A student learning state recognition system based on skin electrical signals

Abstract

The application belongs to the technical field of intelligent teaching, and particularly relates to a student learning state recognition system based on skin electricity signals, which comprises a skin electricity collecting device, a single-chip microcomputer, a wireless transmission module, a screen display module and an upper computer, wherein the skin electricity collecting device, the wireless transmission module and the screen display module are connected with the single-chip microcomputer respectively; the skin electricity collecting device comprises a finger sleeve electrode and a skin electricity sensor module; the skin electricity sensor module is integrated with a low-power amplification circuit, an A / D conversion circuit, a voltage stabilizing circuit and a reference circuit; through accurate hardware collection of skin electricity physiological signals, software algorithm optimization processing and intelligent classification recognition, objective quantification, real-time monitoring and visual display of student learning concentration and learning state are realized, a teacher is assisted to quickly adjust a teaching scheme, and classroom teaching efficiency and teaching accuracy are improved.

Description

Technical Field

[0001] This invention belongs to the field of smart teaching technology, specifically referring to a student learning status recognition system based on electrodermal signals. Background Technology

[0002] In traditional classroom teaching scenarios, teachers primarily rely on subjective observations of students' external behaviors, including hand gestures, facial expressions, body language, and attentive listening. This method is highly dependent on teachers' experience, extremely subjective, and lacks quantitative and objective data support. It fails to accurately capture students' internal fluctuations in attention and emotional state, making it prone to judgment bias. Furthermore, it struggles to identify students who appear to be listening but are actually daydreaming, hindering teachers from adjusting their teaching pace and plans in a timely and precise manner, significantly reducing the relevance and overall efficiency of classroom instruction.

[0003] Existing physiological research shows that physiological signals such as skin electrical activity, heart rate, and respiration are directly regulated by the autonomic nervous system and are not subject to conscious control. They can objectively and truthfully reflect core physiological and psychological characteristics such as concentration, emotional fluctuations, and engagement, making them effective objective indicators for identifying learning states. Currently, some wearable physiological signal acquisition devices are used for emotion recognition and state monitoring in general scenarios. However, existing devices and systems have significant shortcomings: First, there is no dedicated monitoring system adapted to classroom teaching scenarios; the size and wearing method of general physiological monitoring devices are not suitable for students' long hours of classroom learning, writing, answering questions, and other routine operations. Second, existing systems mostly focus on simple signal acquisition, lacking a complete algorithmic process for noise reduction, feature optimization, and accurate classification and recognition specifically for skin electrical activity signals. Third, most devices only have data acquisition functions, lacking classroom-adapted visualization, data statistics, teaching warnings, and state analysis functions, and cannot directly assist teachers in classroom teaching control. Fourth, existing recognition algorithms suffer from feature redundancy, low filtering accuracy, and poor model recognition accuracy, making it difficult to achieve real-time and accurate quantitative identification of students' learning states.

[0004] In summary, the current field of classroom student learning status monitoring generally suffers from technical pain points such as large subjective judgment errors, lack of objective data support, poor real-time performance, low equipment compatibility, and insufficient system integration. Summary of the Invention

[0005] To address the problems of subjective judgment, low accuracy, poor real-time performance, poor device compatibility, and lack of systematic teaching support functions in existing technologies, this invention provides a student learning status recognition system based on electrodermal signals. Through precise hardware acquisition of electrodermal physiological signals, software algorithm optimization and intelligent classification recognition, it achieves objective quantification, real-time monitoring, and visualization of students' learning focus and status. This assists teachers in quickly adjusting teaching plans and improving classroom teaching efficiency and accuracy.

[0006] To achieve the above functions, the technical solution adopted by the present invention is as follows: A student learning status recognition system based on electrodermal signals, comprising an electrodermal signal acquisition device, a microcontroller, a wireless transmission module, a screen display module, and a host computer, wherein the electrodermal signal acquisition device, the wireless transmission module, and the screen display module are respectively connected to the microcontroller; the electrodermal signal acquisition device includes a finger electrode and an electrodermal sensor module, wherein the electrodermal sensor is connected to the microcontroller.

