Attention monitoring method and device and intelligent cushion

Abstract

The invention discloses an attention monitoring method. The method comprises the following steps: S1, acquiring synchronous voltage data and electroencephalogram signals of an intelligent cushion; s2, performing data preprocessing operation on the voltage data to obtain pressure matrix data in a two-dimensional form; s3, performing preprocessing operation on the electroencephalogram signals to obtain electroencephalogram data in a two-dimensional form; s4, performing attention state labeling on the electroencephalogram data by adopting an attention detection equation to obtain an attention state corresponding to the electroencephalogram data; s5, enabling the attention state as a sample label to correspond to a pressure matrix data sample, and carrying out the training, and obtaining an attention monitoring neural network model; and S6, collecting intelligent cushion pressure data of the target object in a preset state, inputting the intelligent cushion pressure data into the attention monitoring neural network model, and predicting the attention concentration condition of the target object by using the attention monitoring neural network model.

Description

Technical Field

[0001] The present invention relates to the technical field of computer monitoring, and particularly relates to an attention monitoring method, device and intelligent cushion. Background Art

[0002] Today, with the rapid development of current information technology, whether students' attention is concentrated directly affects the quality of classroom teaching. However, traditional ways of monitoring students' attention, such as teachers observing students' listening in class and communicating with students through conversations, can no longer meet the current needs. Although electroencephalogram instruments can provide accurate monitoring data, their high prices and inconvenient portability bring many troubles to attention monitoring. Therefore, we urgently need a simple, practical and low-cost monitoring solution. Summary of the Invention

[0003] The purpose of the present invention is to provide an attention monitoring method, device and intelligent cushion to solve the problems raised in the above background art.

[0004] To achieve the above purpose, the present invention provides an attention monitoring method, including the following steps:

[0005] S1. Collect synchronized voltage data of the intelligent cushion and electroencephalogram signals;

[0006] S2. Perform data preprocessing operations on the voltage data to obtain pressure matrix data in two-dimensional form;

[0007] S3. Perform preprocessing operations on the electroencephalogram signals to obtain electroencephalogram data in two-dimensional form;

[0008] S4. Use an attention detection equation to label the attention state of the electroencephalogram data to obtain the attention state corresponding to the electroencephalogram data;

[0009] S5. Use the attention state as a sample label corresponding to the pressure matrix data sample for training to obtain an attention monitoring neural network model;

[0010] S6. Collect the pressure data of the intelligent cushion in the preset state of the target object and input it into the attention monitoring neural network model, and use the attention monitoring neural network model to predict the attention concentration of the target object.

[0011] In a preferred embodiment, in step S5, the attention monitoring neural network model includes an LSTM layer, a Dropout layer and a fully connected layer. In step S6, collecting the pressure data of the intelligent cushion in the preset state of the target object and inputting it into the attention monitoring neural network model, and using the attention monitoring neural network model to predict the attention concentration of the target object includes the following steps:

[0012] Input the intelligent cushion pressure data in the preset state of the target object into the LSTM layer, extract the time series features in the pressure matrix data, and map these features to a higher-dimensional representation space;

[0013] Input the output data of the LSTM layer into the Dropout layer to randomly discard neurons and reduce overfitting;

[0014] Input the output data of the Dropout layer into the fully connected layer to map to the final output result.

[0015] In a preferred embodiment, in step S2, the data preprocessing operations on the voltage data include data standardization, null value filling, feature selection, and feature extraction. In step S3, the preprocessing operations on the electroencephalogram signals include removing useless channels, electrode positioning and extracting experimental segments, differentiating bad leads, artifact removal processing, filtering processing, and segmentation processing; the target object is a student, the preset state is the state of the student in class, the intelligent cushion pressure data is collected every 1.22 seconds on average, and the electroencephalogram signals are segmented, with each segment being 1.22 seconds long on average, and each segment being used as an independent electroencephalogram signal sample.

