Agricultural machinery driving fatigue identification method based on driver heart rate change

Abstract

The invention discloses an agricultural machinery driving fatigue recognition method based on the change of the heart rate of a driver, and belongs to the technical field of agricultural machinery intelligent monitoring and man-machine work efficiency engineering crossing. The problems that an existing physiological signal detection method for driving fatigue is difficult in feature extraction and poor in universality due to individual differences are solved. Comprising the following steps: performing feature extraction based on a heart rate sequence to obtain a time-domain heart rate feature and a nonlinear heart rate feature; meanwhile, frequency domain heart rate characteristics are obtained through calculation based on the power spectrum density; processing is carried out to obtain a corrected time domain fatigue feature, a corrected frequency domain fatigue feature, a corrected nonlinear fatigue complexity feature and a corrected multi-time-mark driving fatigue cooperation feature; calculating multi-feature coupling fatigue energy, and calculating a fatigue accumulated response amount and a fatigue discrimination probability value; and calculating a fatigue strength index and a self-adaptive fatigue grading threshold value, and finally obtaining a fatigue grade judgment result. The method is used for identifying the fatigue state of the agricultural machinist.

Description

Technical Field

[0001] This invention relates to a method for identifying agricultural machinery driver fatigue based on changes in the operator's heart rate, belonging to the interdisciplinary technical field of intelligent monitoring of agricultural machinery and human-machine ergonomics engineering. Background Technology

[0002] With the continuous improvement of agricultural mechanization and intelligence, agricultural machinery has become the core equipment of modern agricultural production. Tractors, combine harvesters, and other agricultural machinery typically perform long hours, high loads, and continuous operations in the fields. During this process, operators not only endure high-intensity mechanical vibration, noise, and dust, but also face monotonous and repetitive operations, drastic changes in day and night lighting, and psychological stress, making them highly susceptible to the rapid accumulation of physiological and psychological fatigue. In a state of fatigue, operators react slowly, making more operational errors, which can lead to serious accidents such as machinery overturning or collisions, threatening personal safety and directly impacting operational efficiency and agricultural production benefits.

[0003] Currently, driver fatigue detection technologies mainly fall into three categories: methods based on vehicle operating parameters, methods based on driver facial visual features, and methods based on physiological signals. Vehicle state methods identify fatigue by monitoring steering wheel angle, lane departure, and speed fluctuations. However, in agricultural machinery operations characterized by low speeds, irregular trajectories, and significant terrain undulations, vehicle parameters are significantly affected by the work content and surface conditions, resulting in a high false detection rate. Facial image methods rely on cameras to capture behavioral features such as blinking frequency, pupil state, and yawning. However, uneven lighting, dust accumulation, and obstructions such as hats and masks worn by drivers are common in agricultural environments, severely impacting image quality and recognition stability. Physiological signal methods (such as electroencephalography, electrocardiography, and electrodermal conductance) directly reflect the state of the human autonomic nervous system and are highly objective. However, under the harsh conditions of strong vibrations and high noise levels in agricultural machinery, physiological signals are susceptible to motion artifacts and electromagnetic interference, leading to decreased signal quality and difficulties in feature extraction. More importantly, agricultural machinery operations have significant unique characteristics: First, the vibration intensity is high and the spectrum is complex, making it difficult for traditional filtering methods to effectively separate electrocardiogram signals from vibration noise; second, the operation cycle is long and the rhythm is inconsistent, with fatigue exhibiting a combination of short-term fluctuations and long-term accumulation, making it difficult for fatigue models on a single time scale to accurately characterize the situation; third, individual differences and environmental load have significant impacts, with different operators having different baseline heart rates and physical conditions, and different fields and different operation stages bringing different physiological loads, making universal fatigue thresholds less applicable.

[0004] Therefore, existing fatigue detection methods face technical bottlenecks in agricultural machinery operation scenarios, including insufficient adaptability, poor stability, and significant interference from individual and environmental factors. Summary of the Invention

[0005] To address the challenges of feature extraction in existing physiological signal detection methods for driver fatigue and the lack of universality due to individual differences, this invention provides a method for identifying agricultural machinery driver fatigue based on changes in the operator's heart rate.