[0007] Furthermore, the finger sleeve electrode has a long cylindrical fully enclosed structure, an elastic tightening structure is provided at the finger joint of the finger sleeve electrode, a rough anti-slip texture is provided on the contact surface of the finger sleeve electrode and the finger, a tightening structure is provided at the end of the finger sleeve electrode, and a button slot is provided on the surface of the finger sleeve electrode for connecting the transmission wire.

[0008] Furthermore, the skin conductance sensor module integrates a low-power amplifier circuit, an A / D conversion circuit, a voltage regulator circuit, and a reference circuit. The low-power amplifier circuit uses an MC607 chip to construct a dual operational amplifier circuit, the A / D conversion circuit uses an MCP3201 chip, the voltage regulator circuit uses an ASM1117-3.3 chip to output a stable 3.3V voltage, and the reference circuit is used to provide a 4.096V reference voltage for the A / D conversion circuit.

[0009] Furthermore, the microcontroller is an STM32F401 microcontroller, which is electrically connected to the skin conductance sensor module, the screen display module, and the wireless transmission module, respectively. It is used to receive the raw skin conductance signal and perform signal preprocessing, feature extraction and filtering, and learning state recognition operations. The signal preprocessing process of the microcontroller includes: using a 0.02Hz~0.2Hz second-order Butterworth low-pass filter to remove high-frequency interference from the skin conductance signal; normalizing the signal to eliminate individual baseline differences; and extracting the effective component of the skin conductance response (SCR) through median filtering. The feature extraction and filtering... The selection process includes: extracting 35 electrodermal signal features from the time domain, frequency domain, and time-frequency domain using statistical methods, Fast Fourier Transform (FFT), and Discrete Wavelet Transform (DWT), respectively; optimizing and filtering the 35 features using a binary particle swarm optimization algorithm; and selecting 18 highly discriminative and representative features as model input features. The microcontroller has a built-in trained intelligent classification model that integrates three recognition algorithms: KNN, RF, and SVM. The SVM algorithm uses a one-to-one method to achieve three-class classification of student learning states, accurately determining the three learning states of focused, general, and distracted learning.

[0010] Furthermore, the wireless transmission module adopts a Zigbee module to establish a wireless data transmission link between the microcontroller and the host computer; the screen display module adopts an OLED screen to display the collected skin electrodermal signal waveform in real time, realizing real-time monitoring of signal quality.

[0011] Furthermore, the host computer has a built-in interactive interface and data processing program. The interactive interface includes a login page, a main page, a sub-page, and a data processing page, used to realize functions such as device status display, learning status statistics, data processing, teaching early warning, and parameter adjustment. The main page of the host computer is used to display device connection status, recognition count, student learning focus, and factors affecting status, and to visualize the distribution of the learning status of all students in the class through a bar chart. The sub-page of the host computer is used to display the real-time sampling point count and learning status prompts, and integrates reminder, warning, waveform display, data saving, and data processing function buttons. The data processing page of the host computer integrates multiple signal processing functions such as low-pass filtering, median filtering, FFT transformation, and DWT transformation, and supports secondary data optimization processing.

