Solar power is a free and clean alternative to traditional fossil fuels.
However, nowadays, solar cells' efficiency is not as high as we would like, so selecting the ideal conditions for its installation is critical in obtaining the maximum amount of energy out of it.
We want to predict the power output for a particular array of solar power generators, knowing some environmental conditions.
Solar power forecasting is very usefull in smooth operation and control of solar power plant. Generation of energy by a solar panel or cell depends upon the doping level and design of solar PV array but the main factors are the amount of solar radiation falling on the panel, environmental factors like atmospheric temperature and humidity and dust present on the panels . These factors are naturally variable and hence the output of solar cell directly depend on it. Also, the solar irradiance as well as all the above-mentioned factors are variable throughout the day. Hourly average or average at a particular interval of time of these parameters received is measured for better prediction of output of a PV module and thus a solar power plant. The lower the sampling rate better will be the predication.
This Project is based on the solar power output prediction using artificial intelligence.Due to limited resources, more focus on pollution free and natural resources made us to focus on methods to improve the output of solar.Records of 7 consecutive days is taken as input data and trained by the algorithms in machine learning.Algorithms like linear, ridge, lasso, decision tree, random forest and artificial neural networks have been implemented.Amazon web Service EC-2 Instance cloud,an Linux Virtual OS Environment run on Amazon web service is used.Data is taken for every half an hour interval in Solcast API. And the latitude and longitude is obtained from accuweather Api, To supply to Solcast Api. The total output generated at the end of the day is calculated earlier and an Email is autogenerated using the Simple Mail Service[SES] of Amazon and send to the Clients.The output of the solar is predicted with the changes of the weather also.It has been observed that improved accuracy is achieved through random forest and ridge regressor.Its very helpful to the Solarpower sations and house in many other fields.
About DATA
- temperature, daily average temperature, in degrees Celsius.
- wind-direction, daily average wind direction, in degrees (0-360).
- wind-speed, daily average wind speed, in meters per second.
- visibility, in kilometers.
- humidity, in percentage.
- average-wind-speed-(period), average wind speed during the 3-hour period de measure was taken in, in meters per second.
- average-pressure-(period), average barometric pressure during the 3-hour period de measure was taken in, in mercury inches.
- power-generated, in KW for each 30-minute period.
- Python 3.8
- Pandas
- Tensorflow 2.2
- Keras
- Matplotlib
- Seaborn
- Numpy
About DATA
- First create an API in accuweather to obtain latitude and longitude,
- sign-up in Solcast Api to obtain weather data for your location.
- create a AWS(Amazon Web service) account, then create an ec2 linux instance , save the key file.
- Download Filezilla and upload this code in it, with the folder name 'solar'.
- create a AWS SES (simple email service), verify your Email to send mail.
- Download Putty. then run code in using the linux comment (cd python3 "Full file path"/main.py)
As a result Solar output is send from your mail to the required user.