[0016] In a preferred embodiment, the attention detection equation is:

[0017]

[0018] where α, β, and θ are the electrical signal oscillations in different frequency ranges, E is the attention concentration index obtained through the detection equation, the larger E is, the more concentrated the attention is. By dividing the range of E, the attention concentration situation is labeled. The attention concentration situations are respectively set as very concentrated attention, generally concentrated attention, and inattentive attention. Among them, when E is less than 0.3, it is labeled as inattentive attention; when E is greater than 0.3 and less than 0.7, it is labeled as generally concentrated attention; when E is greater than 0.7, it is labeled as very concentrated attention.

[0019] In a preferred embodiment, the intelligent cushion pressure data in the preset state of the target object is input into the first LSTM layer. The number of units in the first LSTM layer is set to 128, the shape of the input data is specified, the activation function is set to the ReLU function for introducing non-linearity, and the complete output sequence is set to be returned. The output result of the first LSTM layer is input into the first Dropout layer, and the parameter is set to 0.4, indicating that the dropout ratio is 40%. The output data of the first Dropout layer is input into the second LSTM layer, and the number of units in the second LSTM layer is set to 32. A second Dropout layer is added after the second LSTM layer, and the parameter settings are the same as those of the first Dropout layer. A fully connected layer is added after the second Dropout layer, the output dimension is set to 3, and the softmax function is set as the activation function to convert the model output into a probability distribution. The cross-entropy is set as the loss function, and the Adam optimizer is used to optimize the model parameters.

[0020] The present invention also provides an attention monitoring device, including:

[0021] An acquisition module for acquiring synchronized intelligent cushion voltage data and electroencephalogram signals;

[0022] A voltage data preprocessing module for performing data preprocessing operations on the voltage data of the cushion to obtain pressure matrix data in two-dimensional form;

[0023] An electroencephalogram signal preprocessing module for performing preprocessing operations on the electroencephalogram signals to obtain electroencephalogram data in two-dimensional form;

[0024] An electroencephalogram signal attention state evaluation module for annotating the attention state of the two-dimensional electroencephalogram data using an attention detection equation to obtain the attention state corresponding to the electroencephalogram data;

[0025] A model training module for constructing a training data set with the attention state samples as labels and the corresponding pressure matrix data samples and training using a long short-term memory network model to obtain an attention monitoring neural network model;

[0026] An attention monitoring module for collecting the intelligent cushion pressure data in the preset state of the target object and inputting it into the attention monitoring neural network model, and using the attention monitoring neural network model to predict the attention concentration of the target object.

[0027] In a preferred embodiment, the attention monitoring neural network model includes an LSTM layer, a Dropout layer, and a fully connected layer. In step S6, the intelligent cushion pressure data in the preset state of the target object is collected and input into the attention monitoring neural network model, and the attention monitoring neural network model is used to predict the attention concentration of the target object, including the following steps:

[0028] Input the intelligent cushion pressure data in the preset state of the target object into the LSTM layer, extract the time series features in the pressure matrix data, and map these features to a higher-dimensional representation space;

[0029] Input the output data of the LSTM layer into the Dropout layer, randomly discard neurons to reduce overfitting;

[0030] Input the output data of the Dropout layer into the fully connected layer and map it to the final output result.

[0031] In a preferred embodiment, the attention detection equation is:

[0032]

[0033] Wherein, α, β, and θ are the electrical signal oscillations in different frequency ranges, E is the attention concentration index obtained through the detection equation, the larger E is, the more concentrated the attention is. By dividing the range of E, the attention concentration situation is marked. The attention concentration situations are respectively set as very concentrated attention, generally concentrated attention, and inattentive attention. Among them, when E is less than 0.3, it is marked as inattentive attention. When E is greater than 0.3 and less than 0.7, it is marked as generally concentrated attention. When E is greater than 0.7, it is marked as very concentrated attention.

