Intelligent System for Real-time Health Monitoring of Power MOSFETs

Project Title

Intelligent System for Real-time Health Monitoring of Power MOSFETs

Project Area of Specialization Artificial IntelligenceProject Summary

Power MOSFETs are used in a wide range of industries like Power Electronics, Power control, etc. It is necessary to understand the failure mode and the effect of the performance of the MOSFET when subjected to thermal stress. The most common cause of a short circuit is a broken bond between source and drain. A second typical failure condition is the formation of an oxide on the gate of the Power MOSFET. Oxide formation gets worse as the Power MOSFET body temperature increases due to the high switching frequency.

The RUL is an essential parameter in MOSFET reliability. However, this parameter cannot be directly measured, nor can it be expressed based upon the current or voltage ratings of the device. Data-driven techniques will be used to predict the RUL of power MOSFETs. Python libraries like 'NumPy, pandas, and scipy', will be used to build and train neural networks on Jupyter Lab. Determining how much to apply voltage to the gate of the power MOSFET will maintain its performance.

The overall process includes:

• MOSFET Data Collection (NASA Ames Prognostics Repository)
• Data Cleaning and Pre-processing
• Applying Regression and Classification Techniques to predict RUL
• Health Assessment

Project Objectives

We'll use a data-driven technique that uses neural networks to predict the performance of a power MOSFET. Neural networks are used to analyze the data and determine the degradation of the power MOSFETs. We can do this more quickly and inexpensively than if we had to collect and analyze that data ourselves.

NASA has already collected large amounts of data on MOSFETs. This is very useful for determining their degradation, but the data alone isn't enough. We'll use neural networks to determine their prognosis. Using the Python libraries, like 'NumPy, pandas, and scipy', we can analyze the data to make predictions about power MOSFET degradation.

Project Implementation Method

1. MOSFET Data Collection (NASA Ames Prognostics Repository)
2. Data Cleaning and Pre-processing using Matlab & Python
3. Applying Regression and Classification Techniques to predict Remaining Useful Life of MOSFETs
4. Health Assessment

Benefits of the Project

The benefits of our prognostics project are many.

For one, it reduces the amount of time and money spent on diagnosing and replacing parts that are still functional. For another, it can help manufacturers improve the quality of their products by providing early data about possible issues with production processes.

Finally, our prognostics project will make it easier for industries to take advantage of opportunities in the market.

Technical Details of Final Deliverable

A machine-learning and deep-learning-algorithm backed research results that will help future researchers and industries in gaining an in-depth knowledge of the system health assessment - specifically MOSFET degradation.

Final Deliverable of the Project Software SystemCore Industry OthersOther IndustriesCore Technology Artificial Intelligence(AI)Other Technologies Big DataSustainable Development Goals Quality Education, Decent Work and Economic Growth, Industry, Innovation and Infrastructure, Life on LandRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	7000
Research Paper Printing & Compiling	Miscellaneous	1	7000	7000