[0012] The beneficial effects achieved by adopting the above-described solution in this invention are as follows: 1. This invention relies on skin electrophysiological signals as the core identification indicator. The physiological signals are not affected by the student's subjective behavior and can truly reflect the student's internal attention and learning state. Combined with multi-level signal denoising, feature optimization and screening and multi-algorithm fusion classification model, it completely solves the problems of large subjective judgment bias and lack of objective data support in traditional manual observation, and greatly improves the accuracy and reliability of learning state identification. 2. This invention uses the high-performance STM32F401 microcontroller for high-speed computing and processing, combined with Zigbee low-latency wireless transmission technology, to achieve real-time linkage of the entire process of skin conductance signal acquisition, processing, recognition, uploading, and display. It can provide millisecond-level feedback on students' learning status, support teachers to grasp classroom learning in real time, quickly adjust teaching strategies, and significantly improve classroom teaching efficiency. 3. This invention adopts a special finger sleeve electrode structure. The long tube fully wraps around the electrode, with elastic tightness, anti-slip texture, and end tightening design. It fits snugly and securely, is comfortable and lightweight, and does not affect students' normal operations such as writing, answering questions, and raising their hands in class. It supports long-term continuous classroom monitoring and is suitable for classroom teaching scenarios of all grades. 4. The hardware integrates a complete set of dedicated circuits for low-power amplification, precise A / D conversion, voltage regulation, and reference calibration, effectively avoiding signal distortion caused by environmental interference and voltage fluctuations, and ensuring the stability and accuracy of skin conductance signal acquisition; the software realizes a closed-loop function of signal preprocessing, feature extraction, intelligent recognition, data visualization, and teaching assistance, making the system highly complete and practical. 5. This invention adopts a modular hardware and software design, with each functional module being independently controllable and easy to iterate and upgrade. It can be subsequently expanded to monitor multi-dimensional physiological signals such as heart rate and EEG, and the accuracy of the recognition algorithm model can be optimized. At the same time, the host computer's dedicated classroom interaction interface is adapted to teachers' teaching habits, and the classroom teaching effect can be quantified, providing reliable technical support for the monitoring of learning progress in the construction of smart classrooms. Attached Figure Description

[0013] Figure 1 This is a system module diagram of the present invention; Figure 2 This is a hardware device structure diagram of the present invention; Figure 3 This is a structural diagram of the finger sleeve electrode of the present invention; Figure 4 This is a circuit diagram for signal acquisition in this invention; Figure 5 This is a diagram of the voltage regulator circuit of the present invention; Figure 6 This is the reference circuit diagram for the present invention; Figure 7 This is the host computer interface of the present invention.

[0014] Among them, 10 is the contact surface between the finger electrode and the finger, 11 is the finger electrode and the finger joint, 12 is the surface of the finger electrode, and 13 is the end of the finger electrode. Detailed Implementation

[0015] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.

[0016] like Figure 1 As shown, the system logic architecture of the student learning state recognition system based on electrodermal signals of the present invention includes a data layer, an algorithm layer, and a model layer. In the data layer, the system acquires the electrodermal signals between the index and middle fingers of volunteers using an electrodermal acquisition device, serving as experimental data. In the algorithm layer, the microcontroller processes the data and extracts features. In the model layer, the system trains a learning state recognition model using the experimentally acquired electrodermal data, and deploys the trained model into the microcontroller, ultimately achieving the recognition and prediction of learning states.

[0017] like Figures 2 to 7 As shown, a student learning status recognition system based on electrodermal signals comprises five components: an electrodermal signal acquisition device, a microcontroller, a wireless transmission module, a screen display module, and a host computer. The specific implementation methods of each component are as follows: The skin conductance acquisition device includes finger-shaped electrodes and a skin conductance sensor module; the finger-shaped electrodes, such as... Figure 3 As shown, the overall structure adopts a long, fully enclosed cylindrical design. Specifically: an elastic tensioning structure is provided at the finger joint 11 of the finger sleeve electrode to further secure the electrode and prevent misalignment; a rough, non-slip texture is provided on the contact surface 10 between the finger sleeve electrode and the finger to increase friction and prevent accidental detachment during data collection; a tightening structure is provided at the end 13 of the finger sleeve electrode to further enhance its stability; and a button slot is provided on the surface 12 of the finger sleeve electrode for connecting the transmission wire.

