Human body fatigue detection method and device, medium and equipment

Abstract

The invention belongs to the technical field of human body detection, and particularly discloses a human body fatigue detection method and device, a medium and equipment, and the method comprises the steps: collecting the human body index information of a to-be-detected operator; preprocessing the human body index information of the to-be-tested operating personnel; a human body fatigue detection model is constructed and trained; and inputting the preprocessed human body index information of the to-be-detected operating personnel into a trained human body fatigue degree detection model so as to detect the fatigue degree grade of the to-be-detected operating personnel. According to the method, the fatigue degree of the operator can be quantitatively evaluated, so that the detection accuracy can be improved.

Description

Technical Field

[0001] The present application belongs to the technical field of human body detection, and specifically relates to a method, device, medium and equipment for detecting human body fatigue. Background Art

[0002] At present, traditional methods for detecting human fatigue, especially the fatigue state of workers, often rely on subjective judgment or simple physiological index monitoring, such as heart rate, blood pressure, etc. However, these methods rely too much on manual observation and empirical judgment, and are easily affected by the subjective factors of the observer, resulting in the results being not objective enough and difficult to truly reflect the fatigue state of the human body. Summary of the invention

[0003] In view of the problems existing in the prior art, the main purpose of this application is to provide a method, device, medium and equipment for detecting human fatigue, which can accurately detect the fatigue of operators.

[0004] To achieve the above objectives, this application provides the following technical solutions:

[0005] A method for detecting human fatigue comprises: collecting human index information of a worker to be tested; preprocessing the human index information of the worker to be tested; constructing a human fatigue detection model and training it; inputting the preprocessed human index information of the worker to be tested into the trained human fatigue detection model to detect the fatigue level of the worker to be tested.

[0006] Optionally, the human index information of the operator to be tested includes: blood pressure, heart rate, heart rate variability, blood oxygen saturation, grip strength, vital capacity, critical flicker fusion frequency, visual reaction time, end-tidal carbon dioxide and breath alcohol content.

[0007] Optionally, the human body index information of the worker to be tested is preprocessed, including: cleaning the human body index information; and standardizing the cleaned human body index information.

[0008] Optionally, the human fatigue detection model includes: an input layer for inputting the human index information of the worker to be tested; a feature extraction module for extracting features from the human index information; a feature fusion module for fusing the features extracted by the feature extraction module to obtain fused features; and an output module for feature mapping the fused features to output a fatigue level.

[0009] Optionally, the input layer includes: a first input layer for inputting time series data in the human body index information; a second input layer for inputting non-time series data in the human body index information;

[0010] Optionally, the human fatigue detection model is trained through the following steps: collecting index information of the human body at various fatigue levels, and establishing a human index information database after preprocessing the index information; dividing the human index information database into a training set, a validation set and a test set in proportion; setting training parameters, and training the human fatigue detection model through the training set. During the training process, the cross entropy loss function of the model is calculated, and the loss function is optimized using the Adam optimization algorithm until the loss function converges; verifying the model with the validation set, and during the verification process, the accuracy, recall rate and F1 score are used as evaluation indicators to evaluate the performance of the model. When each indicator reaches a preset value, the model is verified; otherwise, the training parameters are adjusted or the training samples are expanded to retrain the model until the model is verified; testing the model with the test set, and during the verification process, the accuracy, recall rate and F1 score are used as evaluation indicators to evaluate the performance of the model. When each indicator reaches a preset value, the model is verified; otherwise, the training parameters are adjusted or the training samples are expanded to retrain the model until the model is verified.

[0011] In addition, the present application also provides a human fatigue detection device, which includes: an acquisition module for collecting human index information of the workers to be tested; a preprocessing module for preprocessing the human index information of the workers to be tested; a model construction and training module for constructing a human fatigue detection model and training it; a detection module for inputting the preprocessed human index information of the workers to be tested into the trained human fatigue detection model to detect the fatigue level of the workers to be tested.

