From the Smart Factory to the thinking factory

An industrial research and experimental development project to transform industrial data into operational intelligence, integrating predictive algorithms, advanced diagnostics, and real-time monitoring throughout the entire ceramic supply chain.

FUNDING

Ministry of Enterprise and Made in Italy – Agreements for Innovation (Ministerial Decree of December 31, 2021)

PROJECT CODE

F/310087/01-05/X56

DURATION

May 2023 — December 2026 (ongoing)

THE CONTEXT

Too much data, too little intelligence

In recent years, the Italian ceramic industry has achieved advanced levels of digitization. Sensors, industrial networks, and management systems now collect enormous amounts of data along production lines, making factories more connected and better able to monitor what is happening on their plants in real time.

Yet this abundance of information has not yet translated into a real ability to anticipate problems and optimize processes. Traditional statistical techniques struggle to manage the complexity and speed of the data generated. The result is a paradox: companies have more information than ever before, but not the tools to transform it into operational decisions.

The ceramic sector, with its energy-intensive processes and product quality that depends on complex interactions between raw materials and processing parameters, is the ideal testing ground for addressing this challenge. Here, the ability to predict is worth much more than the ability to correct after the fact.

It is in this scenario that the idea of an evolutionary leap takes shape: moving from the Smart Factory, based on automation and monitoring, to the Intelligent Factory, capable of learning from data and autonomously optimizing its operations.

Lack of integration of multi-source process variables in decision-making models

>50% of waste identified only downstream of the process

<10% of line adjustments supported by predictive models

−20% potential for energy consumption reduction not intercepted

THE PROJECT

Three integrated dimensions

START was created to lay the foundations for this new production paradigm. It is not a question of adding technologies to an existing system, but of rethinking the way data is collected, processed, and transformed into concrete actions.

The project, currently underway, works on three complementary fronts that together define the architecture of smart industry.

01

Predictive models

The first area concerns the development of machine learning algorithms capable of predicting the quality of ceramic products based on the composition of raw materials and process parameters. The goal is to shift quality control from the end to the beginning of the production phase, identifying potential critical issues before they result in waste and rework.

02

Edge-to-cloud architecture

The second front addresses the infrastructure problem: how to process large volumes of data while ensuring immediate responsiveness and strategic analysis capabilities. The answer is a hybrid architecture that distributes intelligence between the production periphery, where instant responses are needed, and the center, where the overall picture is constructed.

03

From process to product

The third front extends beyond the confines of the factory. A residential prototype with a ventilated ceramic envelope is collecting data on the material's performance in real-world conditions, feeding predictive models for indoor comfort and creating a bridge between manufacturers and users.

These three fronts do not proceed in parallel but are intertwined: predictive models need infrastructure to function, infrastructure makes sense if it feeds concrete decisions, and monitoring in operation provides information that improves models. It is a systemic approach, where sustainability emerges as a natural result of smarter processes.

THE PROCESS

An evolving project

The project is being developed over a three-year period, with each phase building on the results of the previous ones. START is now in its middle phase, with its foundations consolidated and the first operational trials already underway.

PHASE 1
Completed ✓

Foundations

The work began with the construction of the reference framework: analysis of requirements, mapping of available data sources, and design of the conceptual architecture of the system. At the same time, through workshops and focus groups with operators and technicians, a framework was developed for the design of artificial intelligence systems that comply with the principles of transparency and human supervision. The characterization of national and European raw materials provided the database for training the models.

PHASE 2
Completed ✓

Development

The second phase translated the conceptual framework into operational tools. Machine learning models for quality prediction, trained on years of industrial data, were implemented and validated. Three non-destructive diagnostic systems based on ultrasonic and thermographic techniques were developed. The edge-to-cloud architecture was designed and a thermoeconomic model for assessing process sustainability was developed.

PHASE 3
In progress

Experimentation

The current phase is bringing the research results into the operational environment. The data collection infrastructure has been installed on the pilot atomization plant. The housing prototype has been completed and is collecting data on the behavior of the ceramic envelope in different seasons. The edge-to-cloud architecture has passed all validation tests. The digital twin of the factory is in an advanced stage of development.

PHASE 4
Next developments

Integration

The final phase will lead to the completion of the integrated system: testing of predictive control algorithms directly on the plant, development of the production chatbot to support operators, and thermodynamic assessment of the factory according to the developed model.

RESULTS

Concrete and measurable innovation

The project is producing tangible results which, when integrated, form the backbone of the system under construction.

