Confused about your exam or college applications?
Be the First to Know
Get Access to Latest Updates
Researchers from the Indian Institute of Technology Madras (IIT Madras) have been able to develop algorithms which facilitate unique applications of Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning. These algorithms have been developed to solve different engineering problems across different engineering fields.
Sources state that the IIT Madras researchers will also be setting-up a startup which will be used to open up the AI-Software which they call as ‘AISoft’. The objective of the startup has been designed to focus on formulating solutions to problems that occur in different engineering fields like in Thermal Management, Aerospace, Automobile, Semiconductors and Electronic Cooling Applications.
The applications of AI, ML and Deep Learning have been adopted by engineering fields for more than a decade. However, it’s applications were only seen in areas like speech recognition, signal processing, image reconstruction, and prediction. Globally, there are have been few attempts at developing algorithms that solve engineering issues in thermal management, electronic colling industries and various automobile issues such as fluid dynamics in both automobile and aerospace industries.
An Assistant Professor, Dr. Vishal Nandigana of Fluid Dynamics Laboratory under the Department of Mechanical Engineering at IIT Madras, led a team of researchers to develop the AI and Deep Learning-based algorithms to solve various engineering issues. Sources state that the algorithms will be taking a data-driven approach to come up with solutions instead of solving a physics law or a physics-based partial differential equation (PDE).
As per sources, the AI and Deep Learning-based algorithms is a new concept and has been worked upon by limited research groups, globally. However, unlike IIT Madras’ algorithms, the other research groups utilize CNN or Convolutional Neural Networks or Conditional Generative Adversarial Network (C-GAN) to develop solutions to the engineering problems.
Dr. Vishal Nandigana highlighted several aspects of the algorithms, stating that AISoft has undergone different testing procedures to solve thermal management issues. The tests revealed that the software is a million-times quicker to solve different issues in comparison to the current methods. He added that the algorithms will be the solution to removing the bottleneck of developing solutions to most engineering obstacles, which is the computational time.
Dr. Nandigana further stated that the researchers used a “data-driven AI and Deep Learning Model” in order to come up with solutions for the engineering issues. Artificial Intelligence is first trained with data sets, which further facilitates the development of the solutions. He added that the data sets can either be acquired from current big data significant to the industry, where pools of experimental data are available. In areas with limited or no data, the AI can be trained with the help of Computational Fluid Dynamics software.
The Professor affirmed that the algorithms will be using a Recurrent Neural Network (RNN) and Deep Neural Network (DNN), a unique method, to develop engineering issues. He stated that the IIT Madras AI-algorithms will not need information from the left and right grid in order to solve the grip Point of Interest, making it independent of grid and mesh. He also declared that the AI will be to utilize scarce data sets and develop solutions, further making it unique from commercially available software.
Hardware products have also been developed by the researchers at IIT Madras with the help of multi-threading processing and GPU. The hardware is able to solve problems in the thermal and electronic colling industries. Using both software and hardware products, the solutions are developed quicker than commercially available open-source software and numerical method software.
The development of the algorithms will clear up different crucial issues in the industries, while also being used for the purposes of education.