Nowadays, artificial intelligence software and tools are constantly advancing and are being used in more and more fields every day. It seemed only natural that Sirea should integrate and test the interest of using artificial intelligence to improve the performance of its energy control models. R&D work has therefore been carried out and has brought us to create a prediction server using the NeurEco software. NeurEco facilitates the development of predictive models, the curved widget of the new MicroSERVER interface materializing the results obtained.
Adagos is a Toulouse-based company founded in 2011 by Mohamed Masmoudi and has been working jointly with Sirea since January 2022. Adagos has developed a new type of AI software integrating the NeurEco generator. NeurEco is a tool for generating parsimonious models that improve the accuracy of quantitative predictions while limiting the amount of data needed to build the models. Less data used implies less computing power and less energy consumption of the server designing the model. This technology opens the way to more applications in the fields of connected objects and microcontrollers with integrated logic.
MicroSERVER offers users the ability to remotely monitor and control programmed equipment. It has graphical capabilities and integrates curves and graphs to visualize and compare different data over time or between them. Many widgets are available to facilitate the customization of new supervision screens. They allow the visualization of data in different forms as well as to control a remote equipment by sending, for example, a command from the Switch widget.
The prediction models developed by Sirea will inform users of future production and consumption to help them make decisions that will lead them to optimize their consumption by taking into account of the displayed forecasts.
To do so, you simply need to collect and provide NeurEco with past production or consumption data, hour by hour or day by day, as well as meteorological data. The data is then processed to produce the results. After validation of these results, a model is then generated by the prediction server using the NeurEco tool.
The model can then be used by MicroSERVER to deliver energy management recommendations based on the estimated forecasts. The delivered indications will guide users to optimize their energy consumption and reduce their carbon footprint.
The prediction models are specific to each use case. Companies can use the predictions to organize and modulate their production schedules by taking into account the specific consumption of each machine or schedule operational tasks at the most favourable periods when solar energy is most available.
The image below represents a graph of photovoltaic production of an AEA cabinet installed at a private home. As we can observe, over a period from June 10, 2022 to June 18, 2022, the calculated consumption predictions are remarkably accurate. The light blue curve represents the user’s prediction curve, which remarkably follows the dark blue curve that represents the real values of the site.
Thanks to its predictions, MicroSERVER by NeurEco will encourage users to improve their energy sobriety by helping them to better plan their uses and contribute to improving the energy performance of the controlled installations.