goal of the project is a contribution to a higher Overall Equipment Effectiveness (OEE) in manufacturing systems.
The aim of WP6 is to develop a framework to support the design of long-living production systems through integrating availability, degradation and maintenance aspects into the design process; and consequently enabling the system designers to maximise the diagnostic capabilities of custom designed machines and systems.
The biggest challenge is to ensure that the heterogeneous information sources being considered in the project are well reflected in a coherent common data model that ensures interoperability and information reuse.
WP6 represents the main interface between the SelSus project environment and the system design domain (CAD, PLM, FMEA, etc). This makes it important to put in place a common data model that represents the various types of data in the project environment. This common data model activity interacts with all of the development and demonstration tasks in the project. Moreover the work package also is providing methods for realising the information/knowledge interface between the design phase and the system operation phase where the majority of SelSus’ developments take place.
The work package is developing a common data model as well as new methods for standardised FMEA generation taking into account the provisioning of availability indicators as part of the FMEA exercise during the design phase. In addition to that a new method for generating Bayesian diagnostic models from semi-structured domain knowledge such as FMEA is being developed in the work package along with new approaches for using the maintenance operational records for the validation and tuning of diagnostic models.
The experiences gained from previous projects such as Transparency, IDEAS, and EXPRES were important in informing the SelSus data modelling approach. Also the link between operational-phase knowledge and design phase decision support draws from previous experiences and lessons learnt from previous projects such as TRANSPARENCY.
So far, the common data model has been developed covering most of the concepts used in SelSus, and this model has been instantiated for the component-level demonstrators. FMEA has been generated for three different demonstrators with enhanced availability information included as well. Moreover, a method for generating Bayesian diagnostic models from structured FMEA records is being tested and demonstrated using an integrated software user interface.
To demonstrate the added-value and feasibility of utilising structured design domain knowledge sources for easier provisioning of enhanced diagnostic and availability capabilities in complex manufacturing systems.
To identify a host component, system and production facility to install sensors into for the demonstration of the enhancements this can give to condition monitoring and self-diagnosis. Then, to design and complete the installation of those sensors, and begin extracting data for analysis.
The challenges have been finding suitable machines, incorporating such a wide variety of sensors in complex machines without disrupting their work, processing the data in such a way that it meets all partners’ needs and the expectations of the project.
It is the most visible evidence of how the work we’re doing on Selsus would be realised in a production environment, and it gives everyone a chance to witness the physical installation and real-time data.
We went back to the very basics in our analysis of the machine, and created a completely new machine FMEA and built on this with boundary diagrams representing the critical sub-systems. This isn’t in itself innovation, but as a final customer and the associated members of the team it was enlightening to think in such a focused way on failure modes, consequences and prevention techniques in such fine detail.
I’m new to research projects, but my experience as an engineer at Ford allowed me to quickly identify the processes and operations which would benefit from the technology being discussed. This should make the results of the project relevant to us and other industrial customers.
We are scheduling the installation of sensors in the RTV machine in the UK this month, and should immediately start transmitting data to the cloud
To prove and demonstrate that we can make genuine improvements to machine availability in Ford by improved capturing of information, and taking advantage of the self-diagnosis and healing capability of our flexible machines.
WP2 is the central work package for the development of smart devices and machines capable for condition monitoring and self-healing. The devices developed within the work package build a new generation of intelligent equipment on shop floor which supports the main idea of SelSus for increasing the efficiency of manufacturing by in-situ repair and self-healing processes and components. The main target of WP3 is the development and integration of advanced sensor systems with embedded intelligence for data capture and interpretation within an assembly station. To assist in selecting the right maintenance, renovation and repair strategy a decision support environment is needed. Therefore, WP4 focus on developing a new diagnostic approach/concept focused on detecting potential operating inefficiencies, through utilizing various types of sensory information like energy consumption, vibration, pressure, and temperature that also integrates life cycle assessment strategies into the classical availability-based predictive approach to maintenance, renovation and repair activities. To be able to enhance the fab planning, ramp-up and optimization during production as well as during periods of renovation and repair, WP5 provides the possibility to simulate the impact of modification in production environments based on enhanced LCx data, gathered from real production. A decision support system assists in defining the optimal strategy for optimizing productions based on ROI calculations of simulation results.
WP6 develops a Core Data model to capture the mechanical and functional capabilities of dedicated production resources such as assembly machines linking fragmented information of machine components into one holistic model. This allows designers to check and validate their design proposals at an early stage.
All of these RTD work packages stay close connected and therefore share their knowledge among each other to ensure a real collaborative project with the aim of an optimal overall efficiency and effectiveness.
Taking up these five core developments, WP7 implements all these innovations and demonstrates and validate the approached of Self-Monitoring Intelligent Manufacturing Devices, Self-Healing Production Systems and Decision Support Systems for Optimizing Large Production Systems.
All these research and development efforts has been kicked-off and embraced by a work package (WP1), which cares for solid foundation of industrial requirements and business models, but also for industrial validation.
The external Industrial Advisory board not only gives valuable input to the industrial requirements gathered in WP1, but also serves as a spring board for the Standardization, Dissemination and Exploitation efforts (WP8) of SelSus. By organizing dissemination activities not only at the end, but also intermediate within the developments, it promotes iterative RTD and incorporation of industrial feed-back into the developments.