An internship experience at ZF Wind power

During my internship at ZF Wind Power Coimbatore Private Limited, which spanned from July 9th to August 8th, 2024, I had the opportunity to apply my academic knowledge in Artificial Intelligence and Data Science to real-world challenges within the quality control department. I was part of a team of four members they are, Arun Prasath , T. Dilan Melvin & Jaya Surya . Our internship experience was guided and supported by our mentor, Mr Arun Sundar S , whose insights and expertise were invaluable throughout the program.

About ZF Wind Power

ZF Wind Power is a global leader in providing advanced gearbox solutions for both onshore and offshore wind turbines. The company’s commitment to innovation and high-quality manufacturing is evident in its global presence and cutting-edge facilities, including the state-of-the-art plant in Coimbatore, India.

Our Internship Projects

1. Concentration Drop Prediction for Parts:

?? - Objective: Our first project focused on identifying and quantifying the concentration drop caused by each part, which isn’t directly measurable by sensors.

?? - Methodology: We gathered and cleaned four years’ worth of historical data, then used a Weighted Average Calculation algorithm to determine the concentration drop attributed to each part. This real-time calculation allowed for better monitoring and adjustments to part usage, ensuring optimal concentration levels were maintained.

?? - Outcome: We developed an interface to store the calculated values, which captured crucial data such as part types, usage dates, and current concentration levels.

2. Time Series Prediction of Concentration Values:

?? - Objective: The goal of our second project was to predict future concentration values to ensure they remained within specified thresholds.

?? - Approach: We implemented two predictive models—a Gradient Boosting Regressor and a Long Short-Term Memory (LSTM) model. Both were trained on historical data, with the LSTM model proving particularly effective in capturing temporal dependencies.

?? - Outcome: These models provided accurate predictions, which are essential for maintaining optimal concentration levels and preventing potential issues related to dips in concentration.

Conclusion

This internship was a significant milestone in our education, allowing us to apply the theoretical knowledge we’ve gained to practical challenges in a real-world setting. Working together as a team, and under the guidance of Mr Arun Sundar S , we were able to enhance our technical skills and build a solid foundation for our future careers in engineering and data analysis.