Research Projects

Research Gap : Estimating non-destructively the volume of tender coconut remains a challenge as no device exists for the same. Traditional methods, such as manual tapping or weight-based assessments, are subjective, whereas advanced techniques like MRI and X-ray imaging are accurate but prohibitively expensive and impractical for large-scale use. A scalable device that leverages acoustic and vibration signals remains unexplored and could optimize quality control, pricing strategies, and consumer satisfaction in the coconut industry. Scientific Rationale and Importance : Traditional approaches are inconsistent, and sophisticated imaging technologies are inaccessible for routine application. Studies have shown that acoustic signals and mechanical vibrations effectively predict internal properties of fruits and assess maturity, demonstrating potential for tender coconut water volume estimation. This study hypothesizes that a tailored device utilizing these principles could provide an affordable, accurate, and repeatable solution for the coconut industry. Objectives 1. Design and develop systems for acoustic and mechanical vibration-based estimation of tender coconut water volume. 2.Validate both methods, assess predictive accuracy, and develop machine learning (ML)-based predictive models. 3.Select the superior method (or a hybrid) and integrate it into a non-invasive device. Methodology Phase 1: Method Validation Acoustic Method: A tapping mechanism will generate sound waves, recorded using an omnidirectional microphone in a soundproof chamber, Denoising and frequency domain transformation will extract features such as resonant frequency, amplitude, and sound decay rates Mechanical Vibration Method: A vibrational exciter will induce vibrations, captured by piezoelectric sensors. A diverse sample of tender coconuts will be used, spanning various maturity stages. Signals from both methods will be recorded, and actual water volumes will be measured to validate predictions. Phase 2: Modelling and Data Analysis Feature Extraction: Signal features will undergo normalization and scaling. Modelling: ML models like Support Vector Regression (SVR) and Random Forest (RF) will predict water volume. Comparative Analysis: Determine the superior method or develop a hybrid model using combined features. Phase 3: Device Development Includes tapping mechanism or vibrational exciter, sensors (microphones or piezoelectric), data acquisition module, and ML-based predictive software. Calibration and Testing: The device will be calibrated using known water volumes and tested under various conditions to ensure reliability.