Predict Electricity Consumption of household using Neural Networks.
With the development of smart meters, electric utilities can obtain precise and detailed data of single consumers. This has lead to the generation of various big data sets with previously unattainable information leading to efficient smart grid performance in areas such as load forecasting. Accurate load forecasting promotes effective functioning of smart grids and power system ultimately leading to proper utilization of resources and avoidance of power outages. Meter level predictions, also known as domestic level forecasting, for short period of time finds applications in load scheduling, management of demand response and storage systems. This report describes the use of an approach combined using Seasonal Autoregressive Integrated Moving Average Exogenous (SARIMAX) and Long Short Term Memory networks (LSTM) with regression models to predict daily energy consumption of customers located in London area.
- In Smart Meter Data Analytics, we will study databases and try to discover patterns and on that basis will create a data models.
- Our study includes, Combining all blocks into a single data frame- keeping on relevant columns.
- Use day-level energy consumption data per household to normalize data for inconsistent household count.
- Explore relationships between weather conditions and energy consumptions and create clusters for the weather data.
- With the help of data models we will try to find out the suitable algorithm for our dataset for data processing.
Since the inception of electricity deregulation and market-driven pricing throughout the world, utilities have been looking for a means to match consumption with generation. Nonsmart electrical and gas meters only measure total consumption, providing no information of when the energy was consumed. Smart meters provide a way of measuring this sitespecific information, allowing utility companies to charge different prices for consumption according to the time of day and the season.
We will use data from the London data store, that contains the energy consumption readings for a sample of 5,567 London Households that took part in the UK Power Networks led Low Carbon London project between November 2011 and February 2014. We also collected weather data for London area from the darksky to explore relationships between weather conditions and energy consumptions and create clusters for the weather data- using which we can add weather identifiers to day-level data. We will try to discover patterns by studying these datasets and on that basis will create a data models to find out the suitable algorithm for data processing. We will use Python and Pandas. Python provides constructs that enable clear programming on both small and large scales. And pandas is a software library written for the Python programming language for data analysis. On that basis we will create a system that would help the consumers recognize their electric usage patterns and be able to optimize their usage.
This system will help the customers to analyse their electricity usage patterns which will help them to optimize their usage. It will also be able to predict the electricity usage for the future.
Clone this Repository using,
git clone https://github.com/mayursrt/customer-segmentation-using-k-means.git
Install jupyter from here or use
pip install jupyter
After installing jupyter notebook Just run jupyter notebook in terminal and you can visit the notebook in your web browser.
Install missing dependencies using,
pip install pandas matplotlib sklearn statsmodel tensorflow pyflux keras