Emotion prediction method and device based on electroencephalogram signals

Abstract

The invention provides an emotion prediction method and device based on electroencephalogram signals. The method comprises the steps that the electroencephalogram signals of a target user are acquired; performing time-frequency analysis on the electroencephalogram signal to obtain a target electroencephalogram characteristic signal, and extracting electroencephalogram characteristics of different frequency bands from the target electroencephalogram characteristic signal to obtain multi-dimensional electroencephalogram characteristics; calibrating the resting emotional state baseline according to the multi-dimensional electroencephalogram features to obtain a calibrated emotional state baseline, and extracting features corresponding to the calibrated emotional state baseline to obtain calibrated electroencephalogram features; inputting the calibrated electroencephalogram features into an emotion prediction model, performing emotion prediction by using the emotion prediction model, and outputting to obtain an emotion prediction result; and executing real-time emotion monitoring and feedback based on the emotion prediction result. According to the method and the device, the problem of low emotion recognition precision caused by neglecting data individual differences and non-stationary features in an emotion recognition process in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of electroencephalogram (EEG) signal analysis and processing, and in particular to an EEG signal-based emotion prediction method, an EEG signal-based emotion prediction device, a computer-readable storage medium, and an emotion prediction system. Background Art

[0002] In recent years, with the development of affective computing and neuroscience, emotion recognition technology based on electroencephalogram (EEG) signals has gradually become an important research direction in the fields of mental health monitoring and intelligent human-computer interaction. As physiological data reflecting brain activity, EEG signals contain rich information about emotional states and can show unique advantages in terms of non-invasiveness and real-time performance. However, due to the high individual differences, non-stationarity, and susceptibility to noise interference of EEG signals, traditional emotion recognition methods are difficult to achieve high-precision and personalized emotion prediction. In addition, existing emotion prediction systems often lack real-time emotion fluctuation monitoring and feedback mechanisms, making it difficult to meet the needs of dynamic emotion monitoring in complex application scenarios. Therefore, designing an efficient, accurate, and adaptive emotion prediction method has become a technical problem that needs to be solved in the field of emotion recognition. Summary of the invention

[0003] The main purpose of the present application is to provide an emotion prediction method based on EEG signals, an emotion prediction device based on EEG signals, a computer-readable storage medium and an emotion prediction system, so as to at least solve the problem in the prior art of ignoring individual differences and non-stationary characteristics of data in the emotion recognition process, resulting in low emotion recognition accuracy.

[0004] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a method for predicting emotions based on EEG signals is provided, comprising: obtaining an EEG signal of a target user; performing time-frequency analysis on the EEG signal to obtain a target EEG characteristic signal, and extracting EEG characteristics of different frequency bands from the target EEG characteristic signal to obtain a multi-dimensional EEG characteristic, wherein the target EEG characteristic signal is an EEG signal related to emotion; calibrating a resting emotional state baseline according to the multi-dimensional EEG characteristic to obtain a calibrated emotional state baseline, and extracting characteristics corresponding to the calibrated emotional state baseline to obtain a calibrated EEG characteristic, wherein the resting emotional state baseline is a baseline formed by characteristics of the brain activity state of the target user when the user is not subject to emotional fluctuations or external stimuli; inputting the calibrated EEG characteristics into an emotion prediction model to perform emotion prediction using the emotion prediction model, and outputting an emotion prediction result, wherein the emotion prediction model is a convergent model obtained by iteratively training a predetermined model using a neural network using multiple sets of training data, and each set of the training data includes sample EEG characteristics and the emotional state corresponding to the sample EEG characteristics; and performing real-time emotion monitoring and feedback based on the emotion prediction result.

[0005] Optionally, the EEG signal is subjected to time-frequency analysis to obtain a target EEG characteristic signal, including: decomposing the EEG signal using wavelet transform to generate a first EEG characteristic signal, wherein the first EEG characteristic signal is a multi-scale and multi-band global EEG signal; and decomposing the first EEG characteristic signal into multiple groups of intrinsic mode functions using an empirical mode decomposition method to obtain characteristic signals corresponding to emotions of different frequency components in each of the intrinsic mode functions to obtain the target EEG characteristic signal.