[0006] The present invention provides a method for identifying agricultural machinery driver fatigue based on changes in operator heart rate, comprising:

[0007] The discrete electrocardiogram (ECG) signal sequence of the operator is acquired and preprocessed to obtain the heart rate sequence;

[0008] Feature extraction is performed based on heart rate sequences to obtain time-domain and nonlinear heart rate features; simultaneously, frequency-domain heart rate features are calculated based on power spectral density.

[0009] Based on the time-domain heart rate characteristics, frequency-domain heart rate characteristics, and nonlinear heart rate characteristics, the time-domain fatigue-sensitive heart rate characteristics, frequency-domain fatigue imbalance characteristics, and nonlinear fatigue complexity characteristics are calculated.

[0010] Multi-timescale driving fatigue collaborative features were calculated based on time-domain fatigue-sensitive heart rate features and frequency-domain fatigue imbalance features.

[0011] The time-domain fatigue-sensitive heart rate characteristics, frequency-domain fatigue imbalance characteristics, and nonlinear fatigue complexity characteristics are corrected to obtain the corrected time-domain fatigue characteristics, corrected frequency-domain fatigue characteristics, and corrected nonlinear fatigue complexity characteristics; then, combined with the multi-timescale driving fatigue collaborative characteristics, the corrected multi-timescale driving fatigue collaborative characteristics are calculated.

[0012] The fatigue energy coupled with multiple features is calculated based on the corrected time-domain fatigue features, the corrected frequency-domain fatigue features, and the corrected nonlinear fatigue complexity features. The fatigue cumulative response is also calculated. The fatigue discrimination probability value is obtained by combining the corrected multi-timescale driving fatigue collaborative features.

[0013] The fatigue intensity index is calculated based on the fatigue discrimination probability value and the modified multi-timescale driving fatigue collaborative features; and the adaptive fatigue grading threshold is calculated based on the modified multi-timescale driving fatigue collaborative features.

[0014] The fatigue level determination result is obtained by combining the fatigue strength index and the adaptive fatigue grading threshold.

[0015] According to the present invention, a method for identifying agricultural machinery driver fatigue based on driver heart rate changes represents the driver's discrete electrocardiogram signal sequence as follows: The preprocessing process includes:

[0016] The discrete electrocardiogram signal sequence of the operator is processed according to the selected frequency band. Frequency band constraints:

[0017] ,

[0018] In the formula The ECG signal after frequency band constraint This represents the number of ECG signals after frequency band constraints. The number of discrete ECG signals from the operator. It is a bandpass filter operator;

[0019] ECG signals after frequency band constraints Adaptive smoothing and differential calculations are performed to obtain the enhanced electrocardiogram signal. :

[0020] ,

[0021] In the formula For The length of the local smoothing window centered on the center. For the summation variable, ; Wavelet denoising operator;

[0022] ,

[0023] In the formula Based on smooth window length, To enhance the adjustment parameters;

[0024] Based on enhanced ECG signal Constructing cardiac saliency sequences :

[0025] ,

[0026] Significant sequences of heartbeats Peak values ​​were detected under amplitude threshold constraints and physiologically normal RR interval range constraints to obtain the heartbeat location sequence. :

[0027] ,

[0028] In the formula For the index of the kth heartbeat position, For peak detection operators, The significance level threshold. The minimum RR interval;

[0029] Calculate the corrected RR interval based on the positions of adjacent heartbeats. :

[0030] ,

[0031] In the formula This represents the median of the local RR interval;

[0032] Thus, the heart rate sequence is calculated. :

[0033] .

[0034] According to the present invention, a method for identifying agricultural machinery driver fatigue based on operator heart rate changes includes the following calculation methods for time-domain heart rate features, frequency-domain heart rate features, and nonlinear heart rate features:

[0035] Set the time window length for heart rate analysis Then within the time window length Heart rate subsequence within for:

[0036] , ;

[0037] Time-domain heart rate characteristics for:

[0038] ,

[0039] In the formula Time window length Inner strength is quantity, Time window length Internal mean heart rate;

[0040] Frequency domain heart rate characteristics for:

[0041] ,

[0042] In the formula This refers to the low-frequency band. For the entire frequency band, For frequency variables, Time window length Power spectral density of the inner rate sequence;

[0043] Nonlinear heart rate characteristics for:

[0044] .

