High-throughput data mined prediction of design and preparation of flexible carbon-based conductive materials in energy storage

Abstract

Carbon-based material have been attracted significant attention and widely studied for its excellent electrical conductivity. The prediction of conductivity of carbon-based materials is indispensable to designing and fabricating flexible carbon conductive materials in smart wearable before experimentation. To achieve this objective, ensemble data mining is utilized to automatically search for relationship between factors (filler type, output conductivity, filling fraction, drying temperature, molding temperature, sample thickness, ultrasonic treatment time, stirring time, etc.) and the target (composite conductivity), providing exceptional insight into the design and preparation ability of flexible carbon conductive materials. Specifically, we established an uni-component carbon filler database and proposed an optimized machine learning model based on Light Gradient Boosting Machine. The importance ranking of the characteristic variables in the fabrication of flexible carbon composites has been demonstrated in terms of the constructed uni-component carbon filler database. We trained our developed Light Gradient Boosting Machine model on selected data and attempted to predict the optimal design strategy for flexible carbon-based conductive materials in energy storage in smart wearable. The appealing results revealed effectiveness and dependability of Light Gradient Boosting Machine. The strategy established in this work presents an effective approach for accelerating the research and development of flexible carbon-based conductive materials in energy storage in smart wearable.