[0006] Optionally, EEG features of different frequency bands are extracted from the target EEG feature signal to obtain multi-dimensional EEG features, including: generating a time-frequency spectrum diagram according to the target EEG feature signal; and using a convolutional neural network to perform feature extraction on the time-frequency spectrum diagram to generate the multi-dimensional EEG features.

[0007] Optionally, before calibrating the multi-dimensional EEG features according to the resting emotional state baseline to obtain the calibrated EEG features, the method also includes: acquiring resting EEG signal data when the target user is in a resting state, the resting state being the brain activity state of the target user when not affected by emotional fluctuations or external stimuli; extracting the characteristic frequency band corresponding to the resting EEG signal data by analyzing the resting EEG signal data; and determining the resting emotional state baseline of the target user based on the characteristic frequency band.

[0008] Optionally, the resting emotional state baseline is calibrated according to the multi-dimensional EEG features to obtain a calibrated emotional state baseline, including: using a sliding window method to segment the multi-dimensional EEG features and extracting a first EEG feature from each sliding window; and comparing the first EEG feature with the resting emotional state baseline, and weighted updating the resting emotional state baseline to obtain the calibrated emotional state baseline.

[0009] Optionally, before inputting the calibrated EEG features into the emotion prediction model to use the emotion prediction model to perform emotion prediction and outputting the emotion prediction results, the method also includes: determining a model structure of the emotion prediction model, the model structure including an input layer, an intermediate convolutional layer and an output layer, the intermediate convolutional layer including a convolutional neural network layer, a long short-term memory network layer, a convolutional layer and a recurrent layer, the convolutional neural network layer is used to extract EEG feature signals related to emotions, the long short-term memory network layer is used to capture changes in the EEG feature signals in a time series, the convolutional layer is used to receive the EEG feature signals of the convolutional neural network layer and perform feature processing on the EEG feature signals to obtain emotion features, the recurrent layer receives changes in the EEG feature signals output by the long short-term memory network layer in a time series, and identifies changes in the emotion features in the time series; constructing an initial emotion prediction model according to the model structure; and iteratively training the initial emotion prediction model using a neural network using multiple groups of the training data to obtain the emotion prediction model.

[0010] Optionally, based on the emotion prediction result, real-time emotion monitoring and feedback are performed, including: generating a feedback signal based on the emotion prediction result in a closed-loop feedback system, and recording the history of changes in the target user's emotional state according to the feedback signal.

[0011] According to another aspect of the present application, an emotion prediction device based on EEG signals is provided, the device comprising: an acquisition unit, used to acquire the EEG signals of a target user; an analysis unit, used to perform time-frequency analysis on the EEG signals to obtain a target EEG characteristic signal, and extract EEG characteristics of different frequency bands from the target EEG characteristic signal to obtain a multi-dimensional EEG characteristic, wherein the target EEG characteristic signal is an EEG signal related to emotion; a calibration unit, used to calibrate a resting emotion state baseline according to the multi-dimensional EEG characteristic to obtain a calibrated emotion state baseline, and extract features corresponding to the calibrated emotion state baseline to obtain a calibrated EEG characteristic. The resting emotional state baseline is a baseline formed by the characteristics of the target user's brain activity state when the target user is not subject to emotional fluctuations or external stimuli; an input-output unit is used to input the calibrated EEG features into an emotion prediction model to use the emotion prediction model to perform emotion prediction and output an emotion prediction result, wherein the emotion prediction model is a convergent model obtained by iteratively training a predetermined model using a neural network using multiple sets of training data, and each set of training data includes sample EEG features and the emotional state corresponding to the sample EEG features; a monitoring and feedback unit is used to perform real-time emotion monitoring and feedback based on the emotion prediction result.