[0034] The present invention also provides an intelligent cushion for attention monitoring, including an intelligent cushion body, an internal pressure sensor, and a data acquisition circuit. An internal pressure sensor is laid in the intelligent cushion body. The internal pressure sensor is a matrix-structured resistive thin-film pressure sensor. The matrix-structured resistive thin-film pressure sensor is connected to the data acquisition circuit. The data acquisition circuit includes a main control chip. The main control chip is an ESP32 chip. The ESP32 chip converts the analog signal into a digital signal through an analog-to-digital converter and transmits the ADC value to a computer through a built-in WiFi module. The computer converts the ADC value into a pressure value.

[0035] In a preferred embodiment, the data acquisition circuit further includes six 74HC4051 chips, four 74HC4067 chips and four FPC connectors. The row output and column output of the matrix structure resistive thin-film pressure sensor are respectively connected to two of the FPC connectors. The FPC connector connected to the row output is connected to the analog input / output ports of the six 74HC4051 chips through copper wires. The Z pins of the six 74HC4051 chips are respectively connected to 6 ADC channels of the ESP32 chip, and the control pins of the 74HC4051 chips are connected to 3 pins of the ESP32 chip to select channels. The FPC connector connected to the column output is connected to the analog input / output ports of the four 74HC4067 chips through copper wires, and the control pins of the 74HC4067 chips are connected to 4 pins of the ESP32 chip to select channels.

[0036] Compared with the prior art, the beneficial effects of the present invention are as follows: The attention monitoring method and device provided by the present invention can preprocess the collected seat pressure data, preprocess the collected EEG signals and perform attention annotation, correspond the attention state as a label to the pressure matrix data samples and perform training to obtain an attention monitoring neural network model, use the attention monitoring neural network model to predict the pressure data of the target object in a preset state, obtain the attention situation of the target object, and evaluate the attention concentration degree of the target object through the change of the pressure value of the target object, thereby improving the efficiency and convenience of attention monitoring.

[0037] The intelligent seat adopted by the present invention has obvious advantages compared with the existing EEG instruments. First of all, the cost of the intelligent seat is much lower than that of the EEG instrument, which can achieve more economical large-scale deployment and wider coverage of the student group. Secondly, the portability of the intelligent seat enables it to be easily carried to different places and environments for use, providing more flexibility and convenience for educational and teaching activities. Thirdly, the use of the intelligent seat is also safer and more reliable, without involving complex operations and safety risks similar to those of EEG instruments. Finally, students can hardly feel other additional pressures and interferences during the use of the intelligent seat, thus improving the comfort of students' use. Brief Description of the Drawings

[0038] Figure 1 is a schematic flow chart of the attention monitoring method provided by the embodiment of the present invention;

[0039] Figure 2 is a network schematic diagram of the attention monitoring neural network model provided by the embodiment of the present invention;

[0040] Figure 3 is a structural block diagram of the attention monitoring device provided by the embodiment of the present invention;

[0041] Figure 4 It is the circuit diagram for collecting the pressure of the intelligent cushion provided by the embodiment of the present invention; Specific embodiments

[0042] The technical solutions in the embodiments of the present invention will be clearly and completely described below. For the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0043] Embodiment 1

[0044] As Figure 1-2 shown, the monitoring method of the present invention includes the following steps:

[0045] Step S1: Collect the synchronized voltage data of the intelligent cushion and the electroencephalogram (EEG) signals. The present invention collects the electroencephalogram signals through an EEG acquisition instrument, and the EEG acquisition instrument is a head-mounted electroencephalograph. The head-mounted electroencephalograph can adopt the EPOC Flex 32-channel electrode cap from EMOTIV.

[0046] Step S2: Perform data preprocessing operations on the voltage data to obtain pressure matrix data in two-dimensional form. The data preprocessing operations on the voltage data include data standardization, null value filling, feature selection, and feature extraction.

