Breaking News

Innovation in energy field: How machine learning promotes responsible consumption – JAXenter

Artificial intelligence and machine learning technologies have certainly penetrated a wide range of industries in the past few years and particularly in 2019. While mainstream media is still thrilled about AI beating the world’s best players in chess, here at Itransition we see a much bigger potential in this technology.
Today’s global problem, the climate crisis, has been continuously covered in the news and addressed by the world’s leading scientists. A large portion of proposed solutions is very radical, including the complete elimination of fossil-fuel energy production from all sectors of the economy.
AI and ML technologies, on the other hand, can make an impact by reducing emissions and maximizing production efficiency. Hardly surprising, the energy sector has lavish amounts of data to manage, and AI is indeed a perfect fit for this purpose.
Let’s look at how machine learning can benefit the energy sector.
SEE ALSO: Web developers don’t need a math degree to get started with machine learning
ML in renewable energy
At first, renewable energy seems like a perfect answer to the climate crisis. However, this method of energy supply has proved to be highly unreliable, because renewable energy sources are dependent on weather conditions and other external factors that can’t be controlled. This means that renewable energy can undoubtedly have a positive impact yet it can’t entirely substitute other ways of producing energy.
In essence, energy management is all about maintaining equilibrium between energy supply and demand. Currently, renewable energy companies work together with conventional power stations to sustain power supply.
For example, on a sunny day, the majority of demand can be satisfied with solar power plants, which means that coal-fired power stations operators need to quickly react and reduce the output. Such practices are not economically profitable, and this is where ML comes into play.
Yes, weather forecasting is not entirely a new concept, but accurately predicting weather in real-time can only be done when analyzing large amounts of meteorological data. Then those predictions can be used to estimate how much energy wind turbines or solar panels will produce in the next few days. This process is now entirely run by AI- and ML-based applications in countries like Germany, for example. The ability to accurately monitor supply and demand in real-time makes energy cheaper and power supply more reliable.
IBM, which is known for its AI incentives, has developed a system that helps renewable energy companies in weather forecasting. The self-learning weather model and renewable forecasting technology (SMT) uses big data gathered from thousands of weather stations and satellites to make extremely accurate predictions. The ML-based system considers which weather model has been the most successful in a given location and situation. SMT has proved to be twice as accurate as conventional forecasting methods in some regions of the US.

Power outage prevention
Power grids are vital organs of today’s economy. These grids are used to deliver electricity from powerplants to hospitals, offices, homes, factories, etc. However, power grids have many vulnerabilities. A lightning strike, an operator’s error, or equipment failure are all common reasons for power outages.
We may not think about it much, but our lives and economies would quickly turn to chaos without access to electricity. For example, in March 2019, just a 14-hour downtime cost Facebook around $90 million. Now imagine how damaging power outages are for economies worldwide.
ML systems can make data-based predictions about upcoming malfunctions of power grids. Currently, such data is collected from various industrial IoT sensors, automatically analyzed by ML algorithms, and then used by operators to take immediate steps for blackout prevention. Moreover, such smart grids also help with optimizing supply according to seasonal demand, which can significantly reduce maintenance costs.
Smart energy consumption
AI applications have also found their use in smart energy consumption. Modern homes equipped with ML-powered devices can automatically react to electricity price fluctuations and adjust power usage. On a larger scale, such practices are even more impactful. For example, Google utilizes IoT and DeepMind’s AI algorithm to reduce energy consumption by 30%.
Google relies on its huge resource-intensive data centers to run a wide range of applications including Gmail, YouTube, and its search engine. Every few minutes, AI analyzes data from thousands of IoT sensors to efficiently spread cooling load across the equipment, which in turn optimizes energy consumption. Currently, though, every decision still has to be approved by a data center operator.
Interestingly enough, data center operators admit that the AI system came up with new unorthodox approaches to cooling management. In particular, AI smartly took advantage of the cold winter weather to reduce energy consumption needed for cooling down water. Given AI’s strong ‘desire’ to constantly evolve, we can only expect more energy savings and less human intervention.
SEE ALSO: Lyft open sources Flyte for developing machine learning workflows
Data-powered future
Environmental sustainability is a topic of unceasing debates, and rightfully so. Renewable energy is the core element of our collective wellbeing and economic advancement. However, the variability of weather conditions presents a significant challenge for this industry. ML-based systems can help energy suppliers to always be prepared for supply fluctuations coming from renewable energy sources. This will allow more countries to rely on green sources and make their contributions to battling the climate crisis.
Power outages will have even more significant consequences as our dependency on stable electricity supply is only increasing. ML-powered smart grids can make energy management more efficient and considerably lower the probability of blackouts. It’s only a matter of time for energy companies to recognize the immense potential of ML systems for making electricity production more affordable and reliable.
Source: https://jaxenter.com/machine-learning-energy-170668.html