Home > Electrical/Electronics > Computer (I.T) > Data Privacy And Monetization In Sustainable Energy Systems Through Collaborative Analytics

Data Privacy And Monetization In Sustainable Energy Systems Through Collaborative Analytics

Renewable energy sources (RES) forecasting services are divided into two types: Forecast-as-a-Service (FaaS) and Software-as-a-Service (SaaS). The FaaS was adopted by most of the forecast service vendors (Energy & Meteo Systems, Meteologica, Vaisala, WEPROG, etc.) that provide on-demand customer-defined forecasting services supported by a cloud-based framework or a web-service API: the client either receives numerical weather predictions (NWP) or exchanges power data for RES power forecasts of different types (from single to aggregated power plants). Less common are business models based on SaaS where the client purchases a forecasting software license, as is the case of AleaSoft in Spain.

Presently, a large amount of data is being collected from geographically distributed RES, such as wind turbines and photovoltaic panels. These data include power generation and weather measurements like air temperature, wind speed and direction, irradiation, etc. Recent literature suggests that time-series data from spatially distributed RES agents can improve forecasting skill for different time horizons. For instance, a spatial grid of NWP can improve days-ahead forecasts [1]; and geographically distributed measurements can improve forecasting skill up to six hours-ahead for wind and solar energy [2,3].

This context motivates the design of new business models for forecasting services, now driven to exploit data from different owners and create economic incentives for a collaboration that goes beyond the traditional FaaS and SaaS. To achieve this goal, two requirements need to be addressed: data privacy and algorithmic solutions for data markets.

Collaborative analytics and data privacy:

In the framework of the Smart4RES Horizon 2020 Project, INESC TEC has been developing a privacy-preserving protocol for distributed learning and collaborative forecasting, having also submitted a patent to the European Patent Office in December 2020.

With our focus on privacy-preserving protocols for very short-term RES forecasting [4], our main research outcome is a novel combination of data transformation and decomposition-based methods that allow the forecasting model to be fitted in another feature space without decreasing its forecasting skill.

The main advantages of this innovative model are:

(a) asynchronous communication between peers can be addressed;

(b) a flexible privacy-preserving collaborative model can be implemented using two different schemes – centralized communication with a neutral node and peer-to-peer (P2P) communication – and in a way that original data cannot be recovered by the central node or peers.

This distributed learning framework enables different agents or data owners (e.g., RES power plant, market players, forecasting service providers) to exploit geographically distributed time-series data (power and/or weather measurements, NWP, etc.) and improve forecasting skill while keeping data private. In this context, data privacy can either refer to commercially sensitive data from grid-connected RES power plants or personal data (e.g., under European Union General Data Protection Regulation) from households with RES technology. Distributed learning means that, in replacement of sharing data, the model fitting problem is solved in a distributed manner.

READ ALSO  India’s Richest Man Set To Take On Amazon And Flipkart

As mentioned, two collaborative schemes are possible: centralized communication with a central node (central hub) and peer-to-peer communication (P2P).

Centralized and peer-to-peer communication schemes for collaborative analytics (image by author)
Centralized and peer-to-peer communication schemes for collaborative analytics (image by author)

In the centralized model, the central node can be either a transmission/distribution system operator (TSO/DSO) or a forecasting service provider. The TSO or DSO could operate a platform that promotes collaboration between competitive RES power plants to improve the forecasting accuracy and reduce system balancing costs; on the other hand, the forecasting service provider could host the central node and provide APIs and protocols for information (not data) exchange between different data owners, during model fitting, and receive a payment for this service.

The P2P model is suitable for data owners that do not want to rely (or trust) upon a neutral agent.

Potential business models are:

(a) P2P forecasting between prosumers or RES power plants;

(b) smart cities characterized by an increasing number of sensors and devices installed at houses, buildings, and transportation networks.

