SARATROF (Analytical System for the Aggregate Performance of Workers and Robots in Manufacturing Operations) is a research project initiated in 2021 that has aimed to develop a system to characterize the performance of companies through aggregated information from machines and workers, in order to increase the competitiveness of production based on information and communication technologies (ICT) and artificial intelligence (AI) and has been carried out in collaboration with the companies Ancora, Ledisson, Situm and Beta Implants.
This collaborative project is framed in the Conecta Hubs 2021 program, subsidized by the Axencia Galega de Innovación (GAIN) and has the financial support of the European Union (co-financed by ERDF Funds under the Feder Galicia 2014-2020 operational program). It also receives the support of the Second Vice-Presidency and the Department of Economy, Business and Innovation of the Xunta de Galicia.
The value provided by this project is to improve the productivity of companies by integrating the different information silos that usually exist in a company and that usually do not communicate with each other. The goal is that the information generated in each of the silos can be integrated into a single platform and analyze the data to make decisions to improve the productivity of the company from a more global point of view.
In this last milestone of the project, Ednon’s efforts have focused on optimizing the data acquisition and enrichment processes and, fundamentally, on improving the analytical processes, which has allowed to improve the accuracy of the Machine Learning models themselves and to validate the results with the continuous data collected from the Beta Implants plant. All the information obtained allows a better understanding of the factory operation and its performance based on different parameters.
With the work carried out throughout the project, the objectives of the project for which Ednon was responsible have been met:
- Deployment of an interoperable meta-platform.
- Execution of Machine Learning and data analytics processes.
This year, Ledisson has continued to collect data from all types of machines installed in the end-user (Beta). The biggest challenge has been on the FANUC 32i-B CNC machine, which has an integrated PLC and makes it impossible to extract data that way.
We decided to obtain information from the electrical impulses generated by this machine in its production process. We first reviewed the electrical diagrams to detect the most favorable and interesting signals for this purpose.
The only commercial item that is incorporated into the electrical cabinet of the CNC machine is a RevPi* with a digital input module. When we checked the electrical cabinet we could see that most of the signals were connected directly soldered on the PCB board, which made it impossible to bring these signals to a RevPi. With this in mind, when reviewing the electrical schematics we selected in a first screening 22 possible signals which we could connect to our RevPi.
Of these 22 signals, most were discarded for different reasons, but 5 signals did give a satisfactory answer. These 5 signals belong to 5 contactors installed in the electrical cabinet, which provide us with information such as each time the coolant pump is activated, when the spindle cooling is activated, when the finished part extraction belt is activated, as well as 2 contactors installed in the cabinet that did not appear in the electrical diagrams and could help us to discover their usefulness.
Once the RevPi was connected with its signals, it was verified that everything was working correctly and the data collection began.
* A RevPi is a miniature industrial PC and its function in this project is to translate the digital signals from the machine to a visual environment that allows us to see the data collected through a computer screen, which thanks to its Wireless connectivity allows us to consult the data from anywhere.
In this last year Situm has focused mainly on piloting the solutions developed during the project together with Beta to extract the maximum amount of intelligence from the production process.
In this pilot, the geoanalytics developed using real data have been put in value and the platform has been scaled according to the traffic of a real environment such as BETA’s production line.
Additionally, Situm has continued in the line of automatically detecting the scenarios in which the location system reduces its accuracy in order to indicate the possible need for corrective maintenance of the system.
New ideas have also been extracted from the pilot to continue improving the use of geolocation technologies in industrial environments for future projects, such as applying geolocation to materials and the use of trackers instead of cell phones for geolocation.
This project has allowed the acquisition of new knowledge to define new lines of work in the field of Industry 4.0. In addition, it has established a framework and working methodology flexible and modular enough to be extrapolated to other industrial areas. All this opens up a range of opportunities and future collaborations for the company.