Design and Implementation of a Low-Cost Capacitive Sensor Array for Plastic and Textile Classification

Resumo

The present paper introduces the design and implementation of a simple capacitive sensor array to achieve the accurate classification of plastic and textile materials. A capacitance-to-digital conversion system was interfaced with an 8×8 electrode grid constructed on FR4 to record changes in spatial capacitance. Differentiated patterns in response were found whereby plastics had an average capacitance that was 12-18 percent lower than that of textiles. The extracted statistical features such as the mean and the standard deviation facilitated good separation between classes with an increase of 27 percent in inter-class distance after normalization. The overall classification accuracy of a supervised machine learning model was 94.6% with a precision of 93.8 and a recall of 95.2 with a misclassification rate of less than 5 percent. The system was found to be robust with accuracy maintained to greater than 90 percent in the presence of noise up to 15 percent signal distortion. Comparative analysis showed a 18 per cent improvement to baseline procedures. The findings affirm the efficiency of the proposed system as a cost-effective and dependable tool of automated classification of materials in recycling and waste management processes.