[0012] Optionally, the preprocessing module includes: a cleaning submodule, used for cleaning the human body indicator information; and a standardization submodule, used for standardizing the cleaned human body indicator information.

[0013] In addition, the present application also provides a storage medium, which includes instructions, and when the instructions are executed on a computer, the computer executes any of the methods described in the figure.

[0014] In addition, the present application also provides an electronic device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any of the methods described above when executing the program.

[0015] Compared with the existing technology, the present application can bring the following beneficial effects: by constructing and training a human fatigue detection model, the present application can realize real-time and accurate detection of the fatigue level of workers, effectively improve the objectivity and reliability of the detection, overcome the problems of strong subjectivity and limited data processing capabilities of traditional methods, thereby improving work efficiency and safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of a method for detecting human fatigue provided by an embodiment of the present application;

[0017] Figure 2 is a structural schematic diagram of a human fatigue detection model provided by another embodiment of the present application;

[0018] Figure 3 is a structural diagram of a first feature extraction module provided in another embodiment of the present application;

[0019] Figure 4 is a structural diagram of a second feature extraction module provided in another embodiment of the present application;

[0020] Figure 5 is a structural diagram of a feature fusion module provided in another embodiment of the present application;

[0021] Figure 6 is a schematic diagram of the structure of a residual module provided in another embodiment of the present application;

[0022] Figure 7 is a structural diagram of an output module provided in another embodiment of the present application;

[0023] Figure 8 is a structural schematic diagram of a human fatigue detection device provided by another embodiment of the present application;

[0024] Fig. 9 is a schematic diagram of the structure of a storage medium provided by another embodiment of the present application;

[0025] Fig.10 It is a schematic diagram of the structure of an electronic device provided in another embodiment of the present application. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0027] It should be noted that all directional indications in the embodiments of the present application (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0028] In this application, unless otherwise clearly specified and limited, the terms "connection", "fixation", etc. should be understood in a broad sense. For example, "fixation" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0029] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the meaning of "and / or" appearing in the full text includes three parallel schemes. Taking "A and / or B" as an example, it includes scheme A, or scheme B, or a scheme that satisfies both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0030] Figure 1 FIG. 1 is a flow chart of a method for detecting human fatigue according to an embodiment of the present application. The execution device of the method is an electronic device, such as a computer, a server, etc. Figure 1 As shown, the method includes:

[0031] S100: Collecting human body index information of the operator to be tested;

[0032] S200: Preprocessing the human body index information of the worker to be tested;

[0033] S300: Build a human fatigue detection model and train it;

[0034] S400: Inputting the pre-processed human index information of the worker to be tested into a trained human fatigue detection model to output the fatigue intensity level of the worker to be tested.

[0035] By constructing and training a human fatigue monitoring model, this application can output the specific fatigue level of the operator, realize the quantification of the operator's fatigue status assessment, and thus improve the accuracy and objectivity of the operator's fatigue detection.

[0036] In another exemplary embodiment, in step S100, the human index information of the operator to be tested includes: blood pressure, heart rate, heart rate variability, blood oxygen saturation, grip strength, vital capacity, critical flicker fusion frequency, visual reaction time, end-tidal carbon dioxide and breath alcohol content.

[0037] In another exemplary embodiment, in step S200, the human body index information of the worker to be tested is preprocessed, including:

[0038] S201: Cleaning the human body index information;

[0039] In this step, the human body index information is cleaned to first remove or correct erroneous data points, such as removing data that obviously exceeds the normal range of human physiological indicators; second, missing values ​​need to be processed, which can be done by filling (such as mean, median, mode filling), interpolation or deletion of records containing missing values.