[0045] According to the present invention, a method for identifying agricultural machinery driver fatigue based on operator heart rate changes includes the following calculation methods: time-domain fatigue-sensitive heart rate characteristics, frequency-domain fatigue imbalance characteristics, and nonlinear fatigue complexity characteristics.

[0046] Time-domain fatigue-sensitive heart rate characteristics for:

[0047] ,

[0048] In the formula For integration variables;

[0049] Frequency domain fatigue imbalance characteristics for:

[0050] ;

[0051] Nonlinear fatigue complexity characteristics for:

[0052] , ,

[0053] In the formula This represents the difference in heart rate between adjacent heartbeats.

[0054] According to the present invention, a method for identifying agricultural machinery driver fatigue based on operator heart rate changes, the multi-timescale driver fatigue collaborative feature is represented as follows: :

[0055] ,

[0056] In the formula Fatigue-sensitive heart rate characteristics at various time scales The average value, Fatigue imbalance characteristics in the frequency domain at various time scales The average value;

[0057] .

[0058] According to the present invention, a method for identifying agricultural machinery driver fatigue based on operator heart rate changes, the calculation methods for the corrected time-domain fatigue features, corrected frequency-domain fatigue features, and corrected nonlinear fatigue complexity features are as follows:

[0059] Corrected time-domain fatigue characteristics for:

[0060] ,

[0061] In the formula As an individual heart rate baseline mapping factor, This is a nonlinear correction factor for environmental load.

[0062] ,

[0063] ,

[0064] In the formula This represents the average heart rate of a machine operator during prolonged work. For reference heart rate values, This is an indicator of environmental load intensity.

[0065] Corrected frequency domain fatigue characteristics for:

[0066] ;

[0067] Corrected nonlinear fatigue complexity characteristics for:

[0068] .

[0069] The beneficial effects of this invention are as follows: The method of this invention comprehensively applies physiological signal processing, heart rate variability analysis, time-frequency domain feature extraction and intelligent discrimination algorithm to detect the physiological fatigue state of the operator in real time during agricultural machinery driving operations, and realize the identification and graded early warning of the fatigue level of the operator of agricultural machinery such as tractors and combine harvesters.

[0070] The method of this invention can integrate multi-timescale heart rate features, resist strong vibration interference, and has the ability to adapt to individuals and the environment, enabling continuous, accurate, robust monitoring and graded early warning of the operator's physiological state.

[0071] This invention proposes a multi-timescale adaptive fatigue recognition method based on heart rate changes to address the special working conditions of agricultural machinery operation, such as strong vibration, high noise, long working time, and significant individual and environmental differences. This method offers the following advantages:

[0072] 1) High-quality extraction of heart rate signals under strong vibration environment:

[0073] By employing a preprocessing strategy that combines frequency band constraints, adaptive smoothing, and wavelet denoising, the interference of low-frequency vibrations and high-frequency electromagnetic noise from agricultural machinery on electrocardiogram signals is effectively suppressed, improving the accuracy of heartbeat localization and the continuity of heart rate sequences, thus providing stable and reliable input for subsequent analysis.

[0074] 2) Multi-timescale heart rate characteristics comprehensively depict the fatigue evolution process:

[0075] To address the short-term fluctuations, medium-term persistence, and long-term accumulation characteristics of fatigue in agricultural machinery operations, three time windows—short, medium, and long—were designed. Time-domain, frequency-domain, and nonlinear heart rate variability features were extracted to construct a multi-scale fatigue characterization system. This avoids judgment bias caused by a single time scale and better reflects the actual fatigue development patterns in agricultural machinery operations.

[0076] 3) Adaptive adjustment to individual differences and environmental load:

[0077] An individual heart rate baseline mapping factor and an environmental load nonlinear correction factor are introduced to perform bidirectional calibration of fatigue characteristics. This mechanism can dynamically adjust feature weights and discrimination thresholds based on the operator's long-term heart rate statistics and real-time work intensity, significantly improving the model's adaptability and recognition consistency for different drivers and different work scenarios.

[0078] 4) Fatigue state classification and dynamic early warning:

[0079] Based on fatigue assessment, a refined classification from mild to severe fatigue is achieved by nonlinearly reconstructing the fatigue intensity index and combining it with a multi-threshold non-uniform grading strategy. Furthermore, the thresholds can be dynamically updated based on individual and environmental factors, making the early warning mechanism more personalized and real-time.