[0012] According to another aspect of the present application, a computer-readable storage medium is provided, which includes a stored program, wherein when the program is run, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned methods for predicting emotions based on electroencephalogram signals.

[0013] According to another aspect of the present application, an emotion prediction system is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing any one of the above-mentioned EEG signal-based emotion prediction methods.

[0014] By applying the technical solution of this application, emotion-related EEG features are extracted through multi-level time-frequency analysis, and the physiological differences of different users are addressed based on a personalized emotion baseline calibration method to achieve personalized adaptation of emotional states. A deep learning model, i.e., the above-mentioned emotion prediction model, is used to perform time series analysis on the extracted features to improve the accuracy of emotion prediction, and further a closed-loop feedback system is used to monitor emotion fluctuations in real time to provide support for emotion regulation and mental health. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a hardware structure block diagram of a mobile terminal for executing an emotion prediction method based on electroencephalogram signals according to an embodiment of the present invention;

[0016] Figure 2 is a flow chart of an emotion prediction method based on EEG signals according to an embodiment of the present application;

[0017] Figure 3 It is a structural block diagram of an emotion prediction device based on EEG signals according to an embodiment of the present application.

[0018] The above drawings include the following reference numerals:

[0019] 102, processor; 104, memory; 106, transmission device; 108, input and output devices. DETAILED DESCRIPTION

[0020] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0021] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings 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 should fall within the scope of protection of the present application.

[0022] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0023] As introduced in the background technology, the prior art ignores the individual differences and non-stationary characteristics of data in the emotion recognition process, resulting in low emotion recognition accuracy. To solve the above problems, the embodiments of the present application provide a method for executing emotion prediction based on EEG signals, an emotion prediction device based on EEG signals, a computer-readable storage medium, and an emotion prediction system.

[0024] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0025] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 1 is a hardware structure block diagram of a mobile terminal of an emotion prediction method based on EEG signals according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0026] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for predicting emotions based on EEG signals in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, the above method is implemented. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the above networks include but are not limited to the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or send data via a network. The above network specific examples may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0027] In this embodiment, a method for predicting emotions based on electroencephalogram signals is provided, which runs on a mobile terminal, a computer terminal or a similar computing device. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0028] Figure 2 is a flow chart of an emotion prediction method based on EEG signals according to an embodiment of the present application. Figure 2 As shown, the method comprises the following steps:

[0029] Step S201, obtaining the EEG signal of the target user.

[0030] Specifically, by analyzing the EEG signals of the target user, we can understand the patterns and characteristics of their brain activity and thus infer their emotional state. EEG signals can provide objective data about the user's emotions, which is more accurate and objective than traditional self-reporting methods. By analyzing EEG signals, we can identify brain activity patterns associated with different emotional states. For example, emotional states of pleasure, stress, and anxiety will show different characteristics in the brain, thereby more accurately predicting the emotional state of the target user.

[0031] Step S202, performing time-frequency analysis on the EEG signal to obtain a target EEG characteristic signal, and extracting EEG features of different frequency bands from the target EEG characteristic signal to obtain a multi-dimensional EEG feature, wherein the target EEG characteristic signal is an EEG signal related to emotion.

[0032] Specifically, by performing time-frequency analysis on EEG signals, extracting target EEG feature signals and further extracting EEG features in different frequency bands, multi-dimensional EEG features are obtained, thereby realizing the extraction and analysis of EEG signals related to emotions. This can effectively capture emotion-related features in EEG signals, and provide more accurate and comprehensive information for subsequent emotion predictions. By analyzing multi-dimensional EEG features, we can have a more comprehensive understanding of the emotional state of the subjects, thereby realizing accurate prediction of emotions. Through the analysis of EEG signals, we can also explore the relationship between emotions and brain activity.