[0040] S202: Standardizing the human body index information after cleaning;

[0041] In this step, after cleaning, data of different magnitudes need to be converted to a uniform magnitude to facilitate model training and comparison. Specifically, the data can be standardized using methods such as maximum and minimum standardization and Z-score standardization. Among them, for indicators that conform to normal distribution or approximate normal distribution (such as certain measurements of heart rate and blood pressure), the Z-score standardization method can be used to subtract the mean from each data point and divide it by the standard deviation to convert it to a standard normal distribution with a mean of 0 and a standard deviation of 1. For indicators with a relatively fixed range of values ​​(such as blood oxygen saturation), the maximum and minimum standardization can be used to linearly transform the data to the interval [0,1]. In addition, for some highly skewed data (such as some biochemical indicators), logarithmic transformation can be considered to make it closer to normal distribution.

[0042] In another exemplary embodiment, Figure 2 As shown, the human fatigue detection model includes:

[0043] The first input layer is used to input the time series data in the human body index information;

[0044] The second input layer is used to input non-time series data in the human index information;

[0045] A first feature extraction module, used for extracting features from the time series data input by the first input layer to obtain dynamic information that changes over time;

[0046] A second feature extraction module, used for extracting features from the non-time series data input by the second input layer to obtain static information reflecting human health;

[0047] A feature fusion module, used for fusing the features extracted by the first feature extraction module and the second feature extraction module to obtain a fused feature;

[0048] The output module is used to perform feature mapping on the fused features and output a fatigue level.

[0049] Specifically, this embodiment provides a detailed description of each layer and each module described above.

[0050] Since the indicators included in the human index information belong to different types of data (that is, the above 10 indicators can be divided into time series data and non-time series data, among which, for example, heart rate, blood pressure, heart rate variability and blood oxygen saturation belong to time series data, and grip strength, vital capacity, critical flicker fusion frequency, visual reaction time, end-tidal carbon dioxide, and breath alcohol content belong to non-time series data), in order to enable the model to capture useful information from different angles, thereby improving the accuracy of fatigue detection (if all data are directly mixed and input without distinguishing the data type and received and mixed by the same input layer, it may cause information loss or difficulty in effectively extracting features), therefore, different input layers are needed to classify and input the above indicator information.

[0051] The first feature extraction module includes a convolution layer, a normalization layer, a ReLU activation function, a bidirectional long short-term memory network (Bi_LSTM) and a self-attention mechanism connected in sequence. Among them, the convolution layer scans the time series data through the convolution kernel to capture the local correlation therein. The normalization layer is used to normalize the output after the convolution operation to maintain the stability of the model during the training process and prevent the problem of gradient disappearance or explosion. The ReLU activation function is responsible for introducing nonlinear factors, which can set all negative values ​​to zero and keep positive values ​​unchanged, thereby helping the model to learn more complex feature mappings. The bidirectional long short-term memory network (Bi-LSTM) uses forward processing and reverse processing to enable the model to take into account not only past information but also future information, so as to better understand the context of the entire sequence, and then capture the long-term dependencies in the time series data. The self-attention mechanism allows the model to focus on different parts of the sequence when processing time series data, which can enhance the model's ability to focus on important parts of the sequence, so that the model can better understand the key information of the input data. The first feature extraction module can effectively extract multiple types of features from the input time series data, including trend features, periodic features, fluctuation features, frequency domain features, dynamic features, etc., wherein the trend feature refers to the trend of change over time, such as the trend of blood pressure rising or falling; the periodic feature refers to the periodic pattern in the data, such as the rhythm of heartbeat or breathing; the fluctuation feature includes, for example, the fluctuation of heart rate variability; the frequency domain feature refers to the feature in the frequency domain, such as the energy distribution of a specific frequency component; the dynamic feature includes dynamic characteristics such as speed and acceleration that change over time. The above features together constitute a comprehensive representation of the time series data, providing rich information support for subsequent feature fusion.

