RESOURCES
Sofware Services
Software Services (SS) represent the most significant and innovative outcomes generated by the openZDM project. These are the tangible results that have clear potential for future use or further development to benefit the manufacturing industry.
LEADING PARTNER
ASSOCIATED PARTNERS:
INTRA, MSI, IPB, TECNALIA, HTPT, UPORTO
Description
The service delivers a catalogue of data driven Quality Assessment Modules, named Data Analytics Tools or DATs, that analyse shop floor data to support zero defect manufacturing. The catalogue contains fourteen tools, from DAT 0 to DAT 13, grouped into four clusters such as preprocessing, monitoring, prediction and classification, and descriptive and prescriptive analytics. They process heterogeneous data from non-destructive inspection systems, legacy equipment and digital twins in order to clean and enrich data, monitor process behaviour, predict defects, explain root causes and recommend corrective actions. The tools rely on machine learning and deep learning methods such as Random Forest Regression, gradient boosting, neural networks, time series models and explainable artificial intelligence, combined with statistical analysis and optimisation. All modules are containerised with Docker and integrated with the openZDM platform through the AAS middleware, Node RED workflows and dedicated dashboards. They can operate standalone or be stacked on top of Digital Twin models to support both real time monitoring and feed forward what if analysis.
Purpose
The service can be used to build and deploy AI based quality assessment solutions that combine anomaly detection, defect prediction, claim analysis, model performance tracking and energy or environmental impact assessment. It supports inline monitoring of dimensional and gap and flush measurements, thermal profiles, wall thickness and critical KPIs such as mean time between failures and cost of poor quality, using configurable dashboards and APIs, trained on data from different pilots. It also supports prescriptive use cases such as process reconfiguration, setpoint optimisation and battery charging profile optimisation using optimisation driven DATs.
Benefits
- Early detection and prediction of quality deviations from inline NDI streams and legacy sensor data.
- Unified catalogue that covers data preprocessing, monitoring, prediction and prescriptive optimisation in one framework.
- Technology neutral, container-based deployment that can run on premise or in the cloud and integrate with existing data sources.
- Built in explainability, model drift monitoring and retraining strategies that keep models valid under changing operating conditions.
Key stakeholders
- Process and production engineers in discrete and continuous manufacturing.
- Quality and reliability engineers who need inline and near real time quality indicators.
- Data science and digitalisation teams that design and operate analytics pipelines.
- NDI system integrators and machine builders that want analytics tightly coupled to sensors.
Success stories
- In the metal industry pilot, DATs enabled inline measurements of relevant product dimensions for formed and semi-finished products, feeding integrated applications with dimensional quality metrics that were visualised in the platform.
- In the bottle industry pilot, DATs supported prediction of bottle wall thickness and horizontal distribution using thermal images and presented the results through the openZDM dashboards.
Get access
The DAT catalogue is delivered as a set of Dockerised services integrated with the openZDM platform. Deployment is done through the Configurator and platform orchestrator using Docker Compose, with service discovery via Eureka. Access and exploitation follow the project exploitation and licensing strategy, and dissemination is foreseen through partner industrial offerings and future platform releases. Interested users should contact LMS to discuss integration, piloting or licensing options.
Contact
LEADING PARTNER
ASSOCIATED PARTNERS:
MSI, INTRA, IPB, TECNALIA, HTPT, UPORTO
Description
This service is the framework that creates, trains, deploys and updates data driven Quality Assessment Modules. It comprises the DAT Training Application together with the deployment pipeline through the Configurator and the openZDM platform. The DAT Training App provides a web interface built in React and served by an Express backend. Users can create new modules or update existing ones, select a methodology and model, upload training data in formats such as CSV, Excel, point cloud-based classifier or image zip, configure model hyperparameters, and choose feature and target columns. A FastAPI based Model Service trains the models and stores them and their metadata in a SQL Server database. Once a model is approved, the backend automatically builds a Docker container that exposes a REST API and registers the new module in the Eureka service, making it discoverable by the Digital Twin Toolset and the platform. Updates follow the same flow, with the original configuration pre-loaded so that retraining or parameter changes can be applied without redesigning the module.
Purpose
The framework is used to streamline the end-to-end creation of data driven quality assessment services. It supports experimentation with multiple models and parameter sets, selection of the best performing models, and one click creation of containerised modules with REST APIs. It supports retraining and updating existing services when new data, features or operating conditions appear, with consistent registration to the openZDM service registry and exposure to other tools such as the Digital Twin Toolset and Decision Support Tool.
