A novel feature fitting simulation algorithm for estimating electric vehicle demand

Electric vehicles will drive the future, therefore forecasting and simulating `transportation electrification' demand over the coming years has become important for utilities. Ever since electricity was commercialized there has been a need for demand forecasting and simulation because electricity provider's ability to produce energy far exceeds their ability to store energy. This is an industry worth billions of dollars and therefore even a marginal improvement in the way it's predicted can have a great impact. Demand forecasting is critical for unit commitment and broadly effects the wholesale electricity market price. With the addition of transportation electrification this process has become even more challenging. Traditional ways are hard to model and computationally intensive, nowadays this type of problem is of great interest in the field of machine learning as well, because of the availability of large datasets from utilities. However datasets for transportation electrification still remain a huge challenge therefore more work needs to be done in forecasting electric vehicular loads. This paper tackles this new problem and presents a new method called Feature Fitting Simulation Algorithm to estimate electric vehicle charging demand profiles. The simulation was performed on MATLAB and Excel using various tools and functions to ensure the algorithms run on optimum efficiency. The novel feature of the algorithm is its hybrid structure of considering both historical and simulation data for temporal predication, secondly it introduces two key variables scaling and sensitivity to better control the time series output. FFSA is vetted against machine learning algorithms and the results indicate a better performance achieved by FFSA.