In Ineco's Innovation 2025 Call for Proposals, one of the six winning proposals was submitted by Sergio del Pino Sánchez-Chiquito and Luis Manuel Lozano, focusing on the generation of synthetic data to optimise artificial intelligence models applied to predictive maintenance.
The initiative proposes the use of advanced algorithms based on deep neural networks (known as Deep Learning). Specifically, Generative Antagonistic Networks (GAN) are capable of reproducing temporal patterns of real data. This technology makes it possible not only to simulate the normal operation of systems, but also to generate specific errors, facilitating customised training of AI models.
The project is a response to common limitations in the availability of real data, such as sensor failures or time gaps in the records. Thanks to the synthetic data generated, the quality of training sets can be improved without the need to collect large volumes of real data, reducing costs and development time.
This proposal represents a significant advance in improving the accuracy and robustness of predictive models, with direct applications in operational efficiency and maintenance sustainability in complex technological environments.