Benefits
- Single framework to train, package and register heterogeneous AI models as production ready Quality Assessment Modules.
- Support for tabular, image and point cloud datasets, with guided configuration of methods and hyperparameters.
- Automated containerisation, service registration and deployment through the Configurator and openZDM platform, reducing manual DevOps work.
- Built to scale, with secure communication, model catalogue APIs and flexible extension to new models and data formats.
Key stakeholders
- Data scientists and AI engineers who design and train Quality Assessment Modules.
- Software and DevOps engineers responsible for deployment and lifecycle management of analytics services.
- Process and quality engineers who need to update models with new data without low level coding.
- Platform integrators who embed DAT services into digital twin workflows and business applications.
Success stories
- The framework was used to train and deploy and finally validate DATs. Some of the included multi target predictive models for steel trailing arms and classification models for defect prediction and explanation.
Get access
The DAT Training App and its deployment pipeline are provided as Docker images and configuration files, integrated into the openZDM platform. Access is restricted to authorised platform users through Keycloak, typically administrators and expert users. Industrial adoption is expected through integration projects, where LMS and partners install and configure the framework within the customer.
Contact
LEADING PARTNER
Description
This service provides a point cloud-based methodology and application for building Quality Assessment Modules that measure dimensional defects of steel products. It is integrated into the DAT Training framework and the openZDM platform. The Point Cloud Configurator is a Dash based user interface that uses Plotly and Open3D to visualise three-dimensional point clouds from inline non-destructive inspection systems. Users define bounding boxes and measurement regions directly on the visualisation and submit configurations to a FastAPI service that performs feature extraction on the point cloud. The backend computes dimensional metrics such as distances, hole locations and other geometry descriptors and stores both configuration and results in a PostgreSQL database. When a robust configuration is identified, the same workflow creates a Dockerised module for inline execution, integrated with NDIs and the openZDM platform. Validation in the steel pilot demonstrated inline three-dimensional measurement of relevant dimensions for formed and semi-finished products, with results visualised in the platform dashboards and used by integrated applications.
Purpose
The service can be used to configure and deploy inline three-dimensional dimensional inspection modules for steel products, such as trailing arms and other formed components. It allows users to transform raw point cloud data from inline sensors into stable dimensional indicators, feed these indicators into dashboards and downstream analytics, and update the configuration when product designs or tolerance requirements change. It also provides an efficient path to wrap these pipelines into containerised services that can operate in real time on the shop floor.
Benefits
- Direct use of raw point clouds from inline laser-based inspection systems for dimensional quality assessment.
- Interactive definition of measurement regions and parameters, with immediate feedback on extracted dimensions and sensitivity to configuration choices.
- Automatic packaging of validated point cloud processing pipelines as reusable Quality Assessment Modules that can run inline.
- Seamless integration with openZDM data, NDIs and dashboards, enabling unified monitoring of dimensional quality along the line.
Key stakeholders
- Production and process engineers in steel manufacturing lines that require inline dimensional control.
- Quality engineers responsible for tolerance verification and defect classification on complex shaped steel parts.
- Metrology and NDI engineers who work with three-dimensional sensor systems and point cloud data.
- Data scientists who design and tune point cloud-based models for dimensional quality assessment.
Success stories
- In the VDLWEW steel pilot, the solution enabled inline three-dimensional measurement of key product dimensions, with visualisation of the measurements in the openZDM interface and data provision to integrated applications for monitoring and decision support. Separate configurations were used for formed and semi-finished products, proving the flexibility of the method.
Get access
The Point Cloud Configurator and its FastAPI backend are delivered as containerised services within the DAT Training framework and the openZDM platform. The software is intended for deployment in industrial environments with inline point cloud acquisition systems and is integrated project wise by LMS and its partners. Interested steel manufacturers can access it through collaboration with the consortium for pilot installation, adaptation to their products and integration into existing inline inspection setups.
Contact
LEADING PARTNER
Description
This service provides a point cloud-based methodology and application for building Quality Assessment Modules that measure dimensional defects of steel products. It is integrated into the DAT Training framework and the openZDM platform. The Point Cloud Configurator is a Dash based user interface that uses Plotly and Open3D to visualise three-dimensional point clouds from inline non-destructive inspection systems. Users define bounding boxes and measurement regions directly on the visualisation and submit configurations to a FastAPI service that performs feature extraction on the point cloud. The backend computes dimensional metrics such as distances, hole locations and other geometry descriptors and stores both configuration and results in a PostgreSQL database. When a robust configuration is identified, the same workflow creates a Dockerised module for inline execution, integrated with NDIs and the openZDM platform. Validation in the steel pilot demonstrated inline three-dimensional measurement of relevant dimensions for formed and semi-finished products, with results visualised in the platform dashboards and used by integrated applications.
