CN-121987202-A - Geological operation-oriented real-time monitoring method and device for mental fatigue of operators

CN121987202ACN 121987202 ACN121987202 ACN 121987202ACN-121987202-A

Abstract

The invention discloses a real-time monitoring method and device for mental fatigue of operators facing geological operations, comprising the steps of S1, determining layout of a measured target and a millimeter wave radar, S2, acquiring a millimeter wave radar intermediate frequency echo, extracting complex phases of corresponding distance information after the measured target is positioned by a distance domain, performing phase expansion and steady state correction to obtain a continuous phase sequence representing chest micro-displacement, S3, applying IIR digital band-pass filtering to the continuous phase sequence and decoupling the continuous phase sequence to form a respiratory channel and a heartbeat channel, S4, extracting time domain and frequency domain features of heart rate variability HRV and respiratory rhythm from the respiratory channel and the heartbeat channel in a sliding time window to form a multidimensional vital sign feature vector set, S5, performing incremental learning by using an OS-ELM and judging and outputting mental fatigue grade in real time, S6, implementing hierarchical risk management, triggering early warning and intervention measures, and S7, performing local storage and visual display on fatigue grade and event log and synchronizing to a remote management platform.

Inventors

LI YIFENG
WANG JIA
WANG LEI
HAN SHUAI
LI HENG
HONG ZHENGQIANG

Assignees

中国建筑土木工程有限公司
香港理工大学

Dates

Publication Date: 20260508
Application Date: 20251218

Claims (10)

1. The real-time monitoring method for the mental fatigue of the operators facing the geological work is characterized by comprising the following steps of: S1, determining the arrangement of a detected target and millimeter wave radars in a cockpit; S2, acquiring a millimeter wave radar intermediate frequency echo, positioning a measured target through a distance domain, extracting complex phases of corresponding distance information, and performing phase expansion and steady-state correction to obtain a continuous phase sequence representing chest micro-displacement; S3, applying IIR digital band-pass filtering to the continuous phase sequence and decoupling the continuous phase sequence into a respiratory channel and a heartbeat channel; S4, extracting time domain and frequency domain features of heart rate variability HRV and respiratory rhythm from a respiratory channel and a heartbeat channel in a sliding time window to form a multidimensional vital sign feature vector set; s5, taking the multidimensional vital sign feature vector set as input, performing incremental learning by using an online sequence extreme learning machine OS-ELM, and judging and outputting the mental fatigue level in real time; s6, implementing hierarchical risk management according to the mental fatigue level, and triggering early warning and intervention measures; s7, carrying out local storage and visual display on the fatigue grade and the event log, and synchronizing to a remote management platform.
2. The real-time monitoring method for mental fatigue of the operators facing geological work is characterized in that in the step S1, a millimeter wave radar is installed at a fixed position facing to the chest and shoulder area of the operators in a cockpit, a visual field covers the upper body movement range of the operators and keeps a stable geometric relation with the seat posture, a base of the millimeter wave radar is fixed through vibration reduction and is grounded through electromagnetic shielding, in the step S6, hierarchical risk management comprises voice prompt, illumination adjustment and short-time rest, in the step S7, a timestamp, anonymous identification of equipment and personnel, data quality score, mental fatigue prediction level and treatment result are recorded, and historical backtracking and strategy optimization are supported.
3. The real-time monitoring method for mental fatigue of operators for geological operations according to claim 1, wherein step S2 specifically comprises: s201, performing fast Fourier transform FFT on the obtained radar intermediate frequency echo in a distance dimension, positioning a measured target and determining corresponding distance information; s202, extracting complex echoes from the distance information, performing phase calculation by adopting an arctangent function, and performing 2 pi phase unwrapping to obtain a continuous phase sequence; s203, performing phase difference calculation and steady state correction on the continuous phase sequence to inhibit slow posture drift and strengthen chest wall micro-displacement components, and obtaining a continuous phase sequence representing chest micro-displacement.
4. The real-time monitoring method for mental fatigue of operators for geological work according to claim 1, wherein in the step S3, a filter bank consisting of two IIR butterworth band-pass filters is constructed, one filter bank is used for forming a respiratory channel by low-frequency components related to respiration and suppressing high-frequency components, and the other filter bank is used for forming a heartbeat channel by high-frequency components related to heartbeat and suppressing low-frequency components and respiratory harmonics.
5. The real-time monitoring method for mental fatigue of operators facing geological operations according to claim 4, wherein the breathing channel passband is 0.1-0.5 Hz, the stop band is 0.05-0.55 Hz, the heartbeat channel passband is 0.8-2.0 Hz, and the stop band is 0.7-2.1 Hz.
6. The real-time monitoring method for mental fatigue of operators for geological work according to claim 1, wherein in step S4, sliding time windows are divided by a window length of 60 seconds and an overlapping rate of 50%, and the time domain features and the frequency domain features of heart rate variability HRV and respiratory rhythm are extracted in each sliding time window.
7. The real-time monitoring method for mental fatigue of operators for geological operations according to claim 1, wherein the step S5 specifically comprises: S501, classifying mental fatigue states of operators into low, medium and high categories at a work starting or preset node through subjective questionnaires so as to form an initial calibration sample; S502, taking the multidimensional vital sign feature vector set formed in the step S4 as input, constructing an online sequence extreme learning machine OS-ELM with a single hidden layer feedforward structure, and carrying out real-time online judgment on a continuous phase sequence according to a process of prediction and updating and outputting a mental fatigue level; S503, performing recursive sequential increment updating on an online sequence extreme learning machine OS-ELM under the data stream arrival condition, introducing a forgetting factor lambda to reduce the influence of historical samples and improve the weight of recent samples, wherein the data stream arrival condition refers to that after a multidimensional vital sign feature vector set generated by a continuous sliding time window and a fatigue grade mark corresponding to the multidimensional vital sign feature vector set are written into an updating cache, the number of effective samples in the cache reaches a preset batch size.
8. A real-time monitoring device for mental fatigue of operators for geological operations, based on the monitoring method according to any one of claims 1-7, characterized in that it comprises: The data acquisition unit is used for acquiring a millimeter wave radar intermediate frequency echo through a millimeter wave radar measured target, extracting complex phases of corresponding distance information after the measured target is positioned by a distance domain, and carrying out phase expansion and steady-state correction to obtain a continuous phase sequence representing chest micro-displacement; The data processing unit is used for applying IIR digital band-pass filtering to the continuous phase sequence and decoupling the continuous phase sequence into a respiratory channel and a heartbeat channel, extracting time domain and frequency domain features of heart rate variability HRV and respiratory rhythm from the respiratory channel and the heartbeat channel in a sliding time window, and forming a multidimensional vital sign feature vector set; the online prediction unit is used for taking the multidimensional vital sign feature vector set as input, performing incremental learning by using an online sequence extreme learning machine OS-ELM and judging and outputting the mental fatigue level in real time; And the feedback regulation unit is used for implementing hierarchical risk management according to the mental fatigue level, triggering early warning and intervention measures, carrying out local storage and visual display on the fatigue level and the event log, and synchronizing to a remote management platform.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the geological job oriented operator mental fatigue real-time monitoring method according to any of claims 1 to 7 when the computer program is executed by the processor.
10. A computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor realizes the steps of the geological job oriented operator mental fatigue real-time monitoring method according to any of claims 1 to 7.