[0047] Step S3: Perform preprocessing operations on the EEG signals to obtain EEG data in two-dimensional form. The preprocessing operations on the EEG signals include removing useless channels, electrode positioning and extracting experimental segments, differential bad leads, artifact removal processing, filtering processing, and segmentation processing. It can be understood that the EEG signals are easily interfered by the external environment, and it is necessary to perform artifact removal processing and filtering processing on the EEG signals to reduce the interference of external signals. For example, removing pseudo signals such as electrooculogram and electromyogram, and removing power frequency interference.

[0048] Furthermore, the target object is a student, the preset state is the state of the student in class, the intelligent cushion pressure data is collected once every 1.22 seconds on average, and the EEG signals are segmented, with each segment being 1.22 seconds on average, and each segment being an independent electroencephalogram signal sample.

[0049] Step S4: Use the attention detection equation to label the attention state of the EEG data to obtain the attention state corresponding to the EEG data.

[0050] Among them, the attention detection equation is:

[0051]

[0052] Among them, α, β, and θ are the oscillations of electrical signals in different frequency ranges, reflecting the activity levels of the brain in different states. Alpha waves are usually between 8 Hz and 13 Hz and typically appear during rest, relaxation, or sitting quietly, especially when the cerebral cortex is in a less active state; Beta waves are usually between 13 Hz and 30 Hz and typically appear during alertness, concentration, cognitive activities, and emotional arousal, being more prominent when the cerebral cortex is active; Theta waves are usually between 4 Hz and 8 Hz and typically appear during a state of deep relaxation of the body and brain; E is the attention concentration index obtained through the detection equation, and the larger E is, the more concentrated the attention is.

[0053] The attention state labels corresponding to the pressure matrix data in the present invention are divided into three categories: highly concentrated, moderately concentrated, and not concentrated. E is the attention concentration index obtained through the detection equation, and the larger E is, the more concentrated the attention is. By dividing the range of E, the attention concentration situation is marked. The attention concentration situations are respectively set as highly concentrated attention, moderately concentrated attention, and not concentrated attention. Among them, when E is less than 0.3, it is marked as not concentrated attention; when E is greater than 0.3 and less than 0.7, it is marked as moderately concentrated attention; when E is greater than 0.7, it is marked as highly concentrated attention.

[0054] Step S5: Correspond the attention state as a sample label with the pressure matrix data sample and perform training to obtain an attention monitoring neural network model. Figure 2 It is a network schematic diagram of the attention monitoring neural network model provided by an embodiment of the present invention. The attention monitoring neural network model includes an LSTM layer, a Dropout layer, and a fully connected layer.

[0055] Step S6: Collect the intelligent cushion pressure data of the target object in a preset state and input it into the attention monitoring neural network model, and use the attention monitoring neural network model to predict the attention concentration situation of the target object. It includes the following steps:

[0056] Input the intelligent cushion pressure data of the target object in a preset state into the LSTM layer, extract the time series features in the pressure matrix data, and map these features to a higher-dimensional representation space;

[0057] Input the data output by the LSTM layer into the Dropout layer, randomly discard neurons to reduce overfitting;

[0058] Input the data output by the Dropout layer into the fully connected layer and map it to the final output result.

[0059] Specifically, input the intelligent cushion pressure data in the preset state of the target object into the first LSTM layer. Set the number of units in the first LSTM layer to 128, specify the shape of the input data, set the activation function as the ReLU function to introduce non-linearity, and set to return the complete output sequence. Input the output result of the first LSTM layer into the first Dropout layer with the parameter set to 0.4, indicating a dropout ratio of 40% to prevent overfitting. Input the output data of the first Dropout layer into the second LSTM layer with the number of units in the second LSTM layer set to 32. Add a second Dropout layer after the second LSTM layer with the parameter settings the same as those of the first Dropout layer. Add a fully connected layer after the second Dropout layer, set the output dimension to 3, that is, the prediction results are three categories, namely very concentrated attention, moderately concentrated attention, and inattentive attention. Set to use the softmax function as the activation function to convert the model output into a probability distribution for classification. Set to use cross-entropy as the loss function and use the Adam optimizer to optimize the model parameters.

