Fast and Accurate Quality Prediction for Injection Molding: an Improved Broad Learning System Method
Jianghao Lin,Zhigang Ren,Zongze Wu,Zhouhao Ouyang,Aimin Yang
DOI: https://doi.org/10.1109/jsen.2023.3346849
IF: 4.3
2024-01-01
IEEE Sensors Journal
Abstract:Automatic monitoring of product quality has always been the core of intelligent development of injection molding industry. However, there exist several challenges for quality prediction such as the complexity of multi-sensor data processing and feature selection, as well as the imbalance and small-sample problems caused by the randomness of defective sample collection. To tackle these problems, this study develops a quality prediction model for injection molded products via combining the p-Norm optimization method and Bi-Enhancement Broad Learning System, namely pNBEBLS. To ensure the feature representativeness of the products, we collect 192 features and extract 20 typical ones based on Spearman correlation analysis. The raw extracted features are input into the feature layer and the linear features are thus obtained. Meanwhile, the linear features are changed into the enhanced non-linear features via both enhancement layer and learned enhancement feature layer. Then, the proposed model adopts both linear and non-linear features as input defined as A, and it is multiplied by the weight matrix W to get the predicted output Y. It is noted that in the process of training, p-Norm method is employed to optimize the weight matrix W in output layer, while in the process of testing, W is used directly for prediction with the three-dimensional sizes of products as predicted targets. The comparative experiments are then carried out between the proposed method and methods like SVR (Support Vector Regression), KNN (k-Nearest Neighbor), MLP (Multi-Layer Perceptron), RF (Random Forest), CNN (Convolutional Neural Network), BiGRU (Bidirectional Gated Recurrent Unit) and RBLS (Regression Broad Learning System). Experimental results show that the proposed pNBEBLS can obtain the lowest MSE and MAPE values, and highest R2 scores for size1, size2 and size3 prediction tasks, respectively. In practical size categories detection applications, the ablation experiment results show that the p-Norm and Bi-Enhancement function mechanisms can effectively improve the accuracy of BLS, and the proposed pNBEBLS can obtain the highest accuracy rates of size1, size2 and size3, they are, 98.77%, 96.67% and 97.28%. In addition, results of time comparison experiment indicate that the proposed method has fast size prediction response due to the superior framework of Broad Learning System. In a nutshell, the pNBEBLS is able to predict the product quality with higher accuracy, stability and robustness.