Description

Geological operation-oriented real-time monitoring method and device for mental fatigue of operators Technical Field The invention relates to the technical field of geological engineering operation safety monitoring and human engineering, in particular to an operator mental fatigue real-time monitoring and grading early warning method for geological operation scenes based on millimeter wave radar and an online extreme learning machine (OS-ELM). Background The geological engineering operators need to continuously acquire, process and execute the information related to the dangerous sources in a complex dynamic working environment, and long-term high-strength attention and decision-making requirements easily lead to the accumulation of cognitive load and the induction of mental fatigue, so that the reliability of a perception-judgment-operation chain is weakened. Therefore, continuous monitoring of mental fatigue during the working phase is critical. The existing measuring paths are mainly divided into three types, namely a subjective self-evaluation method, such as NASA Task Load Index and a Borg RPE 6-20 equivalent scale, which have time lag and expected deviation and interrupt the operation flow, a wearable sensing method, such as a wrist strap, EEG, an eye movement instrument and the like, which can objectively and continuously monitor, but has poor wearing compliance and comfort in high-dust, high-noise vibration and high-temperature and humidity environments, and high maintenance cost, and a visual method, which is significantly affected by illumination and shielding and acquires privacy and compliance pressure caused by sensitive information such as faces and the like. There is a need for a non-contact sensory monitoring technology that is insensitive to illumination and shielding, privacy friendly and can operate stably for a long period of time in the cockpit. The millimeter wave radar has insensitivity to dust and illumination variation and high resolution capability to fine displacement, and can sense chest wall micro-vibration through distance domain positioning and phase processing on the premise of not acquiring facial images, so that vital sign signals such as respiration and heartbeat are separated. Compared with the radio frequency sensing with Wi-Fi and other longer wavelengths, the millimeter wave wavelength is matched with the chest wall displacement magnitude, and the heart and lung rhythm characteristics can be recovered under multipath and noise background. Existing studies have validated millimeter wave radars for their effectiveness in short-time heart rate and respiration rate measurements, but how to effectively characterize the gradual accumulation of mental fatigue of operators during long-time work remains to be explored. The physiological calculation data in the construction scene has the characteristics of stream arrival, non-stability, obvious individual difference and the like. In the prior art, a great deal of research is carried out by using an offline processing paradigm, firstly downloading original data, respectively carrying out feature extraction and classification modeling, often relying on manual experience and priori rules, and outputting the data with non-negligible delay, so that closed-loop control of 'monitoring-intervention' is difficult to support in time. Meanwhile, the traditional batch learning method, such as SVM, KNN, ANN, has good performance under the offline condition, but needs to store historical data and periodically retrain in the online scene, and has high calculation and time delay cost. Low quality samples are also included in the data stream, and direct inclusion of updates further impairs the stability and extrapolation ability of the model. Aiming at the aging and self-adaptive bottleneck, online learning is more suitable for continuously generated data streams. The online sequence extreme learning machine (OS-ELM) can update parameters in a sequential closed mode when a new window arrives, and playback of historical samples or periodic retraining is not needed, so that the online sequence extreme learning machine is suitable for low-delay online judgment of an edge end. Disclosure of Invention The invention aims to overcome the defects in the prior art, and provides a real-time monitoring method and device for mental fatigue of operators facing geological operations, the invention realizes incremental learning by adopting an OS-ELM introducing forgetting factors on the basis of millimeter wave vital sign sensing and sliding window characterization, the recent samples are weighted more highly to adapt to the working condition drift and individual difference, real-time classification judgment and grading linkage early warning of mental fatigue states are realized, an online wind control closed loop is constructed for a management layer, the site safety and the production efficiency are improved, and continuous online feedback is realized. The