[0018] The skin conductance sensor module includes a low-power amplifier circuit, an A / D conversion circuit, a voltage regulator circuit, and a reference circuit, as detailed below: like Figure 4 As shown, the low-power amplifier circuit uses a dual operational amplifier circuit based on the MC607 chip for signal isolation and tracking. Pins 1 and 2 of the MC607 are connected to form a follower, and are connected to pin 3 of J1 for signal output; pin 3 receives the amplified signal from pin 7 for isolation; pin 4 is grounded; and pin 6 inputs the signal to the A / D conversion circuit.

[0019] The A / D conversion circuit uses the MCP3201 chip. Pin 2 receives the output signal from pin 7 of the MC607 for analog-to-digital conversion; pin 1 receives a 4.096V reference voltage provided by the reference circuit; pins 3 and 4 are grounded; pins 5, 6, and 7 correspond to pins 4, 5, and 6 of J1, respectively, and are connected to the microcontroller via the SPI interface. Pin 1 of J1 is connected to VCC, and pin 2 is grounded.

[0020] like Figure 5 As shown, the voltage regulator circuit uses the ASM1117-3.3 chip. The input VCC is 5V, which is filtered by a 4.7μF capacitor C1 and regulated to 3.3V to power the microcontroller.

[0021] like Figure 6 As described above, pin 1 of the reference circuit is connected to VCC, capacitor C4 (0.1μF) is connected to ground, pin 3 is grounded, and pin 2 outputs a 4.096V reference voltage. This voltage is matched with the 12-bit A / D converter chip MCP3201 to ensure conversion accuracy.

[0022] The microcontroller used is an STM32F401 microcontroller, which performs noise reduction, smoothing, feature extraction and feature selection on the electrodermal signal on-chip.

[0023] Noise reduction: Considering that the effective frequency of the electrodermal signal is mainly concentrated between 0.02Hz and 0.2Hz, which is lower than most noise and interference signals, the system uses a second-order Butterworth low-pass filter to eliminate high-frequency interference, and the filter algorithm is written into the microcontroller in C language.

[0024] Normalization: To eliminate differences in baseline skin conductance levels between individuals, skin conductance signals were normalized.

[0025] SCR component extraction: Median filtering was used to extract the SCR component of the electrodermal signal.

[0026] Feature extraction: Using statistical formulas, Fast Fourier Transform (FFT) and Discrete Wavelet Transform (DWT), 35 features were extracted from the time domain, frequency domain and time-frequency domain respectively for model training.

[0027] Feature selection: The discrete binary particle swarm optimization algorithm is used to combine and optimize the features, and finally the 18 most representative features are selected for model training.

[0028] The model layer employs KNN, Random Forest (RF), and Support Vector Machine (SVM) classifiers to construct a learning state recognition model. When deployed to a microcontroller, the SVM module from the DSP library is used. To identify the two negative states affecting focus, the SVM algorithm needs to be processed into three categories. This invention uses a "one-to-one" method, meaning one SVM model is trained between any two classes of samples, requiring the deployment of three binary classification models. When predicting new samples, each model outputs a classification result, and the final category is determined by voting.

[0029] The wireless transmission module uses a Zigbee module, which connects to the microcontroller to enable wireless data transmission between the microcontroller and the host computer. Before use, the Zigbee module's address and channel are pre-configured using a serial port assistant to ensure normal communication with the host computer.

[0030] The screen display module uses an OLED screen, which, when connected to a microcontroller, can display the collected electrodermal data in real time, facilitating timely adjustments to the experimental scheme and ensuring the accuracy of signal acquisition.

[0031] The host computer interface is as follows Figure 7 As shown, it includes a login page, a data processing page, a main page, and a sub-page.

[0032] Login page: Includes an account and password input area, as well as "OK", "Change Password" and "Register" buttons.

[0033] The main page is divided into a basic information display area, a student status prediction distribution display area, and a basic operation button area. The basic information display area includes: hardware device connection status, system recognition count, learning focus level, and factors affecting learning focus. The student status distribution display area uses a bar chart to show the student's learning status after each prediction, making it easy to observe the status distribution trend over a period of time. The basic operation area contains three buttons: "System Run / Pause," "Exit System," and "Student Management."