Data markets and monetization:

Since RES agents are most likely competitors in the same electricity market, they are unwilling to share data, particularly power measurements, even if data privacy is ensured. An effective way to encourage agents to share their data is through monetary compensation. A “secondary” market to trade data is necessary to monetize RES forecasting data. This data market should operate in a way that, after some iterations, agents realize which data is relevant to improve its gain, so that sellers are paid according to their data. The buyers’ gain should be a function of the forecast accuracy and value in a specific use case, e.g., imbalance costs reduction in electricity market bidding.

In the Smart4RES project, we are developing algorithmic solutions for data markets in RES forecasting [5]. The first approach was to extend the solution from [6] to a sliding window environment and adapt the gain function for RES forecasting and bidding in the electricity market.

Framework of the data market for renewable energy forecasting (image by author)

Data from the Nord Pool market was used to evaluate the potential of a data market for RES agents, and it was concluded that:

(a) all agents benefit (from the economic point of view) from the data market;

(b) agents that first observe wind-flow (or wind generation) in one location, e.g., at timestep t − 1, provide relevant information to improve the forecasting model (e.g., for t + 1) of neighbor agents in other locations, conditioned by wind direction, and then all agents benefit from the higher revenue accrued either from the data market or the better forecast in the electricity market. In summary, data markets can be a solution to foster data exchange between RES agents and contribute to reducing imbalance costs.

READ ALSO  Why We Are Resuming Digital Switch Over Roll Out - FG

The shape of things to come:

The application of privacy-preserving protocols and data markets is not confined to RES forecasting. The figure below depicts other potential use cases and INESC TEC is applying these techniques to data-driven low voltage control in the EUniversal Horizon 2020 project while considering sparse data from smart metering infrastructures.

Potential use cases for collaborative analytics and data markets (image by author)

In the future, utilities will likely create economic incentives to have consumers sharing their data (from smart appliances, energy gateways, etc.) and use this data to support grid operation (e.g., fault and outage location), as well as to offer energy and non-energy services. One example is the use of smart meter data to rank consumers according to their elasticity to dynamic tariffs and identify demand response potential using causality inference [7].

Acknowledgments

The research leading to this work is being carried out as a part of the Smart4RES project (European Union’s Horizon 2020, No. 864337). The present article reflects only the authors’ view. The European Innovation and Networks Executive Agency (INEA) is not responsible for any use that may be made of the information it contains.

Further reading

[1] J. R. Andrade and R. J. Bessa, “Improving renewable energy forecasting with a grid of numerical weather predictions,” IEEE Trans. Sustain. Energy, vol. 8, no. 4, pp. 1571–1580, Oct. 2017.

[2] L. Cavalcante, R. J. Bessa, M. Reis, and J. Browell, “LASSO vector autoregression structures for very short-term wind power forecasting,” Wind Energy, vol. 20, no. 4, pp. 657–675, Apr. 2017.

[3] R. Bessa, A. Trindade, and V. Miranda, “Spatial-temporal solar power forecasting for smart grids,” IEEE Trans. Ind. Informat., vol. 11, no. 1, pp. 232–241, Feb. 2015.

[4] C. Gonçalves, R.J. Bessa, P. Pinson, “Privacy-preserving distributed learning for renewable energy forecasting,” IEEE Trans. Sustain. Energy, In Press, 2021.

[5] C. Gonçalves, P. Pinson, R.J. Bessa, “Towards data markets in renewable energy forecasting,” IEEE Trans. Sustain. Energy, vol. 12, no. 1, pp. 533-542, Jan. 2021.

[6] A. Agarwal, M. Dahleh, and T. Sarkar, “A marketplace for data: An algorithmic solution,” in Proc. ACM Conf. Econ. Computation, 2019, pp. 701–726.

[7] K. Ganesan, J. Tomé Saraiva, Ricardo J. Bessa, “On the use of causality inference in designing tariffs to implement more effective behavioral demand response programs,” Energies, vol. 12, no. 14, pp. 2666, July 2019.

Source: Energycentral

Total Views: 85 ,
0
0

Leave a Reply

Your email address will not be published. Required fields are marked *