Self-powered wearable electroencephalogram emotion recognizer and preparation and recognition method thereof

Abstract

The invention provides a self-powered wearable electroencephalogram emotion recognizer and a preparation and recognition method thereof, and belongs to the field of wearable electronic equipment, emotion calculation and man-machine interaction. The recognizer integrates a self-powered module, a sensing module and a control and processing module; the sensing module comprises a multi-channel EEG sensor array and is used for synchronously collecting EEG signals, the control and processing module is integrated through a flexible printed circuit board and is used for data processing and wireless transmission, and the method comprises the steps that spatial oversampling is conducted on EEG signals at a single position through the multi-channel EEG sensor array; the signal quality of each channel is evaluated in real time, and the optimal EEG signal is autonomously selected to actively reduce the interference of motion artifacts; and processing the optimal EEG signal through a deep learning process to extract EEG-based emotional features and generate an emotional classification result, thereby realizing stable, autonomous and intervention-free continuous emotional monitoring.

Description

Technical Field

[0001] This invention belongs to the fields of multimodal information processing, artificial intelligence and affective computing, and specifically relates to a self-powered wearable EEG emotion recognizer and its preparation and recognition method. Background Technology

[0002] Affective computing, an interdisciplinary field aiming to identify and interpret human emotions, has made significant progress in recent years thanks to advancements in sensor design, circuit integration, and algorithm optimization. By capturing physiological signals such as speech, facial expressions, and electroencephalograms (EEGs), affective computing can quantitatively and qualitatively acquire a wide range of human emotional information. Compared to other methods, emotion recognition based on physiological signals such as EEGs is more objective and less susceptible to manipulation by subjective consciousness. However, current emotion recognition technologies still face significant technical challenges.

[0003] First, most emotion recognition systems are limited by their power supply and portability. Existing systems generally rely on traditional batteries or wired connections for power, which not only limits the application of devices in long-term, non-invasive monitoring scenarios but also increases the size and weight of the devices, hindering user portability. The need for frequent charging or battery replacement makes continuous, uninterrupted emotion monitoring difficult to achieve.

[0004] Secondly, as a tool for assessing brain condition, EEG has many limitations in its traditional applications. Clinical-grade EEG devices are typically bulky and expensive, require conductive gel, and demand that the subject remain still, which greatly limits their effectiveness in daily life. Furthermore, the operation of these devices usually requires the assistance of professionals, creating operator dependence and potentially leading to discrepancies in results between different operators.

[0005] More critically, wearable EEG sensors are highly susceptible to losing target signals during user movement. For sensors placed on the forehead, facial expressions, speech, and head movements introduce severe motion artifacts and poor electrode contact, significantly impacting EEG signal perception. This is one of the core issues that wearable EEG systems need to address. Consequently, users are often required to remain still to ensure data quality, which contradicts real-world application scenarios.

[0006] While wearable EEG technology has made some progress in overcoming some of the aforementioned problems, existing technologies have failed to address these interconnected challenges: wearable EEG devices are significantly affected by motion artifacts, while the portability of devices using large batteries or wired connections is reduced. Summary of the Invention

[0007] To address the problems mentioned in the background art, this invention provides a self-powered wearable EEG emotion recognition device and its preparation and recognition method, thereby solving the following problems existing in the prior art: traditional batteries or wired connections limit the continuous working time of the device, increase its size and weight, and reduce portability; motion artifacts severely interfere with EEG signals, and body movements and facial expressions can lead to a decrease in data quality or even signal loss, limiting the effectiveness of the measurement; traditional EEG systems are complex to operate and rely on professional personnel, requiring the use of conductive gel and additional equipment, making them unsuitable for daily deployment.

[0008] To achieve the above objectives, the present invention provides the following technical solution: A self-powered wearable EEG emotion recognition device includes a self-powered module, a sensing module, and a control and processing module. The self-powered module includes a flexible organic photovoltaic device for converting ambient light energy into electrical energy to power the recognition device. The sensing module includes a multi-channel EEG sensor array for synchronously acquiring EEG signals. The control and processing module is integrated via a flexible printed circuit board for data processing and wireless transmission. The self-powered module, sensing module, and control and processing module are all encapsulated in a flexible biocompatible elastomer.

[0009] Preferably, the flexible organic photovoltaic device adopts an inverted structure, and its layers from bottom to top are: a flexible transparent cathode ITO-PET, a zinc oxide transport layer, a PM6:Y6 bulk heterojunction active layer, a molybdenum trioxide hole transport layer, and a silver anode.

[0010] Preferably, the multi-channel EEG sensor array comprises an array of eight dry electrodes configured on the forehead for signal acquisition, thereby enabling spatial oversampling of a single location.