[0052] The second feature extraction module includes a first fully connected layer, a second fully connected layer, a third fully connected layer, a normalization layer and a ReLU activation function connected in sequence. The first to third fully connected layers use residual connections, that is, the input of the second feature extraction module and the output of the first fully connected layer are fused and input to the second fully connected layer, the input of the second feature extraction module and the output of the second fully connected layer are fused and input to the third fully connected layer, and the input of the second feature extraction module and the output of the third fully connected layer are fused and input to the normalization layer. In the second feature extraction module, the first to third fully connected layers use residual connections, which can add "shortcuts" or "jump connections" between the layers of the network, so that information and gradients can be directly transmitted in these shortcuts. Specifically, the first to third fully connected layers can achieve the following technical effects through residual connections: First, as the depth of the model increases, it is easy to encounter the problem of gradient disappearance or gradient explosion during training. Residual connections add direct connections between layers, allowing gradients to be more easily propagated back along these connections during back propagation, thereby helping to maintain the stability of the gradient and making model training more feasible; secondly, residual connections enable the model to learn the "residual mapping", that is, the improved part of the current layer relative to the previous layer, rather than directly learning the mapping from input to output. This method makes the model easier to optimize because learning residual mapping is much simpler than directly learning the original mapping; thirdly, since residual connections can transfer gradients more effectively, the model can often converge to a better solution faster during training, thereby shortening the training time. On the contrary, if the fully connected layer does not use residual connections, some adverse effects will occur. For example, during the training process, the model may fall into a local minimum, thus failing to reach the global optimal solution, resulting in poor model performance; for example, in deep learning, as the depth of the network increases, the gradient may gradually disappear or become too large during the back propagation process. Without residual connections, the gradient problem in the deep network may be more serious, causing the training process to become unstable or even stagnant. It can be seen that not using residual connections will affect the training stability, convergence speed and final performance of the model, thereby reducing the effectiveness and reliability of the model in practical applications. Specifically, in the human fatigue detection process described in this application, by adopting residual connections, the second feature extraction module can effectively capture subtle changes in data such as heart rate, blood pressure, and heart rate variability, and can maintain good feature learning capabilities even in the case of a deeper network, thereby helping the model to more accurately identify whether the human body is in a state of fatigue.

[0053] In summary, the second feature extraction module uses residual connections, which can not only improve the training performance of deep neural networks, but also achieve more accurate data analysis and judgment in the application scenario of human fatigue detection.

[0054] The second feature extraction module can extract non-time series features including numerical features, contrast features and threshold features from the input non-time series data, wherein the numerical feature refers to extracting features directly from a single numerical value, such as the size of grip strength or the amount of vital capacity; the contrast feature refers to comparing the relative relationship between different indicators, such as the ratio between grip strength and vital capacity; the threshold feature refers to judging whether a certain indicator is within a normal range by setting a threshold, such as whether the blood oxygen saturation meets the standard.

[0055] The feature fusion module includes a multi-head self-attention mechanism, a linear layer, and an output layer. The multi-head self-attention mechanism first concatenates time series features and non-time series features, and then maps them to three parts: query, key, and value through a linear layer. Next, the similarity score between the query and the key is calculated, and the attention weight is obtained through the softmax function. These weights are multiplied by the value and then fused through another linear layer to obtain the final attention-weighted feature representation. In order to capture more levels of features, this process can be divided into multiple heads, each with its own query, key, and value matrix, and then the results of all heads are concatenated and combined through an additional linear layer, and finally output through the output layer.

[0056] Specifically, the process of the feature fusion module fusing the outputs of the first feature extraction module and the second feature extraction module is described as follows:

[0057] 1. Initialization: The time series feature vector F output by the first feature extraction module t and the non-time series feature vector F n Splice into fusion vector F concat .

[0058] 2. Fusion vector F concat They are projected into three different spaces: query, key, and value. The specific expressions are as follows:

[0059] Q=F concat ·W Q

[0060] K=F concat ·W K

[0061] V=F concat ·W V

[0062] Among them, W Q represents the weight matrix corresponding to the query, W Krepresents the weight matrix corresponding to the key, W V Represents the weight matrix corresponding to the values.