[0033] Step S203, calibrate the resting emotional state baseline according to the multi-dimensional EEG features to obtain the calibrated emotional state baseline, and extract the features corresponding to the calibrated emotional state baseline to obtain the calibrated EEG features. The resting emotional state baseline is a baseline formed by the features of the target user's brain activity state when he is not affected by emotional fluctuations or external stimuli.

[0034] Specifically, based on the extracted multi-dimensional EEG features, individual emotional state baseline calibration is performed. By calibrating the resting emotional state baseline, the emotional characteristics of the target user can be more accurately extracted, thereby improving the accuracy and reliability of emotion prediction. The calibrated EEG features obtained by calibrating the emotional state baseline can better reflect the emotional state of the target user, thereby improving the accuracy and stability of emotion prediction. By comparing the difference between the calibrated emotional state baseline and the actual emotional state, the emotional state of the target user can be more accurately identified and predicted, providing a more effective basis for further emotional intervention and regulation.

[0035] Step S204, input the calibrated EEG features into the emotion prediction model to use the emotion prediction model to perform emotion prediction, and output the emotion prediction result. The emotion prediction model is a convergent model obtained by iteratively training a predetermined model using a neural network using multiple sets of training data. Each set of training data includes sample EEG features and the emotional state corresponding to the sample EEG features.

[0036] Specifically, the calibrated EEG features are input into the emotion prediction model for emotion prediction. The training data is iteratively trained through a neural network to obtain a convergence model, and an emotion prediction model based on a neural network is constructed to achieve emotion prediction. This can more accurately predict the emotional state of an individual, help individuals better understand and manage their emotions in practical applications, and improve the effect of emotion regulation.

[0037] Step S205: performing real-time emotion monitoring and feedback based on the emotion prediction result.

[0038] Specifically, by monitoring the EEG signals of the test subjects, the system can capture their emotional state in real time. Based on the prediction results, the system can provide corresponding feedback, such as reminding the test subjects to relax when their emotions fluctuate greatly, or providing encouragement and support when they are depressed. Real-time emotional monitoring and feedback can help the test subjects better understand their emotional changes and take timely measures to regulate their emotions, thereby improving their emotional management ability and mental health level, helping the test subjects to better understand and control their emotions, and improve their quality of life and work efficiency.

[0039] In this embodiment, emotion-related EEG features are extracted through multi-level time-frequency analysis, and the physiological differences of different users are addressed based on a personalized emotion baseline calibration method to achieve personalized adaptation of emotional states. A deep learning model, i.e., the above-mentioned emotion prediction model, is used to perform time series analysis on the extracted features to improve the accuracy of emotion prediction, and further a closed-loop feedback system is used to monitor emotion fluctuations in real time to provide support for emotion regulation and mental health.

[0040] Specifically, the target user's EEG signals can be obtained through electroencephalogram (EEG), which is a technology that records the brain's electrical activity by placing electrodes on the scalp. Since EEG signals are affected by many factors, such as electrode position, hair, muscle activity, etc., professional processing and correction are required during data collection and processing to ensure accurate emotion prediction results.

[0041] Specifically, time-frequency analysis is a commonly used signal processing technology used to analyze the changing characteristics of signals in time and frequency. It can convert the signal from the time domain to the frequency domain by performing Fourier transform on the EEG signal, thereby obtaining the energy distribution of the signal at different frequencies, which helps to understand the frequency characteristics of brain activity under different emotional states. For example, specific frequency components may be enhanced under anxiety. It can also be used to detect event-related potentials (ERP) in EEG signals, which is a type of brain electrical activity related to specific stimulus events. By analyzing the time-frequency characteristics of ERP, the brain's response patterns to stimuli under different emotional states can be captured more accurately, thereby achieving the prediction and classification of emotions.

