A Novel Stacked Generalization Ensemble-Based Hybrid PSVM-PMLP-MLR Model for Energy Consumption Prediction of Copper Foil Electrolytic Preparation

Zhiming Liao,Mei Su,Guangfu Ning,Jie Zhou,Yonglu Liu,Tong Wang

doi:10.1109/access.2020.3048714

Abstract

At present, the energy consuming during the electrolytic copper foil preparation accounts for more than 75% of the total energy consumption. In real-life production, the process parameters are set by the operator empirically and the system may not work at the operating point with minimum energy consumption. Therefore, it is critical to establish an effective model for predicting electrolysis energy consumption to guide the parameters design. In this paper, a novel hybrid model (named PSVM-PMLP-MLR) based on stacked ensemble learning is proposed. The model is divided into two parts: the base-learning model and the meta-learning model. The support vector machine (SVM) model and multilayer perceptron (MLP) model with different input structures are established by the former first. Then the particle swarm algorithm is employed to determine the optimal value of SVM parameters and the optimal weight of MLP by minimizing the mean absolute percentage error (MAPE). The multiple linear regression (MLR) is finally employed as a meta-learning machine to compute the final predictions. Experimental results show that the regression coefficient of this model reached 0.987, and compared with the traditional SVM and MLP models, the accuracy of the model is improved by 10.29% and 8.28%, respectively.

Highlights

With the continuous development of 5G, industrial intelligence, and new energy vehicles, etc., the demand for copperclad laminate (CCL) and printed circuit board (PCB) is growing rapidly
Six machine learning (ML) models have been introduced in this part, including support vector machine (SVM), multilayer perceptron (MLP), multiple linear regression (MLR), SVM and MLP based on particle swarm optimization (PSO) optimization and PSVM-PMLP-MLR hybrid model. 75% of the production data of electrolytic copper foil were utilized as the training data while the remaining 25% as the test data
To further improve the accuracy, a novel PSVM-PMLP-MLR hybrid model is proposed which based on stacked ensemble learning

Summary

Introduction

With the continuous development of 5G, industrial intelligence, and new energy vehicles, etc., the demand for copperclad laminate (CCL) and printed circuit board (PCB) is growing rapidly. As the basic electronic material, electrolytic copper foil is indispensable and influences the conductivity of circuits and the interconnection of electronic components in producing CCL and PCB. During the whole production process, the electrolytic preparation consumes more than 75% of energy consumption. It is imperative to reduce this energy consumption to save cost. The associate editor coordinating the review of this manuscript and approving it for publication was Firooz B.

Objectives

Methods

Results

Conclusion