[0063] 3. Attention calculation: Calculate the similarity score between the query and the key, and obtain the attention weight through the softmax function, which is calculated based on the following formula:

[0064]

[0065] Among them, A represents the attention weight, d k represents the dimension of the key vector, Used for scaling to prevent the softmax function from saturating.

[0066] 4. Attention weighted summation: Use the attention weight A to weight the value V, that is:

[0067] H=AV

[0068] Among them, H represents the value of the weighted sum of the attention weight A and the value V.

[0069] 5. Multi-head expansion: The above process is extended to multiple heads, and each head independently calculates the attention weight A and weighted sum H, that is:

[0070] H i =AV i

[0071] Among them, i represents the i-th head.

[0072] 6. Fusion of multi-head output: The results of all heads are concatenated and combined through the linear layer and then output by the output layer, that is:

[0073] H multi-head =Concat(H1,H2,...,H n )·W O

[0074] Among them, W O Represents the weight matrix of the output layer.

[0075] The output module includes a fully connected layer and a Softmax activation function, wherein the fully connected layer is used to map the fusion features to the dimensions required for the classification task; the Softmax activation function is used to output the probability distribution of different fatigue levels. The specific process of the output module outputting the fatigue level is as follows:

[0076] First, the fusion vector F concat First, the feature is mapped to a vector of fixed length through a fully connected layer, which implements this mapping through a weight matrix W and a bias vector b:

[0077] Z=F concat ·W+b

[0078] Where Z represents the output vector of the fully connected layer.

[0079] Secondly, the output vector Z of the fully connected layer is transformed by the Softmax activation function, which converts a real vector into a probability distribution so that each element is non-negative and the sum of all elements is 1. Specifically, for each element z in the output vector Z i ,,Softmax activation function is expressed as:

[0080]

[0081] in, It represents the given fusion feature F concat The probability that the model predicts fatigue level i when .

[0082] Finally, according to the probability distribution output by the Softmax activation function, the fatigue level with the highest probability value is selected as the final prediction result.

[0083] A residual module is also provided after the feature fusion module, and the residual module includes two fully connected layers, and each fully connected layer is provided with a batch normalization layer Batch Normalization and a ReLU activation function. In addition, the output of the residual module needs to be added to the input to form a residual connection. By setting the residual module, the model can learn the difference (i.e., residual) between the input and the output, rather than directly learning the mapping from the input to the output. This difference learning is often simpler, thereby reducing the difficulty of model training and making it easier for the model to learn complex features. Adding a residual module after the feature fusion module means that after fusing different types of features, the model can further enhance feature extraction through the residual module. The batch normalization layer following each fully connected layer helps to speed up the training process, and the ReLU activation function can introduce nonlinear factors and enhance the expression ability of the model. In addition, by setting the residual module, the model can better learn the complex patterns in the input data, thereby improving the accuracy of fatigue state detection.

[0084] In another exemplary embodiment, in step S300, the human fatigue detection model is trained by the following steps:

[0085] S301: collecting index information of the human body at various fatigue levels, and establishing a human body index information database after preprocessing the index information;

[0086] S302: Divide the human index information database into a training set, a validation set, and a test set in proportion, for example, the division ratio may be 6:3:1;

[0087] S303: setting training parameters, setting the training batch size to 32, setting the number of training rounds to 100, setting the learning rate to 0.001, and training the human fatigue detection model through the training set. During the training process, calculating the cross entropy loss function of the model, and using the Adam optimization algorithm to optimize the loss function until the loss function converges;

[0088] S304: Validate the model using the validation set. During the validation process, use accuracy, recall and F1 score as evaluation indicators to evaluate the performance of the model. When all indicators reach preset values ​​(for example, accuracy reaches 85%, recall reaches 80%, and F1 score reaches 0.8), the model validation is passed; otherwise, adjust the training parameters (for example, adjust the number of training rounds to 150, adjust the learning rate to 0.005) or expand the training samples (for example, adjust the ratio of training set, validation set and test set to 7:2:1) to retrain the model until the model validation is passed;

[0089] S305: Use the test set to test the model. During the verification process, the accuracy, recall and F1 score are also used as evaluation indicators to evaluate the performance of the model. When each indicator reaches the preset value (for example, the accuracy reaches 85%, the recall reaches 80%, and the F1 score reaches 0.8), the model verification is passed; otherwise, adjust the training parameters (for example, adjust the number of training rounds to 200, adjust the learning rate to 0.05) or expand the training samples to re-train the model until the model verification is passed.