[0011] Preferably, the control and processing module includes a microcontroller unit, a 24-bit analog-to-digital converter, and a wireless transmission module, all integrated on a 4-layer flexible printed circuit board.

[0012] A method for fabricating a self-powered wearable EEG emotion recognition device, the fabrication process comprising the following steps: A1: The ITO-PET substrate is cleaned by ultrasonic treatment in diluted detergent, acetone, deionized water and isopropanol in sequence, followed by 20 minutes of ultraviolet treatment. A2: Spin-coat a zinc oxide precursor solution onto an ITO-PET substrate to form a zinc oxide transport layer for electron transport, and anneal at 150°C for 15 minutes; A3: In a nitrogen-filled glove box, an active layer solution-formed heterojunction active layer is deposited on the zinc oxide transport layer using a spin coating process, and then annealed at 100°C for 10 minutes; A4: At a pressure less than 1e-5mbar, a 15nm thick molybdenum trioxide hole transport layer and a 100nm thick silver anode were sequentially deposited by thermal evaporation to complete the fabrication of flexible organic photovoltaic devices. A5: Integrate control circuit components, including microcontroller units and analog-to-digital converters, onto a single 4-layer flexible printed circuit board (fPCB). A6: The prepared flexible organic photovoltaic device, the fPCB, and the multi-channel EEG sensor array are integrated into a system and connected to an external reference electrode via leads; A7: The entire component, except for the external reference electrode, is encapsulated using medical-grade silicone elastomer to form the final patch-type identifier.

[0013] An emotion recognition method for a self-powered wearable EEG emotion recognition device includes the following steps: B1: Multi-channel EEG signals are simultaneously acquired via the aforementioned identifier; B2: The representations of the multiple EEG signals are input into a VGG16 model, which is trained to evaluate signal quality and thus autonomously select an optimal one-dimensional EEG time series signal; B3: The optimal one-dimensional EEG time series signal is converted into a two-dimensional time spectrum by continuous wavelet transform, and the time spectrum is input into a visual transformer model to extract high-dimensional emotion features and perform classification.

[0014] Preferably, the visual transformer model described in B3 uses its internal multi-head self-attention mechanism to capture long-range dependencies and global contextual information in the temporal spectrogram to identify emotional states.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. By integrating flexible organic photovoltaic devices, the system achieves self-powered operation, freeing it from the constraints of traditional batteries and wires, enabling wearable and portable operation, and supporting long-term continuous emotion monitoring.

[0016] 2. By employing a multi-channel EEG array for spatial oversampling and combining it with machine learning algorithms to autonomously select the optimal signal channel in real time, the interference of motion artifacts is avoided, significantly improving the robustness and data quality of EEG signals under user motion conditions, and overcoming the main defects of traditional wearable EEG.

[0017] 3. By integrating dry electrodes, flexible circuits, and self-powered modules into a single patch and automating the data processing flow, the dependence on conductive gel and professional operation is eliminated, allowing users to wear and use it themselves, and standardizing the data acquisition and interpretation process. Attached Figure Description

[0018] Figure 1 This is a system overview diagram of a self-powered wearable EEG emotion recognition device involved in this invention; Figure 2 This is a performance characterization diagram of the FOPV self-powered module involved in this invention; Figure 3 This invention relates to an autonomous EEG channel selection mechanism based on the VGG16 model. Detailed Implementation

[0019] To facilitate understanding of the technical content of this invention by those skilled in the art, the invention will be further described in detail below with reference to the accompanying drawings and specific examples. It should be understood that the specific examples described herein are merely illustrative and not intended to limit the scope of the invention.

[0020] like Figure 1 As shown in the diagram, a represents the device's wearability and layered structure, while b represents the system hardware and data flow frame. The self-powered wearable EEG emotion recognition device employs an integrated flexible patch design suitable for contact with human skin. It integrates a self-powered module, a sensing module, and a control and processing module. The self-powered module includes a flexible organic photovoltaic (FOPV) device to convert ambient light energy into electrical energy to power the system. The sensing module includes a multi-channel EEG sensor array for synchronously acquiring EEG signals. The control and processing module is integrated via a flexible printed circuit board (fPCB) for data processing and wireless transmission. All modules are encapsulated in a flexible, biocompatible elastomer (Ecoflex 00-30). The emotion recognition method first performs spatial oversampling of the EEG signal at a single location (FP1) using the multi-channel EEG sensor array, and uses the VGG16 model to evaluate the signal quality of each channel in real time, autonomously selecting the optimal EEG signal to actively reduce the interference of motion artifacts. Then, the optimal EEG signal is processed through a deep learning process to extract EEG-based emotion features and generate emotion classification results, thereby achieving stable, autonomous, and intervention-free continuous emotion monitoring.