[0080] 5) Strong overall robustness and good practicality:

[0081] This invention is based entirely on physiological signal analysis and is not limited by visual conditions such as farmland lighting, dust, or obstruction. The method has a clear structure and high computational efficiency, supports deployment in embedded systems or mobile terminals, and is suitable for real-time online monitoring of various agricultural machinery such as tractors and harvesters. It helps to improve the safety, intelligence level, and management efficiency of agricultural machinery operations.

[0082] In summary, the method of this invention closely integrates with the special working conditions of agricultural machinery operations. Through a three-pronged technical approach of vibration suppression, multi-timescale analysis, and adaptive correction, it achieves accurate, stable, and personalized identification of operator fatigue, and has strong engineering application value and promotion prospects. Attached Figure Description

[0083] Figure 1 This is a flowchart of a method for identifying agricultural machinery driver fatigue based on changes in the driver's heart rate, as described in this invention. Detailed Implementation

[0084] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0085] Specific Implementation Method 1: Combination Figure 1 As shown, this invention provides a method for identifying agricultural machinery driver fatigue based on changes in the operator's heart rate, including:

[0086] The discrete electrocardiogram (ECG) signal sequence of the operator is acquired and preprocessed to obtain the heart rate sequence;

[0087] Feature extraction is performed based on heart rate sequences to obtain time-domain and nonlinear heart rate features; simultaneously, frequency-domain heart rate features are calculated based on power spectral density.

[0088] Based on the time-domain heart rate characteristics, frequency-domain heart rate characteristics, and nonlinear heart rate characteristics, the time-domain fatigue-sensitive heart rate characteristics, frequency-domain fatigue imbalance characteristics, and nonlinear fatigue complexity characteristics are calculated.

[0089] Multi-timescale driving fatigue collaborative features were calculated based on time-domain fatigue-sensitive heart rate features and frequency-domain fatigue imbalance features.

[0090] The time-domain fatigue-sensitive heart rate characteristics, frequency-domain fatigue imbalance characteristics, and nonlinear fatigue complexity characteristics are corrected to obtain the corrected time-domain fatigue characteristics, corrected frequency-domain fatigue characteristics, and corrected nonlinear fatigue complexity characteristics; then, combined with the multi-timescale driving fatigue collaborative characteristics, the corrected multi-timescale driving fatigue collaborative characteristics are calculated.

[0091] The fatigue energy coupled with multiple features is calculated based on the corrected time-domain fatigue features, the corrected frequency-domain fatigue features, and the corrected nonlinear fatigue complexity features. The fatigue cumulative response is also calculated. The fatigue discrimination probability value is obtained by combining the corrected multi-timescale driving fatigue collaborative features.

[0092] The fatigue intensity index is calculated based on the fatigue discrimination probability value and the modified multi-timescale driving fatigue collaborative features; and the adaptive fatigue grading threshold is calculated based on the modified multi-timescale driving fatigue collaborative features.

[0093] The fatigue level determination result is obtained by combining the fatigue strength index and the adaptive fatigue grading threshold.

[0094] During agricultural machinery operation, the operator's electrocardiogram (ECG) signal is collected. Under conditions of strong vibration and long-term continuous operation, the ECG signal is preprocessed and the heartbeat is located to output a stable and continuous heart rate sequence, providing a unified input for subsequent multi-timescale feature extraction.

[0095] Operator ECG signal acquisition and frequency band constraint: retaining the effective frequency band components of the ECG signal and reducing non-ECG interference: by using an ECG acquisition device installed on the operator's body, discrete ECG signal sequences are acquired. Frequency band constraint processing is applied to the ECG signals to address low-frequency drift and high-frequency mechanical vibration present in the agricultural machinery operation environment.

