Carbon footprints and electricity forecasts

Forecasting electricity demand by combining mathematics, machine learning and statistical techniques

Like Comment
Read the Article

By: Dan Crisan, Imperial College, London, UK

Electric energy efficiency is extremely important for carbon footprint reduction. In this inter-connected, information-rich era, technology has become ubiquitous, and the energy sector is facing mass digital disruption. Whatever your position in the sliding scale from centrally controlled energy markets, through control distributed generation, to consumers of batteries and electric vehicles, accurate and timely forecasts for peak demands are a fundamental part of any efficient energy management plan. With the rollout of smart meters, renewables and low carbon technologies, individual level forecasts are becoming increasingly important.

 To manage our electric energy consumption and have a flexible, efficient and reliable system, we need good forecasts. This critical ingredient forms the basis of all kinds of applications: storage control, system modelling, making the case for low carbon technology uptake, running different scenarios for policy development, etc.

  The new book Forecasting and Assessing Risk of Individual Electricity Peaks | M. Jacob, C. Neves, D. Vukadinovic Greetham | SpringerLink is the culmination of a joint research project with a major player in the energy sector. It can be openly shared, including the code and data used in it. Each chapter is self-contained (although there are possible interconnections between chapters, depending on the particular applications one has in mind) and is suitable for teaching or to promote independent learning. Focusing on peaks, the book combines mathematical fundamentals, machine learning techniques and specific statistical methodologies for extreme values in order to address impending problems of practical importance in short-term forecasts. In particular, for the first time in book form, it includes a review of permutation-based distances and forecasts that allow small time-shifts between electric profiles. The application of extreme value theory to improve on time-series modelling of electricity load profiles is carried out in a truly novel setting, making the book unique in its field.

Dan Crisan is Professor of Mathematics at the Department of Mathematics of Imperial College London and Director of the EPSRC Centre for Doctoral Training in the Mathematics of Planet Earth, and one of the Editors-in-Chief for the series SpringerBriefs in Mathematics of Planet Earth.



Go to the profile of Guest Contributor

Guest Contributor

Guest Contributor

No comments yet.