Periodic Reporting for period 1 - AI-PROFICIENT (Artificial Intelligence for improved PROduction efFICIEncy, quality and maiNTenance)
Reporting period: 2020-11-01 to 2022-04-30
Different sectors of manufacturing and process industry are characterized by the need for adaptive configuration of the control parameters and rapid response to fluctuating production demands (sometimes intraday or even intrahour). To answer the production responsiveness demands, the industry requires innovative solutions for highly adaptable and flexible production systems. This implies the need for reconfigurable production systems, relying on, in the early production stage, fast ramp-up processes to reach full production capacity for new products and, in the mature production stage, an optimal operation execution at the plant level, while considering component-level degradation and required maintenance. To cope with such a challenge, introduction of advanced AI technologies in manufacturing plants is a must. Digitalization of factories is bringing several technological advances that facilitate the fulfilment of these requirements. To start with, there is an increased amount of sensor data and operation information that can now be deeply explored and exploited by AI systems: from reactive to proactive management. Furthermore, in order to embrace the upcoming Industry 5.0 it is necessary to enable effective ways for human intelligence to work in harmony with AI systems.
In line with these AI-based production challenges, the AI-PROFICIENT project which is a Research and Innovation Action (RIA), aspires to bring advanced AI technologies to manufacturing and process industry, while improving the production planning and execution, and facilitating the collaboration between humans and machines. So, main objective is to bring the advanced AI technologies to manufacturing domain through an evolution from hierarchical and reactive decision making to self-learning and proactive control strategies, underpinned by predictive and prescriptive AI analytics at both component and system level, by cross-fertilizing edge and platform AI, while leveraging the human knowledge and feedback for reinforcement learning.
In line with these AI-based production challenges, the AI-PROFICIENT project which is a Research and Innovation Action (RIA), aspires to bring advanced AI technologies to manufacturing and process industry, while improving the production planning and execution, and facilitating the collaboration between humans and machines. So, main objective is to bring the advanced AI technologies to manufacturing domain through an evolution from hierarchical and reactive decision making to self-learning and proactive control strategies, underpinned by predictive and prescriptive AI analytics at both component and system level, by cross-fertilizing edge and platform AI, while leveraging the human knowledge and feedback for reinforcement learning.
In addition to WPs structuring, the project decided, during the WP1 development, to also map the WPs to the different Use Cases (UC) promoted in the project. These UCs are related to the three demonstration pilots of the project and correspond to specific current production/maintenance/quality situations to be improved by AI. These improvements should be supported by AI services developed in the WPs knowing that a same service (result of one WP) but with different configurations can be implemented in different UCs. 8 UCs have been selected: 5 for CONTINENTAL and 3 for INEOS.
In relation to these WPs, at M18, all are in progress but not all the tasks. Indeed, the tasks T5.5 T6.2 T6.3 T6.4 T6.5 T7.4 T7.5 are not yet started “officially”.
Concrete results about the work developed now in the 8 WPs are visible through the Deliverables provided and submitted to the EU portal (18 Deliverables), and the Milestones achieved (M1, M2, M3).
Globally, the work carried out in the first 18 months of the AI-PROFICIENT project laid the foundations and the first developments for the upcoming activities. In that way, the WP1 led by the end-users, allowed according to the three pilots, at least to clearly define the Use Cases (i.e. operational scenarios), and the AI services in link with them. All this information is enclosed in D1.3. D1.4 and D1.5. which now serve as reference for the development WPs.
In addition, “Ethics by design approach” has been initiated in WP1, from identifying all the issues that might cause ethical issues for the project. This approach is used as guiding principle of the project, to ensure that whatever work is carried out in relation to AI, it will always be ethically responsible.
About AI services development, all the tasks of WP2, WP3, and WP4 are in progress. Even if some deviations have been underlined, the mitigations measures adopted should normally allow these WPs to be carried out.