[0042] In the specific implementation process, the EEG signal is subjected to time-frequency analysis to obtain the target EEG characteristic signal, which can be achieved through the following steps: the EEG signal is decomposed by wavelet transform to generate a first EEG characteristic signal, which is a multi-scale and multi-band global EEG signal; the first EEG characteristic signal is decomposed into multiple groups of intrinsic mode functions by empirical mode decomposition method to obtain the characteristic signals corresponding to the emotions of different frequency components in each intrinsic mode function, and the target EEG characteristic signal is obtained.

[0043] In this scheme, the target EEG characteristic signals are obtained through time-frequency analysis, including global EEG signals of multiple scales and multiple frequency bands and characteristic signals corresponding to emotions of different frequency components. By decomposing and processing EEG signals, emotion-related features in brain activity can be captured more accurately, thereby predicting emotions, providing more accurate and reliable data support for emotion recognition and emotion monitoring, and helping to further study the relationship between emotions and EEG signals.

[0044] Specifically, the EEG signal data is first decomposed using wavelet transform to generate multi-scale and multi-band EEG feature signals (i.e., the first EEG feature signal), thereby extracting the global feature signal, and the EEG signal (i.e., the first EEG feature signal) is decomposed into a group of intrinsic mode functions IMF (i.e., multiple groups of intrinsic mode functions) using the empirical mode decomposition (EMD) method to obtain the emotion corresponding features of different frequency components in each IMF, that is, to extract each local feature signal, obtain EEG features of different frequency bands, and thus obtain the target EEG feature signal.

[0045] It should be noted that the above-mentioned wavelet transform is a signal processing technology that can decompose signals into frequency components of different scales. The wavelet transform is used to process the EEG signals to extract features related to emotions, that is, the collected EEG signals are preprocessed, such as removing noise, filtering and other operations, and the wavelet transform is used to decompose the EEG signals into components of different frequencies. Different wavelet basis functions can be selected to achieve different decomposition effects, and the energy or other characteristics of each frequency component are calculated to reflect the activity level of the EEG signals in different frequency ranges.

[0046] It should be noted that the above-mentioned empirical mode decomposition (EMD) is a signal processing method that can decompose non-stationary and nonlinear signals into multiple intrinsic mode functions (IMFs). The EMD method is used to decompose the EEG signal into multiple IMFs, and the spectral characteristics and energy distribution of each IMF are analyzed to obtain emotion-related information in the EEG signal.

[0047] Specifically, the EEG signal is first decomposed by EMD to obtain multiple IMF components, and then each IMF is spectrally analyzed to calculate its spectral characteristics, such as frequency distribution and frequency intensity. At the same time, the energy distribution of each IMF can also be calculated to understand the signal energy size in different frequency ranges. Finally, by combining the spectrum and energy information of all IMFs, an emotion prediction model can be established to predict the emotional state of the subject by analyzing the IMF characteristics in the EEG signal.

[0048] In the specific implementation process, EEG features of different frequency bands are extracted from the target EEG feature signal to obtain multi-dimensional EEG features. This can be achieved through the following steps: generating a time-frequency spectrum diagram based on the target EEG feature signal; using a convolutional neural network to extract features from the time-frequency spectrum diagram to generate multi-dimensional EEG features.

[0049] In this scheme, by extracting EEG features of different frequency bands, generating time-frequency spectrograms, and using convolutional neural networks to extract features from the time-frequency spectrograms, multi-dimensional EEG features are obtained, which can more comprehensively capture the information in EEG signals and improve the accuracy and reliability of emotion prediction. Using multi-dimensional EEG features for emotion prediction can more accurately analyze the emotional state of an individual and help understand the emotion regulation mechanism and the occurrence mechanism of emotional disorders. At the same time, through the application of deep learning algorithms, a large amount of EEG signal data can be effectively processed to improve the efficiency and accuracy of emotion prediction.

[0050] Specifically, a time-frequency spectrum graph is generated for the EEG signal data (i.e., the target EEG feature signal) after time-frequency analysis, and a convolutional neural network (CNN) is used to extract features from the time-frequency spectrum graph to generate a multi-dimensional EEG feature signal for emotion prediction.