[0096] Furthermore, the discrete electrocardiogram signal sequence of the machine operator is represented as The preprocessing process includes:

[0097] The discrete electrocardiogram signal sequence of the operator is processed according to the selected frequency band. Frequency band constraints:

[0098] ,

[0099] In the formula The ECG signal after frequency band constraint This represents the number of ECG signals after frequency band constraints. The number of discrete ECG signals from the operator. It is a bandpass filter operator; the selected frequency band is 1-40Hz;

[0100] In response to the slow changes in vibration and noise of agricultural machinery, the electrocardiogram signal after frequency band constraint was analyzed. Adaptive smoothing and differential calculations are performed to obtain the enhanced electrocardiogram signal. :

[0101] ,

[0102] In the formula For The length of the local smoothing window centered on the local smoothing baseline is used to calculate the local smoothing baseline; The variable to be summed represents the index of the sampling points traversed within the window. Taking the nth sampling point as the center, take half the window length to the left / right; This is a wavelet denoising operator used to eliminate non-stationary noise;

[0103] Adaptive adjustment based on the rate of change of electrocardiogram signal:

[0104] ,

[0105] In the formula Based on smooth window length, To enhance the adjustment parameters;

[0106] Based on enhanced ECG signal Constructing cardiac saliency sequences To highlight the instantaneous energy peak corresponding to the R-wave:

[0107] ,

[0108] Significant sequences of heartbeats Peak values ​​were detected under amplitude threshold constraints and physiologically normal RR interval range constraints to obtain the heartbeat location sequence. :

[0109] ,

[0110] In the formula For the index of the kth heartbeat position, For peak detection operators, The significance level threshold can be set to [value]. A number of times the sequence mean (e.g., 1.5 times). The minimum RR interval is calculated based on the sampling frequency and the physiological limit heart rate.

[0111] The RR interval (heartbeat interval) is constructed and abnormalities are corrected to eliminate the impact of false positives and false negatives on heart rate calculation.

[0112] Calculate the corrected RR interval based on the positions of adjacent heartbeats. :

[0113] ,

[0114] In the formula This represents the median of the local RR interval;

[0115] Based on the corrected RR interval, the heart rate sequence was calculated. :

[0116] .

[0117] Using heart rate or inter-cardiac interval sequences as input, and considering the characteristics of short-term fluctuations, phased changes and long-term accumulation of agricultural machinery driver fatigue at different time scales, the temporal, frequency, and nonlinear features of heart rate variability (HRV) are extracted in multiple time windows to construct a multi-timescale HRV feature set.

[0118] Multi-timescale heart rate sequence segmentation:

[0119] Based on the time evolution characteristics of fatigue state, three analysis time scales are set, and the corresponding heart rate sequence is extracted at each time scale.

[0120] The calculation methods for time-domain heart rate characteristics, frequency-domain heart rate characteristics, and nonlinear heart rate characteristics are as follows:

[0121] Set the time window length for heart rate analysis Then within the time window length Heart rate subsequence within for:

[0122] , ;

[0123] Time-domain heart rate feature extraction: To reduce the impact of abnormal heartbeats on features, an exponentially suppressed heart rate deviation metric is introduced to construct time-domain heart rate change features.

[0124] Time-domain heart rate characteristics for:

[0125] ,

[0126] In the formula Time window length Inner strength is quantity, Time window length Internal mean heart rate; It is a natural exponential function;

[0127] Frequency domain heart rate feature extraction:

[0128] Within a time window T, spectral analysis is performed on the heart rate sequence to calculate the energy distribution of heart rate changes across different frequency components, constructing frequency domain heart rate characteristics. Within the heart rate sequence, spectral analysis is performed, and the corresponding heart rate power spectral density is calculated using discrete Fourier transform, Welch spectral estimation, or equivalent frequency domain analysis methods. .

[0129] Frequency domain heart rate characteristics for:

[0130] ,

[0131] In the formula This refers to the low-frequency band. For the entire frequency band, For frequency variables, Time window length Power spectral density of the inner rate sequence;

[0132] Nonlinear heart rate feature extraction: To characterize the complex changes in the heart rate regulation process, nonlinear mapping features based on adjacent heart rate changes are introduced.

[0133] Nonlinear heart rate characteristics for:

[0134] .

[0135] In the formula It is the hyperbolic tangent function. This is for variance calculation.

[0136] Heart rate feature extraction related to driver fatigue:

[0137] Using multi-timescale heart rate features as input, this study constructs heart rate features highly correlated with driving fatigue from three aspects: time domain, frequency domain, and nonlinearity, targeting the characteristics of increased heart rate fluctuations, spectral structure imbalance, and decreased nonlinear regulatory capacity caused by driving fatigue, and establishes a multi-timescale collaborative representation.