The final results of these developments will be integrated to the AI-PROFICIENT platform which is being developed in WP5. As these developments are not fully achieved at M18, only T6.1. is active within WP6. From a management and dissemination/exploitation point of view, work done in WP8 and WP7 respectively are in line with expected works.
Looking at the progress of the project, the overall level of activities done until M18 are in line with the expectations.
In relation to these WPs, at M18, all are in progress but not all the tasks. Indeed, the tasks T5.5 T6.2 T6.3 T6.4 T6.5 T7.4 T7.5 are not yet started “officially”.
Concrete results about the work developed now in the 8 WPs are visible through the Deliverables provided and submitted to the EU portal (18 Deliverables), and the Milestones achieved (M1, M2, M3).
Globally, the work carried out in the first 18 months of the AI-PROFICIENT project laid the foundations and the first developments for the upcoming activities. In that way, the WP1 led by the end-users, allowed according to the three pilots, at least to clearly define the Use Cases (i.e. operational scenarios), and the AI services in link with them. All this information is enclosed in D1.3. D1.4 and D1.5. which now serve as reference for the development WPs.
In addition, “Ethics by design approach” has been initiated in WP1, from identifying all the issues that might cause ethical issues for the project. This approach is used as guiding principle of the project, to ensure that whatever work is carried out in relation to AI, it will always be ethically responsible.
About AI services development, all the tasks of WP2, WP3, and WP4 are in progress. Even if some deviations have been underlined, the mitigations measures adopted should normally allow these WPs to be carried out.
The final results of these developments will be integrated to the AI-PROFICIENT platform which is being developed in WP5. As these developments are not fully achieved at M18, only T6.1. is active within WP6. From a management and dissemination/exploitation point of view, work done in WP8 and WP7 respectively are in line with expected works.
Looking at the progress of the project, the overall level of activities done until M18 are in line with the expectations.
The main “Research and Innovation” results planned for the AI-PROFICIENT project (at M36) and considered in progress with regards to the SoA, are related to:
- SMART COMPONENTS FOR EMBEDDED AI AT SYSTEM EDGE
* AI enabled FDD (Fault Detection and Diagnostics) developed for component self-diagnostics and health assessment and embedded at MEMS (Micro Electro Mechanical Systems) and PLCs (Programmable Logic Controller) at system edge.
- AI PROGNOSTICS FOR SYSTEM DEGRADATION AND HEALTH STATE ASSESSMENT
* Machine vision analytics (based on gAnalyzer) extended with CNNs (Convolutional Neural Network) and DNNs (Deep Neural Network) for improved quality assurance and KPIs evaluation.
* Edge AI for component-level proactive maintenance strategies.
- IIOT FOR SMART COMPONENT INTEGRATION AND INTEROPERABILITY
* Set up of IIOT (Industrial Internet of Things) environment for system integration and interoperability based on communication middleware, message broker, LoRa (Low-Power wide area network).
* Definition of M2M (Machine-to-Machine) communication protocols and interfacing based on Data Gateway and CDM (Canonical Data Model) messaging format leveraging OPC UA (Open Platform Communications Unified Architecture).
- SEMANTIC LIFTING AND MODEL AGNOSTIC TECHNIQUES FOR XAI
* Semantic data lifting, fusion and enrichment through development of knowledge-graph-oriented micro-service architecture.
* Development of role-specific HMI (Human Machine) interfaces for the connected worker and shop floor assistance (dashboards and smart handheld devices).
- HYBRID DIGITAL TWINS AND PROCESS MODELLING
* Advancing the physical modelling of chemical processes and thermodynamics by using the operation data and domain knowledge.
* Physical process modelling and advancement of digital twins of manufacturing components (e.g. machine tools and cells).
- GENERATIVE OPTIMISATION OF PRODUCTION PROCESSES (HUMAN IN THE LOOP)
* Deployment and set up of rule-based data stream processing engine and AI enabled CEP (Complex Event Processing) and ECA (Event Condition Action) rules extraction.