[0051] It should be noted that the above-mentioned convolutional neural network (CNN) is a deep learning model, which is widely used in fields such as image recognition and speech recognition. It can extract features from EEG signal data to help identify and predict emotional states. It extracts feature information at different levels through multiple convolutional layers and pooling layers, and outputs the prediction results of emotional states through fully connected layers and softmax layers. During the training process, the back-propagation algorithm can be used to optimize the parameters of CNN so that it can more accurately predict emotional states.

[0052] In the specific implementation process, before calibrating the multi-dimensional EEG features according to the resting emotional state baseline to obtain the calibrated EEG features, the above method also includes the following steps: acquiring resting EEG signal data when the target user is in a resting state, the resting state being the brain activity state of the target user when not affected by emotional fluctuations or external stimuli; extracting the characteristic frequency band corresponding to the resting EEG signal data by analyzing the resting EEG signal data; and determining the resting emotional state baseline of the target user based on the characteristic frequency band.

[0053] In this solution, obtaining the target user's EEG signal data in a resting state can help determine the user's brain activity state when not affected by emotional fluctuations or external stimuli. By analyzing the resting EEG signal data and extracting the corresponding characteristic frequency bands, the user's resting emotional state baseline can be determined. Based on the baseline, the user's emotional state can be calibrated more accurately, thereby improving the accuracy and reliability of emotional prediction, and better understanding and analyzing the user's brain activity, providing a stronger basis and support for emotional prediction.

[0054] Specifically, EEG signal data is collected when the user is in a resting state, and the data is analyzed to extract characteristic frequency bands and determine the user's initial resting emotional baseline parameters, where the resting state refers to the state of brain activity when the user is not subject to obvious emotional fluctuations or external stimuli.

[0055] In the specific implementation process, the resting emotional state baseline is calibrated according to the multi-dimensional EEG features to obtain the calibrated emotional state baseline, which can be achieved through the following steps: the multi-dimensional EEG features are segmented using the sliding window method, and the first EEG feature is extracted from each sliding window; the first EEG feature is compared with the resting emotional state baseline, and the resting emotional state baseline is weighted updated to obtain the calibrated emotional state baseline.

[0056] In this scheme, the accuracy and stability of emotion prediction can be improved by calibrating the emotional state baseline. By utilizing multi-dimensional EEG features for segmented processing and extracting the first EEG feature, and comparing and weighted updating it with the resting emotional state baseline, it can better reflect the changes in the individual's emotional state, help improve the accuracy of emotion prediction, and make the prediction results more reliable and effective.

[0057] Specifically, the real-time EEG signals collected subsequently are processed in segments using the sliding window method. New EEG features are extracted from each sliding window and compared with the resting emotion baseline. The time length and step size of the sliding window are set according to the specific needs of emotion monitoring. The feature data generated by each window will perform a weighted update on the resting baseline parameters.

[0058] It should be noted that the above sliding window method is a commonly used signal processing technology, which is used to analyze local features in continuous signals, extract time domain and frequency domain features of EEG signals, and input these features into the emotion classifier for emotion prediction. First, the continuous EEG signal is divided into windows of fixed length, and then the features of the EEG signal are calculated in each window. The features may include time domain features (such as mean, variance, slope, etc.) and frequency domain features (such as power spectral density, frequency band energy, etc.). Then, they can be used to train the emotion classifier to predict the emotional state of the subjects.

[0059] In the specific implementation process, before the calibrated EEG features are input into the emotion prediction model to use the emotion prediction model to perform emotion prediction and output the emotion prediction results, the method also includes the following steps: determining the model structure of the emotion prediction model, the model structure includes an input layer, an intermediate convolutional layer and an output layer, the intermediate convolutional layer includes a convolutional neural network layer, a long short-term memory network layer, a convolutional layer and a recurrent layer, the convolutional neural network layer is used to extract EEG feature signals related to emotions, the long short-term memory network layer is used to capture the changes of EEG feature signals in time series, the convolutional layer is used to receive the EEG feature signals of the convolutional neural network layer and perform feature processing on the EEG feature signals to obtain emotion features, the recurrent layer receives the changes of the EEG feature signals output by the long short-term memory network layer in time series, and identifies the changes of emotion features in time series; constructing an initial emotion prediction model according to the model structure; iteratively training the initial emotion prediction model using a neural network using multiple sets of training data to obtain an emotion prediction model.