[0090] In another exemplary embodiment, the cross entropy loss function is expressed as:

[0091]

[0092] Where C represents the number of categories, i.e., the fatigue level; y ic represents an indicator variable. If the i-th sample belongs to the c-th class, then y ic is 1, otherwise it is 0; It represents the probability that the model predicts that the i-th sample belongs to the c-th class, and N represents the number of samples.

[0093] Compared with other loss functions, such as the square loss function, the cross entropy loss function given in this application is easier to calculate the gradient, especially when using the Softmax function as the output layer activation function, it can be directly combined with the probability distribution, simplifying the calculation process. In addition, the cross entropy loss function proposed in this application directly measures the difference between the probability distribution predicted by the model and the actual label. In multi-category classification tasks such as human body detection, the model outputs the probability estimate of each category, and the cross entropy loss function proposed in this application can effectively measure the gap between these estimates and the actual label, thereby providing an intuitive and effective loss signal. Furthermore, the cross entropy loss function proposed in this application is defined by logarithmic loss, which means that when the model predicts correctly (that is, the predicted probability is close to 1), the loss function value is very small, and when the model predicts incorrectly (that is, the predicted probability is far from 1), the loss function value will be very large. This feature enables the model to be less punished when correctly classified and more punished when misclassified, thereby guiding the model to better learn the classification boundaries. In summary, the cross entropy loss function proposed in this application helps to improve the efficiency of model training and accelerate the convergence of the model, while also ensuring the performance of the model in tasks such as human fatigue detection that require accurate classification.

[0094] Below, the present application takes a certain group of specific human body indicator information as an example to exemplify the method proposed in the present application.

[0095] This application collects a set of human index information, including:

[0096] Blood pressure: 135 / 88 mmHg

[0097] Heart rate: 78bpm

[0098] Heart rate variability: 35ms

[0099] Blood oxygen saturation: 96%

[0100] Grip strength: 42kg

[0101] Vital capacity: 4.2L

[0102] Critical flash fusion frequency: 42Hz

[0103] Visual response time: 230ms

[0104] End-tidal carbon dioxide: 42 mmHg

[0105] Breath alcohol content: 0mg / L

[0106] First, preprocess the above indicators and clean the data to check whether the data is within a reasonable range. For example, blood pressure 135 / 88 mmHg, heart rate 78 bpm, etc. are all within the normal range and do not need to be corrected. Then convert the data of each indicator to a unified magnitude. For example, blood pressure and heart rate can be standardized using Z-score, while blood oxygen saturation can be standardized to the [0,1] interval using the maximum and minimum values.

[0107] Secondly, the preprocessed indicators are input into the trained human fatigue detection model, and the model inputs the time series data (such as heart rate, blood pressure, heart rate variability and blood oxygen saturation) and non-time series data (such as grip strength, vital capacity, critical flash fusion frequency, visual reaction time, end-tidal carbon dioxide, breath alcohol content) in the preprocessed indicators respectively. Then, the dynamic information of the time series data is extracted through the first feature extraction module, and the static information of the non-time series data is extracted through the second feature extraction module. And the two types of information are further fused through the feature fusion module to form a fusion feature.

