| With the rapid development of economy and the continuous emergence of new scientific technologies,manufacturing enterprises have gradually entered the production track of informatization and intelligence.The increasing demand for customization,such as product personalization,complexity,and diversification,has led to significant changes in the nature of assembly work.Workers need more cognitive ability to meet the everevolving requirements of assembly processes.However,prolonged mental labor may lead to overloaded cognitive burden,which can have adverse effects on production efficiency and system safety of the enterprise.Therefore,adopting a “human factor” perspective and using scientifically effective methods to monitor and evaluate the mental workload of workers during the assembly process can serve as a starting point for optimizing human-machine collaboration.The evaluation results can act as a basis for optimizing the division of labor between humans and machines,which is beneficial in improving the overall assembly efficiency and reducing the error rate caused by human factors,ultimately promoting the humanmachine efficiency of the manufacturing system.To begin with,this article utilized the existing laboratory equipment to design three different levels of assembly tasks in a simulated assembly work.During the task execution,the subjective data of the operators were collected using NASA-TLX,and the whole experimental process was recorded using the Ergo LAB cloud platform,which also collected the pulse,respiratory and electromyographic signals.Additionally,the BP EEG equipment was used to collect the EEG signals.Secondly,this paper uses the combination of subjective evaluation measurement method and physiological index measurement method to study mental workload.Different changes of peripheral physiological signals and EEG signals will cause changes in mental workload.The energy spectrum characteristics of EEG,the heart rate variability of pulse signal,the respiratory mean value and respiratory amplitude of respiratory signal,and the root mean square and median frequency of EMG amplitude were extracted.The collected subjective scores,task performance and physiological data were analyzed by SPSS statistical analysis software for analysis of variance,T-test and correlation analysis,and the characteristic indexes with significant differences were extracted.The results showed that respiratory index increased significantly with the increase of task difficulty,pulse index had no significant difference among tasks,Alpha energy wave of EEG signal had significant difference among tasks,and EMG index had no significant difference among tasks.Finally,this paper proposed a multi-modal physiological indicator evaluation method for assessing the mental workload of assembly workers using information fusion theory.The study trained various classification models on the extracted features at the feature level,and compared and analyzed these models with other classifiers.The classification results showed that,among the single-modality indicators,the accuracy of EEG classification was higher than that of respiratory and pulse indicators.Among the bimodal combination indicators,the accuracy of EEG and PPG combination signals was higher than that of EEG and respiratory combination signals and PPG and respiratory combination signals.Lastly,the classification accuracy of multi-modal data was generally superior to that of single-and bimodal data,with the SVM model achieving the highest classification accuracy.In the human-centered assembly operation,under the condition of ensuring production efficiency,the changes of operator’s mental workload are monitored in real time,and the state characteristics of mental workload and work performance are analyzed.It plays an important role in improving the physical and mental health of operators,improving their work efficiency and formulating effective management strategies,and has important practical significance. |
PDF Full Text Request