Process Time Prediction for Metal Structural ComponentsBased on the IWOA-BPNN Model

To address the difficulty of accurate process time prediction due to the numerous and interrelated stages in the manufacturing of large structural components, a solution framework of “feature extraction-model construction-accuracy enhancement-result comparison” is proposed. Based on historical data, principal component analysis (PCA) is used to efficiently filter the feature parameters affecting process time prediction, thereby reducing data redundancy. The structure and initial parameters of a BPNN prediction model are designed to estimate the minimum process time. An improved whale group algorithm is developed to optimize the initial weight and threshold and improve the prediction accuracy. Augmented data generated by Plant Simulation are combined with historical data to form a sample dataset, which is used to verify the effectiveness of the model and the accuracy enhancement method. Results show that the method proposed in this paper achieves smaller error indicators, faster iteration speed and better optimal fitness values, which provides a novel solution for the accurate prediction in process time of large structural components.