Purpose
- Setpoint optimisation for single cycle or multi-month horizons.
- Evaluation of alternative process configurations.
- Trade-off analysis for cost/quality/energy indicators.
- Rapid A/B scenario testing; decision support during commissioning, ramp-up, and continuous improvement.
- “Safe-to-fail” virtual experiments via feed-forward simulation.
- Exploring parameter ranges and constraints pulled from the DTT.
Benefits
- Data-driven optimal setpoints for quality, energy, and cost.
- Fast scenario exploration at scale.
- Plug-and-play optimisation algorithms.
- Seamless DT integration, automatic retrieval of assets.
- Secure, portable deployment.
Key stakeholders
- Process/production engineers.
- Quality & reliability engineers.
- Operations/plant managers & planners.
- Digital-twin & data-science teams.
Success stories
- In an induction-furnace scenario, the DST guided parameter tuning (water-flow coils, energy, temperature, alignment) and produced optimal settings meeting targeted indicator improvements. Internal testing across five parameter-range scenarios confirmed feasible, actionable recommendations; the Reinforcement Learning agent achieved the highest agreement with DT-predicted outcomes. The system handled >26k combinations while maintaining responsive API performance.
Get access
Accessible via the openZDM platform. An external API exposes created scenarios for platform integration. Deployment is delivered as Docker Compose (UI, backend, agents, PostgreSQL, MongoDB, Nginx). Hardware: 8-core CPU, 16 GB RAM, 50 GB storage, Docker Engine v24.0.6+. For licensing/access outside the consortium, contact the Leading Partner.
Contact
LEADING PARTNER
Description
The DTT enables openZDM users to build, deploy, and operate interoperable Digital Twins that drive simulation and decision-making. In the final release, users can (i) auto-discover assets and use cases via the openZDM AAS Middleware, (ii) create data-driven (RFR/SVR/LSTM) or physics-based (ODEs/PINNs) models at asset or system level, (iii) compose workflows that connect models with live/historical data and DATs, and (iv) publish 2D/3D visualisations and Grafana dashboards with real-time overlays. End-to-end validation included an induction-furnace twin: model training, workflow orchestration, live/simulation data flows, and 3D visualisation were executed successfully, deployed models exposed HTTP APIs and were consumable by the DST for “what-if” evaluation. Performance testing showed stable operation with ~100 MB training datasets, worst-case API latency of 39.0 ms during model deployment, and repeatable fresh deployments. Security (HTTPS/WSS, basic auth for Node-RED/Grafana) and external APIs for visualisations and twin discovery are in place, enabling straightforward platform embedding. The DTT is production-ready as Dockerised services across Windows/Mac/Linux, and is already used to supply simulation data to quality assessment modules and the DST. ducts. It is integrated into the DAT Training framework and the openZDM platform. The Point Cloud Configurator is a Dash based user interface that uses Plotly and Open3D to visualise three-dimensional point clouds from inline non-destructive inspection systems. Users define bounding boxes and measurement regions directly on the visualisation and submit configurations to a FastAPI service that performs feature extraction on the point cloud. The backend computes dimensional metrics such as distances, hole locations and other geometry descriptors and stores both configuration and results in a PostgreSQL database. When a robust configuration is identified, the same workflow creates a Dockerised module for inline execution, integrated with NDIs and the openZDM platform. Validation in the steel pilot demonstrated inline three-dimensional measurement of relevant dimensions for formed and semi-finished products, with results visualised in the platform dashboards and used by integrated applications.
Purpose
Create and deploy Digital Twins to simulate process behaviour, test alternative configurations, and stream data to quality and sustainability analytics. Build workflows that connect models to plant data and DATs, and publish 2D/3D HMIs and dashboards. Use twins to support commissioning, ramp-up, continuous improvement, training, and “safe-to-fail” virtual experiments. Expose model endpoints for downstream tools to run feed-forward simulations at asset or system level.
Benefits
- Rapid twin authoring from asset discovery to live visualisation.
- Flexible modelling: data-driven and physics-based.
- Operational workflows with real-time data and DAT integration.
- Secure, containerised, portable deployment with platform APIs.
