基于可解释机器学习算法的工业信息结构与多模态生理认知耦合模型OA

Under the background of industrial intelligent transformation,the increasing complexity of industrial information leads to increased demand of cognitive resources for human-computer interaction.This study constructed a coupling model of industrial information structure and multimodal physiological cognition based on an explainable machine learning algorithm,revealing the dynamic association through a four-layer architecture(data input-feature extraction-coupling analysis-result output).The industrial scenario was simulated by auditory n-back task experiments.Physiological signals such as electroencephalogram,electrooculogram and electromyogram were collected with industrial information complexity as the independent variable,and the redundant features were removed by Pearson correlation coefficient(threshold>0.6).The results show that among the six machine learning algorithms,extreme gradient boosting(XGBoost)has the best performance with an accuracy of 93.1%and an F1-score of about 0.92.The Shapley additive explanations(SHAP)show that Delta waves are the main feature characterizing the cognitive load of this experiment and show a bidirectional influence mechanism on the classification of industrial information complexity at level 0.The model breaks through the limitations of single modality and quantitatively reveals the nonlinear coupling relationship between industrial information complexity and multimodal physiological signals,which provides a data-driven and interpretable decision-making basis for the optimization of industrial human-machine interfaces.