[0060] Embodiment 2

[0061] As Figure 3 shown, the present invention also provides an attention monitoring device, including: a collection module 301, a voltage data preprocessing module 302, an electroencephalogram signal preprocessing module 303, an electroencephalogram signal attention state evaluation module 304, a model training module 305, and an attention monitoring module 306.

[0062] Specifically, the collection module is used to collect synchronous intelligent cushion voltage data and electroencephalogram signals;

[0063] The voltage data preprocessing module is used to perform data preprocessing operations on the voltage data of the cushion to obtain pressure matrix data in two-dimensional form;

[0064] The electroencephalogram signal preprocessing module is used to perform preprocessing operations on the electroencephalogram signals to obtain electroencephalogram data in two-dimensional form;

[0065] The electroencephalogram signal attention state evaluation module is used to label the attention state of the two-dimensional electroencephalogram data using an attention detection equation to obtain the attention state corresponding to the electroencephalogram data;

[0066] The model training module is used to construct a training data set with the attention state samples as labels and the corresponding pressure matrix data samples and train using a long short-term memory network model to obtain an attention monitoring neural network model, where the attention monitoring neural network model includes an LSTM layer, a Dropout layer, and a fully connected layer.

[0067] An attention monitoring module is used to collect the intelligent cushion pressure data of a target object in a preset state and input it into an attention monitoring neural network model, and use the attention monitoring neural network model to predict the attention concentration of the target object.

[0068] Further, collecting the intelligent cushion pressure data of a target object in a preset state and inputting it into an attention monitoring neural network model, and using the attention monitoring neural network model to predict the attention concentration of the target object includes the following steps:

[0069] Input the intelligent cushion pressure data of the target object in the preset state into the LSTM layer, extract the time series features in the pressure matrix data, and map these features to a higher-dimensional representation space;

[0070] Input the output data of the LSTM layer into the Dropout layer, randomly discard neurons to reduce overfitting;

[0071] Input the output data of the Dropout layer into the fully connected layer to map to the final output result.

[0072] Further, the attention detection equation is:

[0073]

[0074] Among them, α, β, and θ are the electrical signal oscillations in different frequency ranges, E is the attention concentration index obtained through the detection equation, the larger E is, the more concentrated the attention is. By dividing the range of E, the attention concentration situation is marked. The attention concentration situations are respectively set as very concentrated attention, generally concentrated attention, and inattentive attention. Among them, when E is less than 0.3, it is marked as inattentive attention; when E is greater than 0.3 and less than 0.7, it is marked as generally concentrated attention; when E is greater than 0.7, it is marked as very concentrated attention.

[0075] Embodiment 3

[0076] The present invention also provides an intelligent cushion for attention monitoring. The intelligent cushion includes an intelligent cushion body, an internal pressure sensor, and a data acquisition circuit. The internal pressure sensor is laid in the intelligent cushion body. The internal pressure sensor is a 44×56 matrix structure resistive film pressure sensor. The sensor has a total of 2288 independent sensing units, and the resistance value of each sensing unit is 1000K ohms. The matrix structure resistive film pressure sensor is connected to the data acquisition circuit. The data acquisition circuit includes a main control chip. The main control chip is an ESP32 chip. The ESP32 chip converts the analog signal into a digital signal through an analog-to-digital converter and transmits the ADC value to the computer through the built-in WiFi module. The computer converts the ADC value into a pressure value.

