Development of a simulation-based training methodology and framework built on Industrial IoT and machine learning.

Project Details

Project identification number: 2018-1.1.2-KFI-2018-00134

Beneficiary name: Training360 Ltd. (Consortium partner: e-Corvina IT Services Ltd.)

Project title: Development of a simulation-based training methodology and framework built on Industrial IoT and machine learning

Contracted support amount: 481,543,350 HUF

Support rate: 66.71%

Planned project completion date: December 31, 2020

Project Description

Training360 Ltd., together with e-Corvina IT Services Ltd., won non-refundable support exceeding 481.5 million forints in the “Support for R&D&I activities of small, medium and large enterprises” tender call. During the project with a total budget of nearly 721.78 million forints, the development of a new methodology and framework will be implemented.

Industry 4.0 is not just a buzzword anymore – we are making the first steps into a new reality, where computers and automation integrate in entirely new ways, enabling the use of artificial intelligence (AI). These steps may lead to the point where machine learning algorithms control, optimize, and fine-tune processes without human intervention.

Several key elements are required to be in place for this reality to emerge, ranging from autonomous robots to horizontal and vertical system integrations. This particular proposal focuses on three key representatives of the technologies transforming industrial production as part of the Industry 4.0 movement: data collection, data analysis using AI approaches, and decision support.

In this project, we aim to collect data from the industrial Internet of Things (machine sensors) and human employees. In addition to quantitative and qualitative analytics, we will also employ machine learning-based AI solutions to connect plant-level incoming data to management decisions.

A storyboard-supported simulation-based educational methodology and framework will be created, which enables better understanding and analysis of decision-making situations. Thus, this proposal not only discusses the optimization of decision making but also reflects on human resource training.

For the successful employment of machine learning algorithms, specific ‘learning sets’ need to be created with appropriate input/output data and parameters. Obtaining these learning sets is a complex task, as plant-level data needs to be linked to management-level decisions, and these links need to be tested as well.

The proposed project aims to overcome the above limitations by developing a generic simulation-based educational methodology that may use input sensorial data and human feedback from any industrial environment. These multi-role storyboards are then enacted by participants; feedback and decisions are recorded, and consequences are reviewed and analyzed.

Thus, the framework results in a growing database of specific scenarios and decisions based on sensorial input from plant level and related management decisions. This database can then be used to develop bots (programmed scripts) which may interact in different roles with other players.

As a consequence, the simulation may be played with plant-level workers, mid-level management, or senior leaders to test and develop appropriate reactions and achieve a real impact on improving the quality of decision making. The same database then serves as the input-output pairs for machine learning algorithms.

During the project, a specific plant and industry will be approached to obtain the necessary data for storyboard building. This specific organization will provide the sensorial and decision-making inputs and outputs, which will be further fine-tuned with close connection to the organization, and necessary infrastructure will be built.

Employing the sample storyboards, the simulations will be tested with the employees of the organization, recording responses and decisions at various levels. Consequently, a generic methodology and framework will be presented, which may be applied in different industries, linking IoT data collection, machine learning, and decision making with employee training and enhanced decision making.