Intelligent LED Certification System in Mass Production

Area of Science:

• Electrical Engineering
• Materials Science
• Signal Processing

Background:

• Accurate measurement of light-emitting diode (LED) efficiency is vital for applications in medicine and telecommunications, especially for automated mass production.
• Existing noise suppression filters (IIR, FIR, adaptive) are insufficient for reliable LED efficiency measurements due to noise interference in electrical circuits.
• Blind signal extraction methods offer a promising alternative as they do not require a reference signal, unlike adaptive filters.

Purpose of the Study:

• To develop and present a novel method for measuring LED efficiency by effectively suppressing noise in electrical measuring circuits.
• To introduce a blind signal extraction technique based on cascading neural networks for reconstructing waveforms in LED efficiency measurements.
• To enhance the reliability and accuracy of LED efficiency measurements, particularly in automated mass production scenarios.

Main Methods:

• A sequential blind signal extraction method utilizing a cascading neural network architecture.
• Statistical analysis of signal and noise probability density functions (PDFs) to identify distinct characteristics.
• Optimization of neural network parameters using generalized statistical moments (skewness and kurtosis) as the objective function, with moment order determined by maximum Mahalanobis distance.

Main Results:

• The proposed method successfully extracts the signal from noisy electrical measurements of LED cooling curves.
• The use of generalized statistical moments effectively characterizes the differing PDFs of the signal and noise.
• The cascading neural network approach enables accurate reconstruction of the waveform for multi-temporal comparison of crystal cooling curves.

Conclusions:

• The developed blind signal extraction method significantly improves the accuracy of LED efficiency measurements.
• This technique is particularly valuable for automated mass production, ensuring reliable quality control.
• The method offers a robust solution for noise reduction in sensitive electrical measurements relevant to optoelectronic devices.