Electrical load forecasting methods

Hello,

I have a 4 years power meter hourly readings of a large scale facility, I am interested to try and predict how this load will act in the upcoming 3-4 years,

I've been exploring the net about forecasting methods, the topic is huge and there are alot of algorithims around,

I am not sure about which one i need to use.

this is done for practical uses at work in order to make sizing for microgrid elements (Battery storage, distributed power sources etc)

, not for academic uses.

is it possible to make with FFT ? - i will appreciate any guidance.

Thanks

Answer 1

[Xiangchun]

Hi Adnan,

To train a model based on time series data for forcasting purpose, one option is to consider long short-term memory networks in the Deep Learning Toolbox. Each specific problem can have its own unique challenges in terms of forcasting time series data. Here is a link to the documentation page for you to explore.

Deep Learning with Time Series and Sequence Data - MATLAB & Simulink (mathworks.com)

Please let us know any questions about the toolbox. Thanks.

Best wishes,

Xiangchun