Optimization algorithms can calculate better plant configurations. Thesting these new configurations in a real plant can be very cost intensive as production may be compromised during configuration and testing. A virtual environment which is able to simulate new parameters and verify them is a big deal as the running production is not compromised during testing and evaluation of new parameters.

The main focus of IMPROVE are learned models. Manual modeling of system models is not suitable for the complex, fast chaning industrial plants we have today. Lots of expert knowlege is needed to manually create a model. Learned models can be created using only data and little to no expert knowledge is required depending on the technology.

UPTIME platform focusses on the use of condition monitoring techniques, e.g. event monitoring and data processing systems, that will enable manufacturing companies having installed sensors to fully exploit the availability of huge amounts of data and to handle the real-time data in complex, dynamics environement in order to get meaningful insights and to decide and act ahead of time to resolve problems before they appear, e.g. to avoid or mitigate the impact of a future failure, in a proactive manner. Moreover, UPTIME proposed unified framework will not be limited to monitoring and diagnosis but it aims to cover the whole prognostic lifecycle from signal processing and diagnostics till prognostics and maintenance decision making along with their interactions with quality management, production planning and logistics decisions.

UPTIME Platform is developed accroding to unified predictive maintenance framework and an associated unified information system to enable the predictive maintenance strategy implementation in manufacturing industries. The UPTIME predictive maintenance system will extend and unify the new digital, e-maintenance services and tools and will incorporate information from heterogeneous data sources, e.g. sensors, to more accurately estimate the process performances. The UPTIME predictive maintenance platform is developed mainly based on five baseline e-maintenance services and tools:

• USG (Universal Sensor Gateway): USG serves as a modular data acquisition and integration device to the Product Lifecycle Management (PLM) ecosystem of a product
• preInO: The preInO Processing Engine is able to detect and predict the state of a whole system or components with respect to mechanical systems such as windturbines, special-purpose vehicles, production machinery, etc.
• PANDDA (ProActive seNsing enterprise Decision configurator DAshboard): PANDDA is a software service that implements (i) proactive decision methods to provide recommendations about mitigating (perfect or imperfect) maintenance actions and the time for their implementation on the basis of real-time prognostic information; and (ii) a Sensor-Enabled Feedback (SEF) mechanism for continuous improvement of the generated recommendations.
• SeaBAR (Search Based Application Repository): SeaBAR is a modular software platform built on Big Data and Enterprise Search technology. The SeaBAR platform supports end users by means of data aggregation, data analysis and visualization.
• DRIFT (Data-Driven Failure Mode, Effects, and Criticality Analysis (FMECA) Tool): DRIFT is a tool that, on the basis of the information gathered in other modules (maintenance data, production, logistics, quality data) use them to identify what are the Failure Modes, Effects and Criticalities of the components and system according to literature available and novel correlation algorithms among modes failures, effects and critical impacts.

UPTIME will enable manufacturing companies to reach Gartner's four levels of data analytics maturity for optimised decision making - each one building on the previous one: Monitor, Diagnose and Control, Manage, Optimize - aims to optimise in-service efficiency and contribute to increased accident mitigation capability by avoiding crucial breakdowns with significant consequences. UPTIME Components UPTIME_DETECT & UPTIME_PREDICT and UPTIME_ANALYZE aim to enhance the methodology framework for data processing and analytics. The key role for the UPTIME_DETECT and UPTIME_PREDICT components are data scientists who are in charge of developing, testing and deploying algorithmic calculations on data streams. In this way, the component is able to to identify the current condition of technical equipment and to give predictions. On the other hand, the UPTIME_ANALYZE is a data analytics engine driven by the need to leverage manufacturers’ legacy data and operational data related to maintenance, and to extract and correlate relevant knowledge.

To ease integration of all UPTIME components, the main programming language used by the components and the integrated platform is Java.

The UPTIME conceptual architecture was designed according to the ISO/IEC/IEEE 42010 “System and software engineering – Architecture description” and mapped to RAMI 4.0 in order to ensure that it can be represent predictive maintenance in the frame of Industry 4.0.

The UPTIME vision converges and synthesizes predictive maintenance, proactive computing, the Gartner’s levels of industrial analytics maturity and the ISO 13374 as implemented in MIMOSA OSA-CBM in order to create a consistent basis for a generic predictive maintenance architecture in an IoT-based industrial environment. In this way, the Operational Technology and the Information Technology can also be converged in the context of Industry 4.0.

UPTIME will provide a unified predictive maintenance framework and an associated unified information system in order to enable the predictive meintenance strategy implementation in manufacturing firms with the aim to maximize the expected utility and to exploit  the full potential of predictive management, sensor generated big data processing, e-maintenance, proactive computing and the four levels of data analytics maturity. The unification of the novel e-maintenance services and tools in the context of the proposed framework will lead to overcoming of the existing commercial software and research prototypes limitations and will conclude in a novel predictive maintenance solution covering the whole prognostic lifecycle.

Some limitiations of existing e-maintenance commercial software and research prototypes that will be addressed by UPTIME Platform, for example:

• The majority of such prototypes and commercial systems focus on product maintenance, i.e. on the service stage of the PLM (e.g. warranty failures) and not on industrial maintenance, i.e. on the manufacturing stage of the PLM.
• The developed systems are mainly based upon physical, domain-specific models that are not easily extensible for other equipment or for other industries.
• They rarely exploit big data processing infrastructures for real-time, sensor data, since they usually use batches of data, while the level of data analytics maturity is usually low.
• They do not allow the embodiment of domain knowledge in order to be used by appropriate methods and to be coupled with the real-time data.
• Each one of them focuses on a specific aspect of predictive maintenance (e.g. condition monitoring, prognostics, maintenance spare parts) instead of having a unified approach for covering all the phases and related aspects in terms of industrial operations management.