[0108] Finally, the probability distribution of the output module based on the fusion feature output is [0.1, 0.2, 0.5, 0.1, 0.1]. According to the fatigue level set in advance, the first probability indicates wakefulness, the second probability indicates mild fatigue, the third probability indicates moderate fatigue, the fourth probability indicates severe fatigue, and the fifth probability indicates extreme fatigue. In the above probability distribution, the probability of moderate fatigue is 0.5, which is the highest value. Therefore, it can be determined that the fatigue level corresponding to the above-collected indicator information is moderate fatigue.

[0109] In another exemplary embodiment, Figure 8 As shown, the present application also provides a human fatigue detection device, which includes: an acquisition module 100, used to collect human index information of the operator to be tested; a preprocessing module 200, used to preprocess the human index information of the operator to be tested; a model construction and training module 300, used to construct and train a human fatigue detection model; and a detection module 400, used to input the preprocessed human index information of the operator to be tested into the trained human fatigue detection model to detect the fatigue level of the operator to be tested.

[0110] Based on the above embodiments, Fig. 9 , for an explanation of the computer-readable storage medium of the exemplary embodiment of the present application, please refer to Fig. 9, the computer-readable storage medium shown is a CD 40, on which a computer program (i.e., a program product) is stored. When the computer program is executed by the processor, it will implement the steps described in the above method implementation, for example, collecting the human index information of the operator to be tested; preprocessing the human index information of the operator to be tested; constructing and training a human fatigue detection model; inputting the preprocessed human index information of the operator to be tested into the trained human fatigue detection model to output the fatigue level of the operator to be tested. The specific implementation method of each step will not be repeated here.

[0111] It should be noted that the computer-readable storage medium includes but is not limited to phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical or magnetic storage media, which are not listed here one by one.

[0112] Based on the above embodiments, the present application also provides an electronic device, as shown below: Fig.10 An electronic device for downloading a file according to an exemplary embodiment of the present application is described.

[0113] Fig.10 A block diagram of an exemplary electronic device 50 suitable for implementing the embodiments of the present application is shown, and the electronic device 50 may be a computer system or a cloud server. Fig.10 The electronic device 50 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0114] like Fig.10 As shown, the electronic device 50 includes but is not limited to: one or more processors or processing units 501, a system memory 502, and a bus 503 connecting different system components (including the system memory 502 and the processing unit 501).

[0115] The electronic device 50 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 50, including volatile and non-volatile media, removable and non-removable media.

[0116] The system memory 502 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 5021 and / or cache memory 5022. The electronic device 50 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the ROM 5023 may be used to read and write non-removable, non-volatile magnetic media ( Fig.10 is not shown in the Fig.10 As shown in FIG. 5 , a disk drive for reading and writing a removable non-volatile disk (e.g., a "floppy disk") and an optical disk drive for reading and writing a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) can be provided. In these cases, each drive can be connected to bus 503 via one or more data medium interfaces. System memory 502 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of each embodiment of the present application.

[0117] A program / utility 5025 having a set (at least one) of program modules 5024 may be stored, for example, in system memory 502, and such program modules 5024 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. Program modules 5024 generally perform the functions and / or methods of the embodiments described herein.

[0118] The electronic device 50 may also communicate with one or more external devices 504 (e.g., a keyboard, a pointing device, a display, etc.). Such communication may be performed via an input / output (I / O) interface 505. Furthermore, the electronic device 50 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 506. Fig.10 As shown, the network adapter 506 communicates with other modules (such as the processing unit 501, etc.) of the electronic device 50 via the bus 503. It should be understood that although Fig.10 Not shown, other hardware and / or software modules may be used in conjunction with the electronic device 50 .

[0119] The processing unit 501 executes various functional applications and data processing by running the program stored in the system memory 502, for example, collecting the human index information of the operator to be tested; preprocessing the human index information of the operator to be tested; constructing and training a human fatigue detection model; inputting the preprocessed human index information of the operator to be tested into the trained human fatigue detection model to output the fatigue level of the operator to be tested. The specific implementation method of each step is not repeated here. It should be noted that although several units / modules or sub-units / sub-modules of the file concurrent downloading device are mentioned in the detailed description above, this division is only exemplary and not mandatory. In fact, according to the implementation mode of the present application, the features and functions of two or more units / modules described above can be concretized in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into multiple units / modules to be concretized.