[0077] Further, as Figure 4 shown, the data acquisition circuit further includes six 74HC4051 chips, four 74HC4067 chips, and four FPC connectors. The row output and column output of the matrix-structured resistive thin-film pressure sensor are respectively connected to two of the FPC connectors. The FPC connector connected to the row output is connected to the analog input / output ports Y0 to Y7 of the six 74HC4051 chips through copper wires. The Z pins of the six 74HC4051 chips are respectively connected to 6 ADC channels of the ESP32 chip. The control pins S0 to S2 of the 74HC4051 chips are connected to 3 pins of the ESP32 chip to select channels. The FPC connector connected to the column output is connected to the analog input / output ports C0 to C12 of the four 74HC4067 chips through copper wires. The control pins S0 to S3 of the chips are connected to 4 pins of the ESP32 chip to select channels.

[0078] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An attention monitoring method, characterized in that: It includes the following steps: S1. Collect synchronized voltage data of the intelligent cushion and electroencephalogram (EEG) signals; S2. Perform data preprocessing operations on the voltage data to obtain pressure matrix data in two-dimensional form; S3. Perform preprocessing operations on the EEG signals to obtain EEG data in two-dimensional form; S4. Use an attention detection equation to label the attention state of the EEG data and obtain the attention state corresponding to the EEG data; S5. Use the attention state as a sample label corresponding to the pressure matrix data sample for training to obtain an attention monitoring neural network model; S6. Collect the pressure data of the intelligent cushion in the preset state of the target object and input it into the attention monitoring neural network model, and use the attention monitoring neural network model to predict the attention concentration of the target object.

2. The attention monitoring method according to claim 1, wherein: In step S5, the attention monitoring neural network model includes an LSTM layer, a Dropout layer, and a fully connected layer. In step S6, collecting the pressure data of the intelligent cushion in the preset state of the target object and inputting it into the attention monitoring neural network model, and using the attention monitoring neural network model to predict the attention concentration of the target object includes the following steps: Input the pressure data of the intelligent cushion in the preset state of the target object into the LSTM layer, extract the time series features in the pressure matrix data, and map these features to a higher-dimensional representation space; Input the data output by the LSTM layer into the Dropout layer to randomly discard neurons and reduce overfitting; Input the data output by the Dropout layer into the fully connected layer to map to the final output result.

3. The attention monitoring method according to claim 2, wherein: In step S2, performing data preprocessing operations on the voltage data includes data standardization, null value filling, feature selection, and feature extraction. In step S3, performing preprocessing operations on the EEG signals includes removing useless channels, electrode positioning and extracting experimental segments, differential bad leads, artifact removal processing, filtering processing, and segmentation processing; the target object is a student, the preset state is the state of the student during class, the pressure data of the intelligent cushion is collected every 1.22 seconds on average, and the EEG signals are segmented, with each segment being 1.22 seconds on average and each segment being an independent EEG signal sample.

4. The attention monitoring method according to claim 1, wherein: The attention detection equation is: where α, β, and θ are electrical signal oscillations in different frequency ranges, E is the attention concentration index obtained through the detection equation, the larger E is, the more concentrated the attention is. By dividing the range of E, the attention concentration situation is labeled. The attention concentration situations are respectively set as very concentrated attention, generally concentrated attention, and inattentive attention. Among them, when E is less than 0.3, it is labeled as inattentive attention; when E is greater than 0.3 and less than 0.7, it is labeled as generally concentrated attention; when E is greater than 0.7, it is labeled as very concentrated attention.

5. The attention monitoring method according to claim 2, wherein: Input the intelligent cushion pressure data of the target object in the preset state into the first LSTM layer. Set the number of units in the first LSTM layer to 128, specify the shape of the input data, set the activation function to the ReLU function for introducing non-linearity, and set to return the complete output sequence. Input the output result of the first LSTM layer into the first Dropout layer with the parameter set to 0.4, indicating a dropout ratio of 40%. Input the output data of the first Dropout layer into the second LSTM layer and set the number of units in the second LSTM layer to 32. Add a second Dropout layer after the second LSTM layer with the parameter settings the same as those of the first Dropout layer. Add a fully connected layer after the second Dropout layer, set the output dimension to 3, and set to use the softmax function as the activation function to convert the model output into a probability distribution. Set to use cross-entropy as the loss function and use the Adam optimizer to optimize the model parameters.