[0060] In this scheme, the model structure of the emotion prediction model is determined, including the input layer, the intermediate convolution layer and the output layer. The intermediate convolution layer includes the convolutional neural network layer, the long short-term memory network layer, the convolution layer and the loop layer, which can extract the EEG characteristic signals related to emotions and capture the changes of the EEG characteristic signals in the time series, and finally obtain the emotion characteristics. Through iterative training, an optimized emotion prediction model is obtained, which can more accurately predict the emotional state of an individual, help researchers and clinicians better understand and predict the emotional state of an individual, and provide an important reference for emotion management and treatment.

[0061] Specifically, a deep learning model is introduced as a baseline adjustment mechanism to dynamically adjust the emotional state baseline parameters according to the user's real-time EEG signal fluctuation information. The emotion prediction model includes a convolutional neural network (CNN) and a long short-term memory (LSTM) network. CNN is used to extract EEG feature signals corresponding to emotions, and LSTM is used to capture the changes in EEG signal features in time series. At the same time, the neural network structure contains a multi-layer fusion model, in which the convolutional layer is used to extract local emotional features, and the recurrent layer is used to identify the time series changes of emotional features.

[0062] It should be noted that the above-mentioned long short-term memory network (LSTM) is a commonly used recurrent neural network model, which is particularly suitable for processing sequence data. It can be used to capture long-term dependencies in EEG signal sequences, so as to more accurately predict the emotional state of the subjects. By introducing gating units to control the flow of information, it can effectively avoid the gradient vanishing and gradient exploding problems in traditional recurrent neural networks.

[0063] In the specific implementation process, real-time emotion monitoring and feedback are performed based on emotion prediction results, which can be achieved through the following steps: based on the emotion prediction results in a closed-loop feedback system, a feedback signal is generated to record the target user's emotional state change history according to the feedback signal.

[0064] In this solution, it can help monitor the user's emotional state in real time, and generate feedback signals based on the prediction results to help the user adjust his or her emotions. By recording the user's emotional state change history, it can better understand the user's emotional change patterns and help users better manage their emotions. It can help users more effectively cope with emotional fluctuations and enhance their emotional management capabilities, thereby improving the user's quality of life and work efficiency.

[0065] Specifically, the closed-loop feedback system generates a feedback signal based on the emotion fluctuation detection results and records the user's emotional state change history.

[0066] The technical solution provided by the above-mentioned embodiments of the present invention realizes personalized and dynamic monitoring of emotional states through advanced time-frequency analysis technology and adaptive emotion baseline calibration. The neural network structure of the present invention constructs a more accurate emotion prediction model for feature extraction and time-dependent processing of EEG signals. In addition, the closed-loop feedback system design realizes real-time tracking and data recording of emotional states, so that it not only has high-precision emotion prediction capabilities, but also can meet the needs of emotion monitoring in complex scenarios. This method effectively improves the stability and applicability of emotion prediction, and expands the application of EEG signals in the fields of emotion recognition and intelligent interaction.

[0067] From the above, it can be seen that in an embodiment of the present invention, in view of the limitations of traditional emotion prediction methods, an emotion prediction method based on EEG signals is proposed, which aims to solve the problems of individual data differences, non-stationary feature processing and real-time emotion feedback in the emotion recognition process. Emotion-related EEG features are extracted through multi-level time-frequency analysis, and the physiological differences of different users are responded to based on a personalized emotion baseline calibration method to achieve personalized adaptation of emotional states. In addition, a deep learning model combining convolutional neural network (CNN) and long short-term memory network (LSTM) is adopted to perform time series analysis on the extracted features to improve the accuracy of emotion prediction, and further a closed-loop feedback system is used to monitor emotional fluctuations in real time to provide support for emotion regulation and mental health.