[0021] The flexible organic photovoltaic device adopts an inverted structure, with its layers from bottom to top being: a flexible transparent cathode (ITO-PET), a zinc oxide (ZnO) transport layer, a PM6:Y6 bulk heterojunction (BHJ) active layer, a molybdenum trioxide (MoO3) transport layer, and a silver (Ag) anode. Figure 2 As shown, Figure 2 In this context, 'a' represents the chemical structural formulas of the donor PM6 and the acceptor Y6. Figure 2 In the diagram, b represents the structure of the FOPV device. Figure 2 The 'c' in the diagram is a comparison of the structures of BHJ and PHJ. Figure 2In the figure, d represents the JV curves of the BHJ and PHJ devices.

[0022] The multi-channel EEG sensor array includes an array of eight dry electrodes configured on the forehead (FP1 position) for signal acquisition, thereby enabling spatial oversampling at a single location.

[0023] The control and processing module includes a microcontroller unit (MCU), a 24-bit analog-to-digital converter (ADC), and a wireless transmission module, all integrated on a 4-layer flexible printed circuit board (fPCB).

[0024] A method for fabricating a self-powered wearable EEG emotion recognition device includes the following steps: 1. Dissolve 109 mg of zinc acetate and 29 μL of ethanolamine in 1 mL of 2-methoxyethanol, wherein the molar ratio of ethanolamine to zinc acetate is 7:25, and then stir vigorously at 600 rpm for 12 hours.

[0025] 2. Bulk heterojunction (BHJ) active layer solution: The donor polymer PM6 and the acceptor Y6 were dissolved together in chloroform with a total concentration of 16 mg / mL and a PM6:Y6 weight ratio of 1:1.2. The mixture was then stirred for 12 hours.

[0026] 3. Planar heterojunction (PHJ) active layer solution: The donor polymer PM6 and acceptor Y6 were dissolved in chloroform and then stirred for 12 hours.

[0027] 4. Clean the flexible ITO-PET substrate by sequentially ultrasonic treatment in diluted detergent, acetone, deionized water and isopropanol, followed by 20 minutes of ultraviolet treatment.

[0028] 5. Spin-coat a ZnO precursor solution onto the surface of an ITO-PET substrate at 5000 rpm for 40 seconds to prepare an electron transport layer, and then heat-anneal the resulting film at 150°C for 15 minutes.

[0029] 6. Deposition of active layer of BHJ structure device (preferred embodiment of the present invention): Select the substrate after the electron transport layer is prepared, and spin-coat PM6:Y6 BHJ active layer solution on the electron transport layer at a speed of 3000 rpm for 30 seconds in a glove box filled with nitrogen, and then perform thermal annealing treatment at 100°C for 10 minutes on the formed film.

[0030] 7. PHJ structure device active layer deposition (preferred embodiment of the present invention): Select the substrate after the electron transport layer is prepared, and spin-coat PM6 and Y6 BHJ active layer solutions on the electron transport layer at a speed of 3000 rpm for 30 seconds in a nitrogen-filled glove box. Then, perform thermal annealing treatment at 100°C for 10 minutes on the formed film.

[0031] 8. Under high vacuum conditions, a 15 nm thick MoO3 hole transport layer and a 100 nm thick Ag anode were sequentially deposited by thermal evaporation to complete the device fabrication; 9. Under standard test conditions, the photoelectric conversion efficiency (PCE) of the fabricated FOPV device with the BHJ structure was measured to be 12.39%, the open-circuit voltage (Voc) to be 0.74V, and the short-circuit current density (Jsc) to be 28.20mA / cm². 2 The fill factor (FF) is 59.64%.

[0032] As shown in Table 1, the performance parameters of the self-powered module, under standard test conditions, the photoelectric conversion efficiency (PCE) of the fabricated FOPV device with PHJ structure was measured to be 10.45%, the open-circuit voltage (Voc) to be 0.76V, and the short-circuit current density (Jsc) to be 23.36mA / cm². 2 The fill factor (FF) is 58.56%.

[0033] Table 1

[0034] It can be seen that the photoelectric conversion efficiency (PCE) and short-circuit current density (Jsc) of the FOPV device with BHJ structure (preferred embodiment of the present invention) are significantly better than those of the device with PHJ structure.

[0035] 10. Control Circuit Integration: The control circuit components, including the microcontroller unit (MCU, model ESP32-WROOM-32E) and the analog-to-digital converter (ADC, model ADS1298), are integrated onto a single 4-layer flexible printed circuit board (fPCB).

[0036] 11. System Integration: The fabricated flexible organic photovoltaic device with a BHJ structure, the fPCB, and a custom 8-channel EEG dry electrode array are integrated into the system. Simultaneously, an external reference electrode (designed as an ear clip for wearing on the earlobe) is connected via a thin lead.

[0037] 12. Encapsulation: The entire component, except for the electrodes, is encapsulated using medical-grade silicone elastomer (Ecoflex-0030) to form the final, soft, and wearable patch-type identifier.