[0034] Subpage: Accessed by clicking the "Student Management" button, this page consists of a prompt window, a function area, and a raw data waveform display area. The prompt window displays the number of sampling points for this prediction and the current prediction status. The function area contains five buttons: "Reminder," "Warning," "Waveform Display," "Data Save," and "Data Processing." Clicking the "Data Processing" button takes you to the data processing page, which includes four functional modules: low-pass filtering, median filtering, FFT, and DWT.

[0035] Workflow: Students wear finger electrodes on their index and middle fingers to collect raw analog signals of skin electrophysiology. These signals are amplified by the skin electrophysiology sensor module's amplifier circuit, regulated by the voltage regulator circuit, and calibrated by the reference circuit. The analog signals are then converted to digital signals by the A / D converter circuit and transmitted to the STM32F401 microcontroller via the SPI communication protocol. The microcontroller preprocesses the received digital signals, extracts and filters features, and inputs the optimized 18 features into an intelligent classification model to achieve real-time identification of the student's learning status. The identification results and raw signal data are uploaded to a host computer in real-time via a Zigbee wireless transmission module. The host computer visualizes the individual and overall learning status of students, signal waveforms, and data parameters through a multi-page interactive interface. Teachers can use this visualized data to monitor classroom learning in real time and adjust the teaching pace and plan accordingly.

[0036] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A student learning state recognition system based on electrodermal signals, characterized in that, It includes a skin conductance acquisition device, a microcontroller, a wireless transmission module, a screen display module, and a host computer; the skin conductance acquisition device, the wireless transmission module, and the screen display module are respectively connected to the microcontroller; the skin conductance acquisition device includes finger electrode and skin conductance sensor module, and the skin conductance sensor module is connected to the microcontroller.

2. The student learning state recognition system based on electrodermal signals according to claim 1, characterized in that, The finger sleeve electrode has a long, fully enclosed structure with an elastic tension structure at the finger joint, a rough, non-slip texture on the surface in contact with the finger, a tightening structure at the end, and a button slot on the surface for connecting wires.

3. The student learning state recognition system based on electrodermal signals according to claim 1, characterized in that, The skin conductance sensor module integrates a low-power amplifier circuit, an A / D conversion circuit, a voltage regulator circuit, and a reference circuit. The low-power amplifier circuit uses an MC607 chip to form a dual operational amplifier circuit. The A / D conversion circuit uses an MCP3201 chip. The voltage regulator circuit uses an ASM1117-3.3 chip to output a stable 3.3V voltage. The reference circuit is used to provide a 4.096V reference voltage for the A / D conversion circuit.

4. The student learning state recognition system based on electrodermal signals according to claim 1, characterized in that, The microcontroller used is an STM32F401 microcontroller, which receives the raw EKG signal and performs signal preprocessing, feature extraction and filtering, and learning state recognition operations. The signal preprocessing includes using a 0.02Hz~0.2Hz second-order Butterworth low-pass filter to remove high-frequency interference, normalizing the signal to eliminate individual baseline differences, and extracting the effective component of EKG response (SCR) through median filtering. The feature extraction and filtering includes extracting 35 EKG signal features from the time domain, frequency domain, and time-frequency domain using statistical methods, Fast Fourier Transform (FFT), and Discrete Wavelet Transform (DWT), and using a binary particle swarm optimization algorithm to select 18 representative features as model input.

5. A student learning state recognition system based on electrodermal signals according to claim 1, characterized in that, The microcontroller is wirelessly connected to the host computer.

6. A student learning state recognition system based on electrodermal signals according to claim 1, characterized in that, The wireless transmission module uses a Zigbee module, and the screen display module uses an OLED screen.

7. A student learning state recognition system based on electrodermal signals according to claim 1, characterized in that, The host computer has a built-in interactive interface and data processing program. The interactive interface includes a login page, a main page, a sub-page, and a data processing page.

Images