Plasma control by pattern recognition in laser induced breakdown spectroscopy

A low-computational intensive laser control approach is proposed for implementing an embedded control system, using pattern recognition by relevant principal component analysis for laser induced breakdown spectroscopy applications. The laser energy is directly related to the resulting spectral pattern and is determined by iterations in the feature space. Results show that single shot iterations until optimum energy can be significantly reduced by pattern recognition. A performance benchmark with minerals, alloys and pellets from material collected from a drill demonstrated an average of 50% improvement, significantly reducing sample deterioration and improving measurement safety.