[0038] An emotion recognition method for a self-powered wearable EEG emotion recognition device includes the following steps: (1) Multi-channel EEG signals are synchronously acquired through the identifier; (2) The representations of the multiple EEG signals are input into a neural network (VGG16) model, which is trained to evaluate signal quality and thus autonomously select an optimal EEG signal; (3) The optimal one-dimensional EEG time series signal is converted into a two-dimensional time spectrum using continuous wavelet transform (CWT), as shown below. Figure 3 As shown, the time-spectrum graph is input into a Vision Transformer (ViT) model to extract high-dimensional emotion features and perform classification.

[0039] Step (3) describes a visual transformer model that uses its internal multi-head self-attention mechanism to capture long-range dependencies and global contextual information in the temporal spectrogram in order to identify emotional states.

[0040] The present invention has been described through the above embodiments. However, it should be understood that the above embodiments are for illustrative purposes only and are not intended to limit the invention to the scope of the described embodiments. Furthermore, those skilled in the art will understand that the present invention is not limited to the above embodiments, and many more variations and modifications can be made based on the teachings of the present invention, all of which fall within the scope of protection claimed by the present invention. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A self-powered wearable EEG emotion recognition device, characterized in that, It includes a self-powered module, a sensing module, and a control and processing module; the self-powered module contains a flexible organic photovoltaic device for converting ambient light energy into electrical energy to power the identifier; the sensing module contains a multi-channel EEG sensor array for synchronously acquiring EEG signals; the control and processing module is integrated through a flexible printed circuit board for data processing and wireless transmission; the self-powered module, sensing module, and control and processing module are all encapsulated in a flexible biocompatible elastomer.

2. The self-powered wearable EEG emotion recognition device according to claim 1, characterized in that, The flexible organic photovoltaic device adopts an inverted structure, and its layers from bottom to top are: flexible transparent cathode ITO-PET, zinc oxide transport layer, PM6:Y6 bulk heterojunction active layer, molybdenum trioxide hole transport layer, and silver anode.

3. The self-powered wearable EEG emotion recognition device according to claim 1, characterized in that, The multi-channel EEG sensor array includes an array of eight dry electrodes configured on the forehead for signal acquisition, enabling spatial oversampling at a single location.

4. The self-powered wearable EEG emotion recognition device according to claim 1, characterized in that, The control and processing module includes a microcontroller unit, a 24-bit analog-to-digital converter, and a wireless transmission module, all integrated on a 4-layer flexible printed circuit board.

5. A method for fabricating a self-powered wearable EEG emotion recognition device, used to fabricate the recognition device as described in claims 1-4, the fabrication process comprising the following steps: A1: The ITO-PET substrate is cleaned by ultrasonic treatment in diluted detergent, acetone, deionized water and isopropanol in sequence, followed by 20 minutes of ultraviolet treatment. A2: Spin-coat a zinc oxide precursor solution onto an ITO-PET substrate to form a zinc oxide transport layer for electron transport, and anneal at 150°C for 15 minutes; A3: In a nitrogen-filled glove box, an active layer solution-formed heterojunction active layer is deposited on the zinc oxide transport layer using a spin coating process, and then annealed at 100°C for 10 minutes; A4: Flexible organic photovoltaic devices were fabricated by sequentially depositing a 15 nm thick molybdenum trioxide hole transport layer and a 100 nm thick silver anode through thermal evaporation at a temperature of less than 1e-5 mbar. A5: Integrate control circuit components, including microcontroller units and analog-to-digital converters, onto a single 4-layer flexible printed circuit board (fPCB). A6: The prepared flexible organic photovoltaic device, the fPCB, and the multi-channel EEG sensor array are integrated into a system and connected to an external reference electrode via leads; A7: The entire component, except for the external reference electrode, is encapsulated using medical-grade silicone elastomer to form the final patch-type identifier.

6. An emotion recognition method for a self-powered wearable EEG emotion recognition device, applied to the recognition device as described in claims 1-4, comprising the following steps: B1: Multi-channel EEG signals are simultaneously acquired via the aforementioned identifier; B2: The representations of the multiple EEG signals are input into a VGG16 model, which is trained to evaluate signal quality and thus autonomously select an optimal one-dimensional EEG time series signal; B3: The optimal one-dimensional EEG time series signal is converted into a two-dimensional time spectrum by continuous wavelet transform, and the time spectrum is input into a visual transformer model to extract high-dimensional emotion features and perform classification.

7. The emotion recognition method of a self-powered wearable EEG emotion recognition device according to claim 6, characterized in that, The visual transformer model described in B3 uses its internal multi-head self-attention mechanism to capture long-range dependencies and global contextual information in the temporal spectrogram to identify emotional states.

Images