[0138] The calculation methods for time-domain fatigue-sensitive heart rate characteristics, frequency-domain fatigue imbalance characteristics, and nonlinear fatigue complexity characteristics are as follows:

[0139] Construction of fatigue-sensitive heart rate features in the time domain: By performing a cumulative nonlinear mapping on the time domain heart rate features, their sensitivity to fatigue state is enhanced.

[0140] Time-domain fatigue-sensitive heart rate characteristics for:

[0141] ,

[0142] In the formula It is an integral variable, and is a dummy continuous variable;

[0143] Construction of frequency domain fatigue imbalance features: For frequency domain heart rate features, a logarithmic-ratio form is introduced to construct fatigue-related frequency domain imbalance features;

[0144] Frequency domain fatigue imbalance characteristics for:

[0145] ;

[0146] In the formula It is the natural logarithm function.

[0147] Construction of nonlinear fatigue complexity features:

[0148] To characterize the decay characteristics of heart rate regulation complexity under fatigue, an exponential decay nonlinear feature is introduced;

[0149] Nonlinear fatigue complexity characteristics for:

[0150] , ,

[0151] In the formula The difference in heart rate between adjacent heartbeats. Let V be the variance of the window's heart rate sequence.

[0152] Construction of multi-timescale driving fatigue synergistic features: In order to comprehensively describe the accumulation and synergistic effect of fatigue at different time scales, multi-timescale driving fatigue synergistic features are constructed.

[0153] Multi-timescale driving fatigue collaborative features are represented as :

[0154] ,

[0155] In the formula Fatigue-sensitive heart rate characteristics at various time scales The average value, Fatigue imbalance characteristics in the frequency domain at various time scales The average value;

[0156] .

[0157] Individual and environmental adaptive correction:

[0158] Using driver fatigue-related heart rate characteristics as input, this paper constructs a multivariate nonlinear adaptive correction model from three levels—time domain, frequency domain, and nonlinearity—to achieve refined correction of fatigue characteristics, addressing the systematic shift in fatigue characteristics caused by individual differences in driver heart rate baselines and changes in agricultural machinery operating environment load.

[0159] The calculation methods for the corrected time-domain fatigue characteristics, corrected frequency-domain fatigue characteristics, and corrected nonlinear fatigue complexity characteristics are as follows:

[0160] Multivariate coupling adaptive correction of time-domain fatigue characteristics: The time-domain fatigue characteristics are corrected by ternary coupling with individual and environmental factors.

[0161] Corrected time-domain fatigue characteristics for:

[0162] ,

[0163] In the formula As an individual heart rate baseline mapping factor, This is a nonlinear correction factor for environmental load.

[0164] Construction of multi-scale mapping factors for individual heart rate baseline:

[0165] Based on the long-term heart rate statistics of machine operators, a log-hyperbolic joint mapping is introduced to construct an individual baseline factor. The logarithmic function compresses the range of individual differences, while the hyperbolic function limits the amplitude of the correction factor to prevent over-amplification.

[0166] During long-term operations, based on continuously output heart rate sequences, heart rate data from multiple work cycles are statistically analyzed to calculate the operator's average heart rate under long-term working conditions. .

[0167] Construction of Nonlinear Composite Characterization Factors for Environmental Load: Based on environmental signals collected during agricultural machinery operations, the statistical fluctuation intensity of these signals within a time window is calculated and used as an indicator of environmental load intensity. ;

[0168] ,

[0169] ,

[0170] In the formula This represents the average heart rate of a machine operator during prolonged work. For reference heart rate values, This is an indicator of environmental load intensity.

[0171] Saturated nonlinear correction of frequency domain fatigue characteristics: Logarithmic-exponential composite correction is introduced to address the sensitivity of frequency domain characteristics to environmental changes.

[0172] Corrected frequency domain fatigue characteristics for:

[0173] ;

[0174] Exponential consistency correction for nonlinear fatigue complexity features: an exponential consistency constraint is introduced for nonlinear fatigue features.

[0175] Corrected nonlinear fatigue complexity characteristics for:

[0176] .

[0177] Multivariate normalization correction of multi-timescale fatigue synergistic features: introduce multivariate normalization and smoothing constraints to multi-timescale fatigue synergistic features.

[0178] Corrected multi-timescale driving fatigue coordination features for:

[0179] .

[0180] In the formula This is for summing over different time windows.

[0181] Heart rate fatigue assessment: based on a set of fatigue characteristics that have been adaptively corrected for individual and environmental factors.

