Publicación: MACHINE LEARNING TO PREDICT VENEER DRYING QUALITY IN THE PINUS RADIATA PLYWOOD INDUSTRY
Fecha
2024
Título de la revista
ISSN de la revista
Título del volumen
Editor
MADERAS: CIENCIA Y TECNOLOGIA
Resumen
MACHINE LEARNING IS A TOOL THAT IS BEING USED TO OPTIMIZE HIGHLY COMPLEX INDUSTRIAL PROCESSES. IN THE PLYWOOD PANEL PRODUCTION INDUSTRY, VENEER DRYING IS ONE OF THE MOST IMPORTANT PROCESSES AS IT ALLOWS TO OBTAIN HIGH QUALITY PRODUCTS. THE BIOLOGICAL NATURE AND HIGH STRUCTURAL VARIABILITY OF WOOD TURNS ITS INDUSTRIAL PROCESSING
MULTIVARIATE AND DIFFICULT TO CONTROL. THE LARGE NUMBER OF VARIABLES PRESENT AND THE INCREASING POSSIBILITY TO MEASURE THEM IN REAL TIME ARE ENABLING THE AVAILABILITY OF A LARGE AMOUNT OF DATA. NOWADAYS, DATA-DRIVEN APPROACH AND INTELLIGENCE ARTIFICIAL TECHNIQUES, SPECIFICALLY MACHINE LEARNING CAN ENABLE ROBUST PREDICTION AND CONTROL APPROACHES. IN THE PROCESS INDUSTRY, WITH HIGH LEVELS OF AUTOMATION, IT IS POSSIBLE TO ENABLE DECISION MAKING TO PREDICT PRODUCT QUALITY BY MONITORING EXPLANATORY CONTROL VARIABLES. THE OBJECTIVE OF THIS WORK WAS TO EVALUATE A MACHINE LEARNING ALGORITHM CAPABLE OF PREDICTING THE QUALITY OF THE VENEER DRYING PROCESS FROM A CONSIDERABLE NUMBER OF INPUT VARIABLES CAPTURED FROM A REAL INDUSTRIAL PROCESS. THE WEKA PLATFORM AND PYTHON CODE WERE USED. THREE ALGORITHMS WERE EVALUATED: K-NEAREST-NEIGHBOR, EXTREME GRADIENT BOOSTING AND
SUPPORT VECTOR MACHINE. VARIABLE AND DIMENSIONALITY REDUCTION, CORRELATION ANALYSIS AND PRINCIPAL COMPONENT ANALYSIS WERE PERFORMED. THE RESULTS SHOWED THAT EXTREME GRADIENT BOOSTING ACHIEVED AN ACCURACY OF 76 % IN PREDICTING QUALITY SCORES. FINALLY, IT IS CONCLUDED THAT BOTH THE DATA ENGINEERING METHODOLOGY AND THE EVALUATED ALGORITHMS WERE EFFICIENT IN PREDICTING INDUSTRIAL DATA.
Descripción
Palabras clave
wood quality, veneer drying, process optimization., predictive model, Machine learning algorithms