Key stakeholders
- Process/production engineers.
- Data scientists / ML engineers.
- Operations/plant managers.
- Digital-twin & data-science teams.
Success stories
- Induction-furnace twin created end-to-end: asset discovery via AAS, RFR/SVR model training, workflow with live/simulated data and DATs, 3D HMI with real-time overlays, and consumption by the DST for optimisation. Internal tests reported low API latency (39.0 ms max during deployment) and robust operation across fresh installations.
Get access
Accessible via the openZDM platform. External APIs expose available twins and visualisations for integration. Delivered as Docker Compose with setup guides, standard licensing/access via the Leading Partner. HW/SW: 8-core CPU, 16 GB RAM, 100 GB storage; Docker Engine v24.0.6+.
Contact
SS #6
Configurator: Low-code Canvas for Deploying and Monitoring Dockerised Microservices
LEADING PARTNER
ASSOCIATED PARTNER: IPB
Description
The Configurator is a low-code, canvas-based tool that lets users visually define, parametrize, connect and deploy Dockerised microservices as complete “apps” across different industrial environments. In openZDM, the Configurator is used to deploy containerised Digital Twins, DATs and other services, thus becoming a generic execution and orchestration environment for containerised components of applications, including third-party ones. Integrated authentication, monitoring and notification features help keep deployments secure and observable.
From a web-based interface, users can drag and drop components representing different services or modules to the central canvas. These microservices can be connected with a click of an arrow, thus defining the interaction protocol between them. When the architecture is completed, the tool automatically generates the docker-compose deployment file and configuration files. Users may then choose a manual deployment or rely on lightweight Launcher agents that retrieve application definitions from the backend and orchestrate services on one or multiple devices. After deployment and during execution, the monitoring tool collects, processes and stores alarms, warnings, logs and other data from the different docker containers to graphically show the status of the microservices. This gives the user more control and insight into the health of the deployed architecture.
Purpose
The Configurator is used to design, deploy and monitor containerised industrial applications built from Dockerised microservices. Typical uses include setting up data acquisition and storage pipelines, connecting Digital Twins with Quality Assessment Modules, and integrating new analytics or optimisation services into existing architectures. Organisations can generate applications visually on the canvas, standardising how their own and third-party software components are integrated, configured and operated across different plants and devices, without having to manually create and maintain Docker Compose stacks for each scenario, thus simplifying the rollout of new microservice-based solutions.
Benefits
- Visual, low-code composition of applications / services: Canvas-based editor to build applications by dragging, connecting and configuring reusable microservice components, without writing deployment scripts.
- Automated deployment artefacts: Automatic generation of docker-compose and related configuration files, standardising how applications are packaged and deployed across environments, and reducing manual errors and deployment time.
- Hybrid and distributed architectures: Support for deploying services on one or multiple devices (edge, on-premise or cloud) using Launcher agents, easing scalability and load distribution.
- Integrated monitoring and observability: Built-in monitoring agents provide health, logs and resource usage for each device and microservice, enabling proactive operations. The monitoring dashboard provides a real-time view of status, logs and performance metrics.
- Bridge to microservice adoption: Helps organisations move from legacy or monolithic solutions to microservice-based, Dockerised architectures using a guided, visual workflow.
Key stakeholders
- Industrial IT/OT teams managing Docker-based infrastructures and distributed applications.
- System/solution integrators delivering services based on tools such as Digital Twins, DATs and other analytics or optimisation solutions.
- Software vendors and third parties offering containerised components that must be combined into customer-specific solutions.
- Process and quality engineers using openZDM tools who need a simpler way to deploy and monitor complete microservice workflows.
Success stories
- Within openZDM, the Configurator has been used to deploy complete workflows that combine different types of data sources, synthetic data generators, Digital Twins and DATs into end-to-end applications, such as the bottle manufacturing demonstrator. These workflows are built directly on the canvas, then deployed manually or automatically via Launcher agents, and finally supervised through the monitoring dashboard. This has reduced the effort needed to create and maintain complex Docker-based stacks and has provided a common way to manage applications, improving transparency and repeatability of deployments.
Get access
The Configurator is provided and maintained by MSI as part of its portfolio of industrial digitalisation solutions. It can be offered as a standalone product or bundled with other MSI software components to configure and monitor both MSI and third-party Dockerised services within customer environments. Organisations interested in evaluating or deploying the Configurator can contact MSI to define the most appropriate deployment option (on-premise or cloud-hosted) and the associated commercial conditions.