[0182] Furthermore, the fatigue energy is constructed by coupling multiple features: a comprehensive fatigue energy function is constructed by introducing coupling terms between features.

[0183] The calculation method for multi-feature coupled fatigue energy is as follows:

[0184] ,

[0185] In the formula For multi-feature coupled fatigue energy, This is the time-domain fatigue feature mapping quantity. This is the frequency domain fatigue characteristic mapping quantity. This is a nonlinear fatigue characteristic mapping quantity;

[0186] Multi-channel nonlinear mapping of fatigue characteristics:

[0187] ;

[0188] Construction of the asymmetric fatigue accumulation function: Introducing a cumulative nonlinear transformation to fatigue energy.

[0189] Fatigue cumulative response for:

[0190] ,

[0191] In the formula For integration variables;

[0192] Fatigue discrimination probability value for:

[0193] .

[0194] The value range is (0,1).

[0195] Fatigue grading and dynamic threshold: Given that the fatigue state has been determined, the degree of fatigue is graded and the threshold is dynamically adjusted based on individual and environmental factors.

[0196] Nonlinear reconstruction of fatigue strength: Constructing a high-resolution fatigue strength index;

[0197] Fatigue Strength Index for:

[0198] ,

[0199] In the formula To prevent extremely small positive numbers with a denominator of zero;

[0200] Dynamic fatigue threshold multi-factor adaptive update: Introducing individual and environmental factors to construct a dynamic threshold;

[0201] Adaptive fatigue grading threshold for:

[0202] ,

[0203] In the formula The basic fatigue discrimination threshold is used.

[0204] Multi-threshold fatigue level mapping function: Fatigue grading is completed through non-uniform threshold intervals.

[0205] Fatigue level assessment results for:

[0206] .

[0207] This implementation method achieves differentiated identification of fatigue during continuous operation of agricultural machinery by using multi-timescale fatigue evolution modeling and individual / environment coupling correction.

[0208] While the invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. A method for identifying agricultural machinery driver fatigue based on operator heart rate changes, characterized in that... include, The discrete electrocardiogram (ECG) signal sequence of the operator is acquired and preprocessed to obtain the heart rate sequence; Feature extraction is performed based on heart rate sequences to obtain time-domain and nonlinear heart rate features; simultaneously, frequency-domain heart rate features are calculated based on power spectral density. Based on the time-domain heart rate characteristics, frequency-domain heart rate characteristics, and nonlinear heart rate characteristics, the time-domain fatigue-sensitive heart rate characteristics, frequency-domain fatigue imbalance characteristics, and nonlinear fatigue complexity characteristics are calculated. Multi-timescale driving fatigue collaborative features were calculated based on time-domain fatigue-sensitive heart rate features and frequency-domain fatigue imbalance features. The time-domain fatigue-sensitive heart rate characteristics, frequency-domain fatigue imbalance characteristics, and nonlinear fatigue complexity characteristics are corrected to obtain the corrected time-domain fatigue characteristics, corrected frequency-domain fatigue characteristics, and corrected nonlinear fatigue complexity characteristics. Then, the corrected multi-timescale driving fatigue collaborative features are obtained by combining the multi-timescale driving fatigue collaborative features calculations. The fatigue energy coupled with multiple features is calculated based on the corrected time-domain fatigue features, the corrected frequency-domain fatigue features, and the corrected nonlinear fatigue complexity features. The fatigue cumulative response is also calculated. The fatigue discrimination probability value is obtained by combining the corrected multi-timescale driving fatigue collaborative features. The fatigue intensity index is calculated based on the fatigue discrimination probability value and the modified multi-timescale driving fatigue collaborative features; and the adaptive fatigue grading threshold is calculated based on the modified multi-timescale driving fatigue collaborative features. The fatigue level determination result is obtained by combining the fatigue strength index and the adaptive fatigue grading threshold.