Z-BRE4K will contribute to the productivity increase of different critical manufacturing processes, such as joining (GESTAMP), cutting (PHILIPS) and forming (GESTAMP, PHILIPS, SACMI) by providing analytics and suggestion in to order to assist in minimizing the machines breakdowns. The main gain is operational and maintenance costs reduction. Furthermore, for the GESTAMP use case a real-time arc welding quality control system, based on infrared images, is being developed

Part of the Z-BRE4K project is the development, of a novel embedded condition monitoring solution with cognitive capabilities, by applying deep learning techniques to reduce the dimensionality of multimodal sensor data associated to a given machine/device, and provide meaningful features to predictive maintenance services on the cloud. Most suitable IoT edge devices, for optimal trade-off between computational power and energy consumption, sensors, providing relevant information of the condition of different components, and signal processing algorithm are proposed for different machines and processes. Data gathering is enabled by the installation of IoT gateways, where data in different protocols are homogenised and sent to the cloud for storage. Real-time data, relevant KPIs and information about components status are visualised through dedicated dashboards.

AUTOWARE platform will push forward and stablish an open CPPS ecosystem. In the AUTOWARE Framework, a collection of enablers has been defined as components/tools that will enable potential users of the tools, be it end users, system integrators or software developers, to easily apply the developed technical enablers to their daily work. Moreover, verification, validation and certification enablers will be introduced in the AUTOWARE platform.

In the framework of Z-BRE4K, an IoT approach is applied to integrate end user machines to the Z-BRE4K platform. Through IoT gateways deployed at the shop floor, machine components are enabled to communicate their conditions, sending sensors data to the cloud where they are stored and analysed to provide predictive maintenance related information.

The UI’s goal is to visualize data from maintainers and components statuses (real time data and relevant KPIs). The SPARQL Web service is used to send custom SPARQL queries against the Semantic Framework RDF repository as a general-purpose querying web service. The UI can visualise the probability of breakdown and RUL. CAD and CAE models would be useful in mapping these values into a 3D visualisation.

Z-BRE4K solution provides a big data analytics framework for the identification of the deterioration trends to extended towards prescriptive maintenance. Advanced data analysis tools are under development, to be applied to the quality and production data to realise zero-defect and zero-break down production. Furthermore, it involves models for anomaly detection, that are capable of identifying the machine states where the operation deviated from the norm.  This is achieved by collecting the data from the machine sensors in chunks of time and processing them in batch through deep learning models. The models are trying to recreate their inputs, and this results in an observable measure called Reconstruction Error, which is used to identify states that the models aren’t capable of addressing sufficiently (which constitutes an anomaly.

The modelling and simulation methods used in Z-BRE4K are mainly Finite Element Methods (FEM) where complex problems and processes from the real world are being simplified and solved using a numerical approach. First, an accurate digital model of the geometry and material properties of all involved objects, boundary conditions between these objects and process data is created (i.e. forces or temperature). 

Then, the complex shape of all objects involved, is approximate using a finite number of simple geometries (i.e. triangles) which simplify the complex mathematical problem. A computer is capable of solving these mathematical operations at a rate impossible for humans and thus enables the user to analyse various scenarios, ranging from mechanical strains within the objects to rise in temperature or material fatigue. This information can be used to predict the remaining useful lifetime of a given tool.

Simulation platform is deployed by the physical equipment to create intuitive maintenance control and management systems. The Z-BRE4K’s platform simulation capabilities will estimate the remaining useful life calling for maintenance and suggesting the optimal times to place orders for spare parts, reducing the related costs. The increased predictability of the system and the failure prevention actions will reduce the number of failures, maximise the performance, reduce the repair/recover times reducing further the costs.

By applying time series analysis, we are able to detect special events that are known (Fault detection) or unknown (anomaly detection) during production. This information, correlated with sensor readings is fed into machine learning algorithms that create estimates of Remaining Useful Life (RUL), Health Indexes (HI) and forecast upcoming events (Likelihood of Failure). Special focus is given in techniques that can provide real-time information (Fast computation and high accuracy) as well as being scalable in order to use new data as it becomes available. Additional information such as meantime between failures based on historical data or an expert opinion, CAE data, quality control data, real time states etc. are also used to the design of machine simulators.

The AUTOWARE apps development will be supported and linked to the FIWARE. AUTOWARE approach is to connect and extend the FIWARE for Industry resources and assets to the end of digital automation community, so synergies emerge both in terms of multi-sided business opportunities and amount of resources that are made available to the cognitive automation community to build their autonomous solutions and apps.

The choice of the AUTOWARE platform was based on the fact that is an open source project, and that is hardware agnostic.

The Z-Bre4k solution is based on the blackboard architectural model. This model is mainly an artificial intelligence approach, where a common knowledge base, the "blackboard", is iteratively updated by a diverse group of specialist knowledge sources, starting with a problem specification and ending with a solution. Each knowledge source updates the blackboard with a partial solution when its internal constraints match the blackboard state. In this way, the specialists work together to solve the problem. The blackboard model was originally designed as a way to handle complex, ill-defined problems, where the solution is the sum of its parts. The blackboard component acts as a central repository system. The rest of the software applications (components) act independently at the common data structure stored on the blackboard, they respond on changes and create new reactions according to changes. Interaction between components is implemented via the blackboard.

AUTOWARE reduces significantly the access complexity to the different isolated tools and speeds up the process by which multi-sided partners can meet and work together.