- AI ENABLED DECISION-MAKING FOR QUALITY ASSURANCE
* Development of predictive and prescriptive AI services for FDD based on ML (Machine Learning) models and empirical trend modelling for quality assurance
* Development of deep learning techniques for proactive maintenance through DNN models for condition monitoring and prognostics.
* Holistic generative optimization via evolutionary GA algorithms (Genetic Algorithms) and reinforcement leaning for optimal production planning and execution.
- ROLE-SPECIFIC VISUALIZATION FOR TRANSPARENT AI DECISION SUPPORT
* Integration under MES (Manufacturing Execution System; Olanet 4.0) and ERP (Enterprise Resource Planning; RPS) solutions
* Application of EHSQ (Environmental Health Safety and Quality) & operational risk management for process safety in smart manufacturing (TenForce EHSQ platform)
* Integration and application under the Epicor ERP.
- ETHICS BY “DESIGN APPROACH”
- SMART COMPONENTS FOR EMBEDDED AI AT SYSTEM EDGE
* AI enabled FDD (Fault Detection and Diagnostics) developed for component self-diagnostics and health assessment and embedded at MEMS (Micro Electro Mechanical Systems) and PLCs (Programmable Logic Controller) at system edge.
- AI PROGNOSTICS FOR SYSTEM DEGRADATION AND HEALTH STATE ASSESSMENT
* Machine vision analytics (based on gAnalyzer) extended with CNNs (Convolutional Neural Network) and DNNs (Deep Neural Network) for improved quality assurance and KPIs evaluation.
* Edge AI for component-level proactive maintenance strategies.
- IIOT FOR SMART COMPONENT INTEGRATION AND INTEROPERABILITY
* Set up of IIOT (Industrial Internet of Things) environment for system integration and interoperability based on communication middleware, message broker, LoRa (Low-Power wide area network).
* Definition of M2M (Machine-to-Machine) communication protocols and interfacing based on Data Gateway and CDM (Canonical Data Model) messaging format leveraging OPC UA (Open Platform Communications Unified Architecture).
- SEMANTIC LIFTING AND MODEL AGNOSTIC TECHNIQUES FOR XAI
* Semantic data lifting, fusion and enrichment through development of knowledge-graph-oriented micro-service architecture.
* Development of role-specific HMI (Human Machine) interfaces for the connected worker and shop floor assistance (dashboards and smart handheld devices).
- HYBRID DIGITAL TWINS AND PROCESS MODELLING
* Advancing the physical modelling of chemical processes and thermodynamics by using the operation data and domain knowledge.
* Physical process modelling and advancement of digital twins of manufacturing components (e.g. machine tools and cells).
- GENERATIVE OPTIMISATION OF PRODUCTION PROCESSES (HUMAN IN THE LOOP)
* Deployment and set up of rule-based data stream processing engine and AI enabled CEP (Complex Event Processing) and ECA (Event Condition Action) rules extraction.
- AI ENABLED DECISION-MAKING FOR QUALITY ASSURANCE
* Development of predictive and prescriptive AI services for FDD based on ML (Machine Learning) models and empirical trend modelling for quality assurance
* Development of deep learning techniques for proactive maintenance through DNN models for condition monitoring and prognostics.
* Holistic generative optimization via evolutionary GA algorithms (Genetic Algorithms) and reinforcement leaning for optimal production planning and execution.
- ROLE-SPECIFIC VISUALIZATION FOR TRANSPARENT AI DECISION SUPPORT
* Integration under MES (Manufacturing Execution System; Olanet 4.0) and ERP (Enterprise Resource Planning; RPS) solutions
* Application of EHSQ (Environmental Health Safety and Quality) & operational risk management for process safety in smart manufacturing (TenForce EHSQ platform)
* Integration and application under the Epicor ERP.
- ETHICS BY “DESIGN APPROACH”