Machine Learning Brings a New Era in the Energy Sector Machine Learning Brings a New Era in the Energy Sector
(This article is co-authored with Rajiv Dinesh) For more than a century, the energy sector has been driven by centralized grid-based systems. But 135 years... Machine Learning Brings a New Era in the Energy Sector

(This article is co-authored with Rajiv Dinesh)

For more than a century, the energy sector has been driven by centralized grid-based systems. But 135 years after the advent of these systems, millions of people are still left behind with insufficient or no electricity. One just needs to visit a developing country to realize how transient and unreliable energy access can be.

Imagine a future where energy is cheap, clean and abundant for all. Where each home or building is self-sufficient for energy.

This article is the story of how are we building this future. Join us on this journey and learn how you, too, can contribute to an energy-secure future.

Episode I: The Current Wars and Centralized system wins the first bout



In the late 1880s and early 1890s direct current (DC)-based Edison Electric Light Company and the alternating current (AC)-based Westinghouse Electric Company were the flag bearers of making electricity the go-to source of energy within a building.

Thomas Edison’s low voltage DC electric distribution “utility” was designed to serve indoor business and residential use, acting as an alternative to the gas- and oil-based lighting of the time. The Westinghouse model featured a central power plant that could generate electricity, and use step up and step down transformers to transmit AC current over long distances with minimum losses in transmission. This was naturally more efficient and cost-effective and also fit well in a world dominated by centralized systems.

Battle lines were drawn, and after a brief but eventful and highly public series of skirmishes, the centralized system based on AC prevailed and established the template for the energy infrastructure of the next century.

Episode II: the return of the DC devices



In the late 1990s with the arrival of millions of battery-powered consumer electronics devices like laptops, mobiles, etc., there was a renewed push for electricity storage. DC was a perfect fit for such devices and lead to a wave of renewed research.

Today we see DC systems are not just storage-based devices, but also in the latest high-energy efficiency devices like Inverter ACs and Inverter refrigerators.

With the growing relevance of DC devices, is the stage set for DC systems to make a comeback? Let’s explore how the arrival of solar makes a compelling argument for DC systems.

Episode III: Harnessing the sun


The early 2000s saw the birth of a clean energy movement across the world. With growing consciousness of pollution and climate change, an aggressive search began for ways of generating ‘green’ energy. Solar was seen as one of the forerunners of alternative energy sources. After all, life on earth has been powered by the sun for millions of years!

Solar panels run for 20 years and beyond, losing less than a fifth of their efficiency over this period, and converting a free source of energy — sunlight — into electricity that can power people’s daily requirements. As no mechanical systems are involved, simple operations like cleaning and health check of the electrical systems can ensure these plants run well. With growing adoption powered by large government contracts, solar has now become cheaper than conventional electricity, even for individual users in most parts of the world.

Episode IV: A world powered by decentralized solar power

In the case of most residences, the solar energy generated from a rooftop solar plant is sufficient to power the entire consumption of the household. There is no reason why each building cannot be self-sustaining and independent when it comes to energy.

On-grid systems, which are non-battery based, power the home during sunlight hours. Excess energy produced during this time is exported to the grid through net meters and drawn back at night to power consumption during non-sunlight hours.

Such systems are already providing cheap electricity to millions of homes across the world. With the growing demand for solar, we are also heading for an inflection point where solar + storage will be competitive with electricity from the existing system, and allow individual buildings to be self-sufficient for energy. With storage removing the dependence on the grid, households and communities can become self-sufficient, paving the way for an ecosystem of decentralized and clean energy.

What lies between us and ‘sunny days’?

So what stops us from immediate mass scale adoption of solar rooftops? A quick look at the challenges faced by a user on their solar journey will give us an idea:

  1. Lack of organized relevant information: To understand what solar means for you, today’s solar journey necessarily requires a physical visit by a solar company engineer to assess your property. The few good ‘online calculators’ give a very rough approximate at best, or require a lot of data input that you may not have an idea about (eg: what is the shadow-free area on your roof?)
  2. Time-consuming offline engagement process: Once an accurate assessment is made, customer education happens over a number of physical meetings with several solar companies in order to form an informed opinion. It is also difficult to leverage the experience of existing solar users, in the absence of an online solar community
  3. Insufficient support: Existing solar users are the best ambassadors of solar adoption. However, as maintenance is not a lucrative option for solar companies, many solar users are left hanging and do not have a positive experience to share with potential solar adopters.

