A carbon emission prediction model of mechanical parts machining based on structural design parameters

Under the Industry 5.0 paradigm that prioritizes human-centric and sustainable manufacturing system design (MSD), the structural design of mechanical parts has a direct impact on the carbon emissions generated during machining. However, accurately predicting carbon emissions stays challenging due to the complex mechanistic relationship between the structural design parameters (such as size and shape) and the resulting emissions. Consequently, implementing low-carbon structural design becomes a formidable task. To address these challenges, a carbon emission prediction model for mechanical parts machining driven by structural design parameters is proposed. First, the influence of structural design parameters and machining processes on carbon emissions is analyzed, and the structural design parameters are categorized accordingly. The initial sample set is constructed using the optimal Latin hypercube sampling (OLHS) method. Subsequently, a radial-basis function neural network (RBFNN) model is developed for predicting carbon emissions, driven by structural design parameters, considering the complex nonlinearity resulting from multiple structural design parameters. The accuracy of the training model is enhanced by implementing the Regularization Coefficient and K-Fold Cross Validation method. Finally, the effectiveness of the proposed method is verified through a case study involving reducer gear machining.