Electricity meets New Age Electricity (A.I.)
This post is about ‘When 20th century electricity meets 21st century electricity i.e. Artificial Intelligence.
Andrew Ng has famously quoted
Artificial Intelligence is the new ‘Electricity’. — Andrew Ng
Recently I came across an opportunity to apply machine learning techniques to renewable energy industry. In the beginning I was skeptical if ML (Machine Learning) could do any good for the renewable energy generation industry, as I spent more time understanding the industry I realized it is a very important and untapped opportunity, and concluded to get started in the field.
Renewable Energy
Renewable energy is energy that is collected from renewable resources, which are naturally replenished on a human timescale, such as sunlight, wind, rain, tides, waves, and geothermal heat. — wikipedia
Types of Renewable Energy
Since my area of interest in Solar Engery this post will heavily focus on that.
Solar Energy
The conversion of sun’s light/heat into a different form of energy generally electricity comes under solar energy. The traditional way is to set up ‘Photovoltaic’ (PV) Solar Panels which generate electricity when the sun is up in the sky.
Here is an image of a solar power plant.
Generally a solar power plant is spread across multiple acres of land and they tend to be huge.
Lets dig deep and get a basic understanding of any solar power system.
Working of a Solar Power System
The very basic working can be explained in simple words as the sun rays fall on the PV solar panel, the panel generates electricity in the form of DC current, the DC to AC converter converts electricity from Direct Current to Alternating Current which is then input to another system e.g. grid or home.
Here is a diagram of house mounted solar power system
The solar modules generally mounted on the house aggregate the power to a inverter the inverter converts the power to AC and passes on to fusebox which supplies the current to household appliances or to the grid depending on the use case.
For a house mounted system or a low consumption (domestic) load the monitoring and logging is not a priority, where as for an industrial grade system such as solar power plant monitoring is very important.
SCADA
SCADA stands for Supervisory control and data acquisition, it's a system which taps into different points in the power plant (here on we will mostly talk about industrial grade systems) and displays on monitors and also stores the data on a permanent storage device such as hard disk.
Basically scada is any system that records the state of the system and monitors it.
Solar power plants also use SCADA systems to monitor various parameters of the plant. Given this data there is a very good chance of performing data analysis and doing machine learning to optimize certain objectives.
Application of ML in Solar
Artificial Intelligence is going to impact a lot of industries and for systems which show determinism to some extent such as solar power plants Machine Learning is going to make a huge change in which the systems are setup and operated.
Of the top of my head I could think of a few things where application of ML makes perfect sense, e.g. energy demand forecast, failure analysis, operation & maintenance, Scheduling … etc.
It is interesting to think how the new age electricity (A.I.) is going to meet the challenges of systems generating old age electricity.
Working in solar industry for a couple of weeks have made me realize how ML could impact solar and how huge the impact will be. We would be automating almost 80 percent of the work and with proper systems in place we would keep the systems up and running for a longer duration and perform preventive management thus increasing the reliability and efficiency.
A lot can be done using ML with power generation and consumption, as this was my intro to solar industry I will not talk much about ML here as I am still exploring the industry and understanding it, I will record my ideas here as I come across and build cool things.
Please get in touch if you have any ideas or want to discuss further. Tweet me at @kaustubhn