2. The method for identifying agricultural machinery driver fatigue based on operator heart rate changes according to claim 1, characterized in that, The discrete electrocardiogram (ECG) signal sequence of the machine operator is represented as follows: The preprocessing process includes: The discrete electrocardiogram signal sequence of the operator is processed according to the selected frequency band. Frequency band constraints: ， In the formula The ECG signal after frequency band constraint This represents the number of ECG signals after frequency band constraints. The number of discrete electrocardiogram signals of the operator. It is a bandpass filter operator; ECG signals after frequency band constraints Adaptive smoothing and differential calculations are performed to obtain the enhanced electrocardiogram signal. : ， In the formula For The length of the local smoothing window centered on the center. For the summation variable, ; Wavelet denoising operator; ， In the formula Based on smooth window length, To enhance the adjustment parameters; Based on enhanced ECG signal Constructing cardiac saliency sequences : ， Significant sequences of heartbeats Peak values ​​were detected under amplitude threshold constraints and physiologically normal RR interval range constraints to obtain the heartbeat location sequence. : ， In the formula For the index of the kth heartbeat position, For peak detection operators, The significance level threshold. The minimum RR interval; Calculate the corrected RR interval based on the location of adjacent heartbeats. : ， In the formula This represents the median of the local RR interval; Thus, the heart rate sequence is calculated. : 。 3. The method for identifying agricultural machinery driver fatigue based on operator heart rate changes according to claim 2, characterized in that, The calculation methods for time-domain heart rate characteristics, frequency-domain heart rate characteristics, and nonlinear heart rate characteristics are as follows: Set the time window length for heart rate analysis Then within the time window length Heart rate subsequence within for: ， ； Time-domain heart rate characteristics for: ， In the formula Time window length Inner strength is quantity, Time window length Internal mean heart rate; Frequency domain heart rate characteristics for: ， In the formula This refers to the low-frequency band. For the entire frequency band, For frequency variables, Time window length Power spectral density of the inner rate sequence; Nonlinear heart rate characteristics for: 。 4. The method for identifying agricultural machinery driver fatigue based on operator heart rate changes according to claim 3, characterized in that, The calculation methods for time-domain fatigue-sensitive heart rate characteristics, frequency-domain fatigue imbalance characteristics, and nonlinear fatigue complexity characteristics are as follows: Time-domain fatigue-sensitive heart rate characteristics for: ， In the formula For integration variables; Frequency domain fatigue imbalance characteristics for: ； Nonlinear fatigue complexity characteristics for: ， ， In the formula This represents the difference in heart rate between adjacent heartbeats.

5. The method for identifying agricultural machinery driver fatigue based on operator heart rate changes according to claim 4, characterized in that, Multi-timescale driving fatigue collaborative features are represented as : ， In the formula Fatigue-sensitive heart rate characteristics at various time scales The average value, Fatigue imbalance characteristics in the frequency domain at various time scales The average value; 。 6. The method for identifying agricultural machinery driver fatigue based on operator heart rate changes according to claim 5, characterized in that, The calculation methods for the corrected time-domain fatigue characteristics, corrected frequency-domain fatigue characteristics, and corrected nonlinear fatigue complexity characteristics are as follows: Corrected time-domain fatigue characteristics for: ， In the formula As an individual heart rate baseline mapping factor, This is a nonlinear correction factor for environmental load. ， ， In the formula This represents the average heart rate of a machine operator during prolonged work. For reference heart rate values, This is an indicator of environmental load intensity. Corrected frequency domain fatigue characteristics for: ； Corrected nonlinear fatigue complexity characteristics for: 。 7. The method for identifying agricultural machinery driver fatigue based on operator heart rate changes according to claim 6, characterized in that, Corrected multi-timescale driving fatigue coordination features for: 。 8. The method for identifying agricultural machinery driver fatigue based on operator heart rate changes according to claim 7, characterized in that, The calculation method for multi-feature coupled fatigue energy is as follows: ， In the formula For multi-feature coupled fatigue energy, This is the time-domain fatigue feature mapping quantity. This is the frequency domain fatigue characteristic mapping quantity. This is a nonlinear fatigue characteristic mapping quantity; ； Fatigue cumulative response for: ， In the formula For integration variables; Fatigue discrimination probability value for: 。 9. A method for identifying agricultural machinery driver fatigue based on operator heart rate changes according to claim 8, characterized in that, Fatigue Strength Index for: ， In the formula To prevent extremely small positive numbers with a denominator of zero; Adaptive fatigue grading threshold for: ， In the formula The basic fatigue discrimination threshold is used.

10. The method for identifying agricultural machinery driver fatigue based on operator heart rate changes according to claim 9, characterized in that, Fatigue level assessment results for: 。

Images