[0120] In the description of the present application, it should be noted that the terms "first", "second" and "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0121] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0122] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0123] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0124] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0125] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that can be executed by a processor. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a cloud server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0126] The above embodiments are only for illustrating the technical concept and features of the present application, and their purpose is to enable people familiar with the technology to understand the content of the present application and implement it accordingly, and they cannot be used to limit the protection scope of the present application. Any equivalent changes or modifications made according to the spirit of the present application should be included in the protection scope of the present application.

Claims

1. A method for detecting human fatigue, characterized in that: The method comprises: Collecting human index information of the workers to be tested; Preprocessing the human body index information of the operator to be tested; Build a human fatigue detection model and train it; The pre-processed human index information of the operator to be tested is input into a trained human fatigue detection model to detect the fatigue level of the operator to be tested.

2. The human fatigue detection method according to claim 1, characterized in that: The human index information of the operator to be tested includes: blood pressure, heart rate, heart rate variability, blood oxygen saturation, grip strength, vital capacity, critical flash fusion frequency, visual reaction time, end-tidal carbon dioxide and breath alcohol content.

3. The human fatigue detection method according to claim 1, characterized in that: Preprocessing the human body index information of the operator to be tested includes: Cleaning the human body index information; The human body index information after cleaning is standardized.

4. The human fatigue detection method according to claim 1, characterized in that: The human fatigue detection model comprises: The input layer is used to input the human body index information of the operator to be tested; A feature extraction module, used for extracting features from the human body index information; A feature fusion module, used for fusing the features extracted by the feature extraction module to obtain fused features; The output module is used to perform feature mapping on the fused features to output a fatigue level.

5. The method for detecting human fatigue according to claim 4, characterized in that: The input layer includes: The first input layer is used to input the time series data in the human body index information; The second input layer is used to input non-time series data in the human body index information.

6. The human fatigue detection method according to claim 1, characterized in that: The human fatigue detection model is trained by the following steps: Collecting index information of the human body at various fatigue levels, and establishing a human body index information database after preprocessing the index information; Dividing the human index information database into a training set, a validation set and a test set in proportion; Set training parameters and train the human fatigue detection model through the training set. During the training process, calculate the cross entropy loss function of the model and use the Adam optimization algorithm to optimize the loss function until the loss function converges. The model is verified using the validation set. During the verification process, accuracy, recall and F1 score are used as evaluation indicators to evaluate the performance of the model. When all indicators reach the preset values, the model is verified. Otherwise, the training parameters are adjusted or the training samples are expanded to retrain the model until the model is verified. The model is tested using the test set. During the verification process, accuracy, recall and F1 score are used as evaluation indicators to evaluate the model performance. When all indicators reach the preset values, the model is verified. Otherwise, the training parameters are adjusted or the training samples are expanded to retrain the model until the model is verified.

7. A human fatigue detection device, characterized in that: The device comprises: A collection module is used to collect the human body index information of the workers to be tested; A preprocessing module, used for preprocessing the human body index information of the operator to be tested; Model building and training module, used to build and train a human fatigue detection model; The detection module is used to input the pre-processed human index information of the operator to be tested into a trained human fatigue detection model to detect the fatigue level of the operator to be tested.

8. The human fatigue detection device according to claim 7, characterized in that: The preprocessing module comprises: A cleaning submodule, used for cleaning the human body index information; The standardization submodule is used to standardize the human body index information after cleaning.

9. A storage medium, characterized in that: The method comprises instructions, which, when executed on a computer, enable the computer to execute the method according to any one of claims 1 to 6.

10. An electronic device, characterized in that: The electronic device comprises: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method according to any one of claims 1 to 6 is implemented.

Images