The heavy offline component involved in today’s solar process makes solar sales expensive and time-consuming. This limits the ability of solar companies to approach more people and adversely affects the growth of solar adoption. As a result, solar adoption remains a project of governments and solar companies and is not backed by a popular participation and engagement of people.

Building a Solar world using Machine Learning

To bring solar out of the 1960s vacuum cleaner sales era and make it the energy platform of the future, we need to look for answers in technologies that are shaping the future.

Our goal is to build a simulator that can help people get an estimate of their rooftop solar potential and ‘test drive’ a virtual solar plant. Users will be able to place solar panels virtually on their roof and see their day to day savings. This will help them to get a first-hand experience of owning a rooftop solar plant customized to their needs and ease the process of learning about solar.

The critical challenge here is the ability to remotely map the features of a roof, including boundaries and obstacles so that the area suitable for solar can be identified. Project Sunroof of Google is trying to solve this problem, too, and has provided partial solutions for certain cities in the U.S. and Europe. However, due to the bad quality of satellite imagery in India (and other developing countries), their solution is not suitable.

Identifying critical roof features from a low-quality satellite image

Existing algorithms don’t really work for rooftop analysis due to the low quality of satellite images available for Indian roofs. A machine learning approach based on training roof datasets is promising to solve a hitherto unsolvable problem. In case you are not interested in a technical discussion of the same, you can choose to skip to the next section to see how you too can contribute to solving this problem

Approach 1: Algorithmically solving the roof detection problem

We started with OpenCV library to see how the roof edges and features can be detected algorithmically. Shown below is a solar rooftop image from Germany.

With the quality of the satellite image, the algorithmic approach works fairly well and with some noise reduction, we will be able to identify the edges correctly. But when we apply the same algorithm on the Indian rooftops, the results are far worse. The image quality from residential areas is particularly bad, and feature identification is made even more difficult by the non-uniform roofs and proximity of neighboring buildings

When we applied different edge detection methods on Indian satellite images, we got the following results:

  1. Canny and Gaussian

2. With some customized code, filters, and Laplace edge detection we can achieve the following

The results are either not satisfactory or not generic enough to work on different scenarios.

Approach 2: Machine Learning

We realized that we needed to train a machine to solve our challenges with Indian rooftops. We have begun this process using a vast repository of hundreds of roofs ranging from residential to industrial facilities that we have physically mapped, measured and documented. We want to use machine learning to identify the following:

1. Type of obstacles in rooftop — We have identified that obstacles on roofs typically belong to around 10 different families of obstacle types, for instance:

(from the top — left — right: Mumpty, Water tanks, and Turbo ventilator)

2. Type of roof — Roofs can also be classified into broadly a few categories, again validated by our manual mapping. One of the interesting articles we came across on rooftop type detection using Keras and TensorFlow. We tested this algorithm and we had ~90% accuracy.

3. Edges of the roof — This is the most challenging part of the problem and we want to train a machine to learn to identify the edges, to be able to distinguish one property from the other and one roof level from the other

The community needs to come together to solve these challenges

Any move toward putting power in the hands of people has always featured the extraordinary involvement of community (think universal franchise or independence movements).

Solar needs to become a people’s movement to realize the vision of a decentralized energy independent future.

Today, we are blessed with access to technology that gets better at solving problems with each participation by the community.

Come, do your bit, so that together we can create the platform that makes going solar affordable and effortless. We need your help with the following:

  1. Mapping roof outlines and identifying obstacles on rooftops: Come map your roof and learn about your solar potential. This activity requires nothing more than the will to contribute — no technical skills are required. Mapping your roof or a neighboring roof will take no more than a couple of minutes and contribute largely to organizing solar information for your area and community. Each activity helps the system improve, and you will be awarded points that you can exchange for rewards in future
  2. If you are a machine learning enthusiast and have experience with image analysis: Contribute to our open source project to improve the algorithms that can do this, and make it available to innovators across the world.
Rudradeb Mitra

Rudradeb Mitra

Author of Creating Value with AI ( | 6 startups | 10 yrs as an AI Engineer/Researcher

Open Data Science - Your News Source for AI, Machine Learning & more