| Driver fatigue is defined as the phenomenon that physiological and psychological function decline and normal driving behavior was affected, due to monotonous driving or driving at night for a long time. Affected by the adverse factors, such as unreasonable design of man-machine interface, poor driving environment, long operating time, complex and laborious driving operation and so on, the combine harvester driver is susceptible to psychological and physiological fatigue, which makes driver’s function decline,the reaction ability and alert level reduce greatly. To study fatigue detection method for combine harvester driver, effectively identify the fatigue state of the driver, and timely warn driver at the necessary moment, which have important practical significance and research value for improving the operation efficiency, ensuring safety driving, reducing the accidents caused by fatigue driving, protecting the driver’s health status, and ensuring the grain harvest smoothly and losslessly.On the basis of deeply analyzing the detection methods of driver fatigue and the research status of driver fatigue in the field of agricultural machinery in China and abroad. Aiming at the existing deficiency and the further studied problems in the field of agricultural mach inery driver fatigue, taking the detection method of combine harvester dri ver fatigue as research object, two kinds of physiological signals of electrocardiograph(ECG) and surface electromyography(s EMG)as fatigue test tools, the detection methods of driver fatigue which suit for combine harvester operation environment and has high-performance as core task, the driver’s ECG and s EMG parameters were researched when driving combine harvester. Characteristics parameters of ECG and s EMG were extracted and analyzed, the optimal characteristics parameters were selected, the evaluation model of fatigue grade of combine harvester driver was built based on information fusion. The research results can objectively judge and evaluate the fatigue of combine harvester driver and provide references for the further research on real-time detection technology of the fatigue of agricultural machinery driver. The main research contents were as follows:(1) Design and implementation of experiment. The experimental scheme for fatigue detection of combine harvester driver was designed according to the purpose of the experiment, which includes personnel, determination period, weather conditions, ground feature, equipment, testing methods, testing procedure and so on. The monitoring experiment of the fatigue of combine harvester driver was carried out by RM-6240 C multi-channel physiological signal acquisition processing system, the ECG and s EMG data of neck muscles and right lumbar muscles of 10 male drivers were recorded for 120 minutes.(2) Subjective evaluation analysis of driver fatigue. Subjective questionnaires of driver fatigue which were obtained in the experiment were analyzed. The change curve of total fatigue degree and the change trends of physical, mental and sensory fatigue symptoms were obtained. The results showed that the total fatigue degree increased with the increase of the driving time, the change trend of fatigue degree was fast at first, and then slow down, at last fast again; most of subjects felt a little fatigue at 60 min, 90% of subjects felt more fatigue at 100 min, 90% of subjects felt very fatigue at 120 min; The fatigue degree of three kinds fatigue symptoms increased after driving, the change degree of physical fatigue was the most obvious, the secondary was sensory fatigue. The neck and lumbar were the most fatigue parts in all parts of the body. The most frequent symptom among sensory fatigue was eye fatigue, the eyelid was droopy, and the eye was dry.(3) Characteristic analysis of ECG. The time domain, frequency domain and nonlinear characteristics parameters of HRV were extracted in the process of driving. The curves of most characteristics parameters showed obvious linear trend, the values of MEAN, SDNN, LF, HF, LFnorm, HFnorm, LF/HF and Samp En before driving were significantly different from that after driving(P<0.05); MEAN, LFnorm, HFnorm and Samp En respectively performed well in the three types of indicators, the curves showed linear trend and lower fluctuation. Compared with the linear indexes, fluctuation range of nonlinear characteristic curves were small, the change trends were more stable, which can effectively reflect the process and degree of fatigue. The key characteristic parameters of HRV detection of driver fatigue degree were determined by the comparison of Pearson Correlation Coefficient with HRV characteristic parameters, which were SDNN, LFnorm, HFnorm and Samp En. The ECG signals in straight section and turn section were compared from the aspects of signal waveform, HR, HRV, difference test and so on. The average values of HR and LF/HF in turn section were larger than that of straight section in the early, middle and later driving periods, the average values of MEAN, LFnorm, Ap En and Samp En in turn section were smaller than that of straight section in the early, middle and later driving periods, and the values of six indexes in turn section were significantly different from that of straight section(P<0.05), which indicated that the working load and labor intensity in turn section were higher than that of straight section.(4) Characteristic analysis of s EMG. The time domain, frequency domain and nonlinear characteristics parameters of s EMG of neck muscles and right lumbar muscles were extracted in the process of driving. The change curves of all characteristics parameters of neck muscles and right lumbar muscles showed obvious linear trend with the extension of driving time, which were significantly different before and after driving(P<0.05); Compared with the nonlinear indexes, the change degree of linear indexes were more significant, the curves showed higher fluctuation, but the fluctuation ranges of nonlinear characteristic curves were small, the curves were more stable, which can effectively reflect the process and the degree of muscle fatigue. The joint analysis of s EMG spectrum and amplitude showed that the increase of muscle strength of the erector spinae muscle and middle scalene muscles were not significant in the process of driving, the degree of muscle fatigue gradually increased. The fatigue degree of right neck muscles was higher than that of left neck muscles by comparing the change degree of different indexes in both sides of the neck muscles. RMS, MF and C0 complexity were the key characteristic parameters of s EMG in lumbar muscles; i EMG, MF and Ap En of right neck muscles were the key characteristic parameters of s EMG in neck muscles.(5) Establishing evaluation model for driver fatigue grade. The key characteristic parameters of HRV and s EMG were fused together; the evaluation model of fatigue grade of combine harvester driver was established based on SVM. The input variables of the SVM model were the SDNN, LFnorm, HFnorm and Samp En of HRV, the RMS, MF and C0 complexity of lumbar muscles, the i EMG, MF and Ap En of right neck muscles; the driving fatigue was divided into three grades, which are alert(level 0), mild fatigue(level 1) and fatigue(level 2), the three fatigue grades were selected as output variables of the SVM model. By the classification inspection of test set, the recognition rates of the three fatigue grades were respective 89.66%, 83.33% and 90.48%; the overall classification accuracy rate was 87.5%; the recognition accuracy rate of fusion feature was higher than that of single feature, the recognition accuracy rate of ECG feature was higher than that of s EMG feature. The performances of evaluation model for driver fatigue grade were good, the classification accuracy was higher, the characteristic parameters fusion of ECG and s EMG can effectively recognize the fatigue grade of combine harvester driver. |
PDF Full Text Request