Accuracy

of predictive models for product quality

Milliseconds

of edge-to-cloud system response time

Reduction

in production waste

Efficiency

of specific energy consumption

Validated edge-to-cloud architecture

The infrastructure connecting the production periphery to the decision-making center has been implemented on the pilot plant and has passed all validation tests. The most significant result is the drastic reduction in response times: operations that used to take an hour can now be completed in a few milliseconds. This change is not only quantitative but also qualitative: for the first time, truly predictive management of production processes is possible.

Simulated Intelligent Factory

The complete model of the intelligent factory was built and tested in a simulation environment. By integrating the data architecture with the plant's production logic, the system demonstrated consistent improvements in both downtime reduction and production capacity increase. The simulation confirmed the robustness of the approach and paved the way for implementation in a real environment.

Predictive models for quality

The machine learning algorithms developed are now able to predict the quality conformity of tiles by analyzing the composition of raw materials and process parameters. Trained on years of industrial data and validated on different supplies, the models have achieved levels of accuracy exceeding the initial objectives and demonstrate their ability to generalize even when the starting conditions change.

Non-destructive diagnostic systems

Three systems for characterizing ceramic materials without compromising their integrity have been developed and validated in the laboratory. The techniques developed, based on ultrasound and thermography, allow mechanical and thermal properties to be measured and defects that are not visible on the surface to be detected. They form the basis for faster and more accurate in-line checks than traditional analyses.

Housing prototype with ceramic envelope

The experimental module installed in Alghero is collecting data on the actual behavior of the ventilated ceramic envelope in different climatic conditions. The monitoring system continuously detects internal and external environmental conditions, while the artificial intelligence-based control system has proven to reduce energy consumption while maintaining high levels of comfort. It is the bridge that connects the factory to the building.

Guidelines for ethical AI

The involvement of operators, technicians, and production managers has made it possible to identify the values that should guide the introduction of artificial intelligence in the factory: transparency of decisions, possibility of human supervision, and system reliability. These principles have been translated into concrete requirements for the design of the production chatbot currently under development.

SYMΞX thermoeconomic model

A framework has been developed that integrates the principles of thermodynamics with management logic, offering a new perspective for assessing the sustainability of industrial processes. The model provides the methodological basis for the thermodynamic assessment of the factory planned for the final phase of the project.

THE PARTNERSHIP

Integrated skills for innovation

START brings together those who produce ceramics, those who build the machines to produce them, and those who study the processes that make it possible. It is an ecosystem of diverse skills working together towards a common goal.

Gruppo Ceramiche Gresmalt

One of Italy's leading manufacturers of porcelain stoneware, it coordinates the project and provides what only an industry can offer: real plants, historical data, and in-depth knowledge of processes. Its factories are where the solutions developed are put to the test.

SACMI

A world leader in technologies for the ceramic industry, it brings to the project the ability to translate research into industrial solutions. It develops edge-to-cloud architecture, control algorithms, and the Intelligent Factory model, integrating expertise in automation, IoT, and information systems.

Libera Università di Bolzano

Contributes to the design of the digital twin and data ontology, as well as guiding the development of guidelines for ethical artificial intelligence. Its Smart Mini Factory Laboratory provides a controlled environment for testing architectures before transfer to the factory

Università della Calabria

Develops predictive models for product quality and non-destructive diagnostic systems. Two departments collaborate, bringing complementary skills: raw material characterization on the one hand, machine learning algorithms and signal processing on the other.

Università degli Studi di Sassari

Designs and builds the housing prototype, conducts monitoring campaigns, and develops models for indoor comfort control. It is the partner that extends the project's scope beyond the factory to the place where the ceramic product finds its final destination.

THE IMPACT

Towards a learning industry

START is building something that goes beyond the boundaries of the project itself. The models, tools, and architectures that are taking shape can be extended to other stages of the ceramic process and adapted to manufacturing sectors with similar challenges.

Edge-to-cloud architecture is a reference point for anyone who needs to manage large volumes of data while balancing responsiveness and strategic vision. The guidelines for ethical artificial intelligence offer a method for introducing intelligent systems without losing human control. The housing prototype paves the way for buildings capable of adapting to environmental conditions and the needs of their inhabitants.

The vision that guides this work is that of an industry where sustainability is not a constraint to be respected but the natural result of more intelligent processes: capable of learning from experience, anticipating problems, and making better use of resources.

The project is ongoing. The results achieved so far show that the transition from Smart Factory to Intelligent Factory is not a future aspiration: it is a concrete path, already under construction.

Page updated

Google Sites

Report abuse