Maintenance operations are currently scheduled by the customer based on the preventive maintenance indications provided by the COMAU manual. The scheduling is contained in a machine ledger, currently in an excel file version. To manage work orders, customers use ERP software. Maintenance operations are planned during production stops to avoid downtime, therefore, machines are over-maintained. Condition-based maintenance can contribute to reduced machine breakdowns resulting in a reduction of maintenance costs and an increase in equipment availability.
A COMAU’s industrial robot has been chosen as the objective of the analytics pipeline since robots are the most common types of machinery in the COMAU’s production lines. In the scope of the project, the analytics efforts have been focused on a precise component of the robot, which is common to many other industrial robots and has an important impact on both robot precision and capabilities, the transmission belt. As it is easily conceivable, having to deal with an industrial robot and its overall complexity is tricky, since there are many different physical and environmental conditions that all contribute in a different and not completely predictable way. To deal with that complexity and with the fact that for having an efficient classification model a labelled dataset is strictly required (containing as well, fault data), it has been decided to focus the study regarding the transmission belt on a test-bed made of just one of the six industrial robot’s axis, the sixth. A system that continuously changes the distance between the motor and the gearbox reducer has been designed and developed, simulating the changes in the transmission belt, tensioning from too low (simulating the loss in performance after too many working hours without maintenance) to extremely high (representing a wrong belt installation from the maintenance crew). That system is made of a slider and a millesimal comparator to precisely monitor the actual distance and thus, the current class that the testbed is representing. A dataset is thus created, gathering data from two different testbeds, containing 5 comparable classes each, to test the model generalization capabilities.
The COMAU demonstrator includes the local deployment of the SERENA platform and its subsystems, integrating all services as Docker containers in the local ICT infrastructure of COMAU. To better study, the belt tensioning problem, a testbed named RobotBox has been developed replicating the key aspects of an industrial robot.. The demonstrator consists of the data acquisition process performed via a local gateway (4G and 5G compatible), predictive analytics as well as the scheduling of maintenance activities and remote operator support to perform the required maintenance activities. In particular, data acquisition is delegated to a REST service on NiFi listening on a specified port for POST requests. Furthermore, the Predictive Analytics service enables the following key tasks:
- Model building triggered by either specific user input or an output of the self-assessment service. It produces as output the prediction model.
- Prediction task triggered by a new robot cycle and generating a prediction label along with a RUL value.
- RUL value triggers the scheduling Service for assigning maintenance tasks to site personnel.
- Real-time visualisation of the monitored equipment superimposing key information is facilitated via a Unity3D engine platform.
The SERENA platform has been working on the COMAU’s premises for more than 14 months, demonstrating its architectural capabilities in terms of distributed computing, reliability, and security since it is working according to the COMAU’s internal network and security policies.
The knowledge and the technologies used to develop the SERENA platform have raised an internal interest in COMAU, which has led to the re-use and extension of many of the SERENA concepts in the COMAU’s IIoT offering. The analytics pipeline has been proved effective not only from the prediction point of view but also from the architectural deployment, giving the possibility to train the model in the cloud and then decide if keeping the prediction as well hosted in the cloud or if deploying it to the edge level. This is a key feature since it gives the possibility to choose time by time what is the best way to deploy the intelligence, not only from a technical point of view but also from business, enabling, as an example, analytics as a service scenario. As happened for the knowledge build, thanks to the SERENA architecture, most of the analytics ideas and procedures were the starting point of COMAU internal integrations and extension.
Furthermore, the SERENA platform has successfully proven that the platform-as-a-service is as efficient as flexible guaranteeing the interworking between services, developed from different companies/universities, just relying on the common data model and REST calls. Indeed, data collection, model prediction, visualization, and then maintenance schedules are all strictly linked together, with consistent and always updated data overall the platform. Finally, the VR maintenance procedure has demonstrated the supposed effectiveness, providing augmented and immersive instructions to perform tasks, even by non-expert personnel. Besides, the digital twin service has proven the technical feasibility of gathering robot real-time data and showing it remotely.
The benefit for the customer is the improvement, in terms of maintainability, of industrial robots exploiting all their components for their entire useful life without over-maintenance activities and with reduced breakdowns. This will in turn facilitate reduced maintenance costs and extension of the robot’s operational life. The AR solution may speed up remote support for maintenance and the robot will be able to return faster to its operation.
| Country: | IT |
| Address: | GRUGLIASCO 10095 |
- No items published
- Not specified
The SERENA project has provided a deep dive into an almost complete IIoT platform, leveraging all the knowledge from all the partners involved, merging in the platform many competencies and technological solutions. The two main aspects of the project for COMAU are the container-based architecture and the analytics pipeline. Indeed, those are the two components that more have been leveraged internally, and which have inspired COMAU effort in developing the new versions of its IIoT offering portfolio. The predictive maintenance solutions, developed under the scope of the project, have confirmed the potential of that kind of solutions, confirming the expectations.
On the contrary, another takeaway was the central need for a huge amount of data, possibly labelled, containing at least the main behaviour of the machine. That limits the generic sense that predictive maintenance is made by general rules which can easily be applied to any kind of machine with impressive results.
More concretely, predictive maintenance and, in general, analytics pipelines have the potential to be disruptive in the industrial scenario but, on the other hand, it seems to be a process that will take some time before being widely used, made not by few all-embracing algorithms, but instead by many vertical ones and applicable to a restricted set of machines or use cases, thus the most relevant.