基于非线性动力分析的脑力疲劳评估和预测应用基础研究

With the accelerated pace of life and increased work pressure, mental fatigue is increasingly becoming one of the biggest negative factors which affect human's mental health and work efficiency. Traditional assessment methods for mental fatigue such as subjective scales and behavioral testing show low reliability and require subjects to take the initiative to the experiments. In recent years, a number of electrophysiological parameters have been applied into the research of mental fatigue.Usually,electrophysiological signals are acquired from nonlinear, time-varying and non-stationary complex systems. Therefore,traditional analytical methods such as power spectrum analysis have significant limitations to analyze such signals. Taking into account the nonlinear dynamics of human brain system and the limitations of a single method, this project proposes the application of nonlinear dynamics theory combined with traditional signal processing methods to study the EEG, ECG, and other electrophysiological signals, designed to reveal the nonlinear nature of mental fatigue and have a better insight into the rhythm of human body.A systematic evaluation based on subjective questionnaires, behavioral assessment and electrophysiological indicators are proposed. This evaluation system can provide a comprehensive evaluation for mental fatigue from both subjective and objective viewpoint. Machine learning technique will be used to give quantitative risk assessment and realize an early warning for mental fatigue. The research will provide a theoretical basis for the real-time mental fatigue monitoring technology based on multi-parameter signals and contribute to an effective means to reduce traffic accidents, operational failures and casualties caused due to mental fatigue.

随着生活节奏的加快和工作压力的加大，脑力疲劳日益成为影响人们身心健康和工作效率的最大负面因素之一。脑力疲劳的传统评估方法如主观量表和行为学测试存在信度低且需要被试者主动配合等局限性。近年来，一些电生理指标已经介入脑力疲劳的研究中，但由于电生理信号源于非线性、时变、非平稳的复杂系统，故而传统信号处理方法存在很大的局限性。考虑到大脑系统的非线性动力学特性和单一研究方法存在的局限性，本项目提出运用非线性动力学理论结合经典信号处理方法来研究脑电、心电等多种电生理信号，旨在揭示脑力疲劳的非线性本质和更好地洞察人体的节律。在电生理指标研究基础上，本课题将进一步融合主观量表和行为学指标，利用机器智能学习方法来系统、综合、定量地对脑力疲劳进行风险评估和早期预警研究。本课题的研究成果将为多参数信号脑力疲劳实时监测技术提供理论基础，更为降低因脑力疲劳造成的交通事故、操作故障、人员伤亡提供理论依据和有效手段。

本课题围绕“非线性动力学理论在电生理信号分析中的应用”和“基于多信息融合的脑力疲劳分级评估与早期预警”两大科学问题展开研究，主要取得了以下几方面的研究成果：1）运用非线性动力学理论结合经典信号处理方法来研究脑电、心电等多种电生理信号，揭示了脑力疲劳的非线性本质。2）将电生理信号的分析与主观量表评价、行为指标相结合，实现了对脑力疲劳的系统、综合、定量评估。3）应用统计学习和机器智能学习等理论和方法，在已有算法的基础上，改进和探索新算法比如极限学习机（ELM）算法，提高了系统实时处理的速度和准确度。4）将课题研究成果应用于脑机接口领域，探讨了在人机交互过程中的脑力疲劳和视觉疲劳问题，使脑机接口的实验开展更加人性化，减轻了被试者的疲劳感。5）研究和试制了脑电、心电采集的硬件系统，为多参数生体电信号采集和监测技术提供了理论基础和原型样机。