6. An attention monitoring device, characterized in that: It includes: A collection module for collecting synchronized intelligent cushion voltage data and electroencephalogram signals; A voltage data preprocessing module for performing data preprocessing operations on the voltage data of the cushion to obtain pressure matrix data in two-dimensional form; An electroencephalogram signal preprocessing module for performing preprocessing operations on the electroencephalogram signals to obtain electroencephalogram data in two-dimensional form; An electroencephalogram signal evaluation attention state module for annotating the attention state of the two-dimensional electroencephalogram data using an attention detection equation to obtain the attention state corresponding to the electroencephalogram data; A model training module for constructing a training data set with the attention state samples as labels and the corresponding pressure matrix data samples and training using a long short-term memory network model to obtain an attention monitoring neural network model; An attention monitoring module for collecting the intelligent cushion pressure data of the target object in the preset state and inputting it into the attention monitoring neural network model, and using the attention monitoring neural network model to predict the attention concentration of the target object.

7. The attention monitoring device according to claim 6, characterized in that: The attention monitoring neural network model includes an LSTM layer, a Dropout layer, and a fully connected layer. In step S6, collecting the intelligent cushion pressure data of the target object in the preset state and inputting it into the attention monitoring neural network model, and using the attention monitoring neural network model to predict the attention concentration of the target object includes the following steps: Input the intelligent cushion pressure data of the target object in the preset state into the LSTM layer, extract the time series features in the pressure matrix data, and map these features to a higher-dimensional representation space; Input the output data of the LSTM layer into the Dropout layer to randomly discard neurons and reduce overfitting; Input the output data of the Dropout layer into the fully connected layer to map to the final output result.

8. The attention monitoring device according to claim 7, wherein: The attention detection equation is: Among them, α, β, and θ are electrical signal oscillations in different frequency ranges, E is the attention concentration index obtained through the detection equation, and the larger E is, the more concentrated the attention is. By dividing the range of E, the attention concentration situation is marked. The attention concentration situations are respectively set as very concentrated attention, generally concentrated attention, and inattentive attention. Among them, when E is less than 0.3, it is marked as inattentive attention; when E is greater than 0.3 and less than 0.7, it is marked as generally concentrated attention; when E is greater than 0.7, it is marked as very concentrated attention.

9. An intelligent cushion for attention monitoring, characterized in that: The intelligent cushion includes an intelligent cushion body, an internal pressure sensor, and a data acquisition circuit. The internal pressure sensor is laid in the intelligent cushion body. The internal pressure sensor is a matrix-structured resistive film pressure sensor. The matrix-structured resistive film pressure sensor is connected to the data acquisition circuit. The data acquisition circuit includes a main control chip, and the main control chip is an ESP32 chip. The ESP32 chip converts the analog signal into a digital signal through an analog-to-digital converter and transmits the ADC value to the computer through the built-in WiFi module. The computer converts the ADC value into a pressure value.

10. The intelligent cushion for attention monitoring according to claim 9, characterized in that: The data acquisition circuit further includes six 74HC4051 chips, four 74HC4067 chips, and four FPC connectors. The row output and column output of the matrix-structured resistive film pressure sensor are respectively connected to two of the FPC connectors; the FPC connector connected to the row output is connected to the analog input / output ports of six 74HC4051 chips through copper wires. The Z pins of the six 74HC4051 chips are respectively connected to 6 ADC channels of the ESP32 chip, and the control pins of the 74HC4051 chips are connected to 3 pins of the ESP32 chip to select channels; the FPC connector connected to the column output is connected to the analog input / output ports of four 74HC4067 chips through copper wires, and the control pins of the 74HC4067 chips are connected to 4 pins of the ESP32 chip to select channels.

Images