[0068] In order to enable those skilled in the art to more clearly understand the technical solution of the present application, the implementation process of the emotion prediction method based on EEG signals of the present application will be described in detail below in combination with specific embodiments.

[0069] This embodiment relates to a specific method for predicting emotions based on EEG signals, which includes the following steps: a data acquisition module, a data processing module, a closed-loop feedback system, and an emotion monitoring module.

[0070] 1) The data acquisition module is used to obtain the user's EEG signals.

[0071] 2) The data processing module is used to carry the algorithm model.

[0072] 3) The closed-loop feedback system is used to continuously receive the real-time data stream provided by the emotion detection module and analyze the emotion fluctuations.

[0073] 4) The emotion monitoring module is used to identify changes in emotional characteristics in EEG signals and generate feedback signals.

[0074] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0075] The embodiment of the present application also provides an emotion prediction device based on EEG signals. It should be noted that the emotion prediction device based on EEG signals of the embodiment of the present application can be used to execute the emotion prediction method based on EEG signals provided by the embodiment of the present application. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions thereof will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0076] The following is an introduction to the emotion prediction device based on EEG signals provided in an embodiment of the present application.

[0077] Figure 3 is a structural block diagram of an emotion prediction device based on EEG signals according to an embodiment of the present application. Figure 3 As shown, the device includes: an acquisition unit 301, an analysis unit 302, a calibration unit 303, an input and output unit 304 and a monitoring feedback unit 305.

[0078] The acquisition unit 301 is used to acquire the EEG signal of the target user.

[0079] The analysis unit 302 is used to perform time-frequency analysis on the EEG signal to obtain a target EEG characteristic signal, and extract EEG features of different frequency bands from the target EEG characteristic signal to obtain a multi-dimensional EEG feature. The target EEG characteristic signal is an EEG signal related to emotion.

[0080] The calibration unit 303 is used to calibrate the resting emotional state baseline according to the multi-dimensional EEG characteristics to obtain the calibrated emotional state baseline, and extract the characteristics corresponding to the calibrated emotional state baseline to obtain the calibrated EEG characteristics. The resting emotional state baseline is a baseline formed by the characteristics of the brain activity state of the target user when not affected by emotional fluctuations or external stimuli.

[0081] The input-output unit 304 is used to input the calibrated EEG features into the emotion prediction model to use the emotion prediction model to perform emotion prediction and output the emotion prediction results. The emotion prediction model is a convergent model obtained by iteratively training a predetermined model using a neural network using multiple sets of training data. Each set of training data includes sample EEG features and the emotional state corresponding to the sample EEG features.

[0082] The monitoring and feedback unit 305 is used to perform real-time emotion monitoring and feedback based on the emotion prediction result.

[0083] In this embodiment, emotion-related EEG features are extracted through multi-level time-frequency analysis, and the physiological differences of different users are addressed based on a personalized emotion baseline calibration method to achieve personalized adaptation of emotional states. A deep learning model, i.e., the above-mentioned emotion prediction model, is used to perform time series analysis on the extracted features to improve the accuracy of emotion prediction, and further a closed-loop feedback system is used to monitor emotion fluctuations in real time to provide support for emotion regulation and mental health.

[0084] In the specific implementation process, the analysis unit includes a first decomposition module and a second decomposition module. The first decomposition module is used to decompose the EEG signal using wavelet transform to generate a first EEG characteristic signal, which is a multi-scale and multi-band global EEG signal; the second decomposition module is used to decompose the first EEG characteristic signal into multiple groups of intrinsic mode functions using the empirical mode decomposition method to obtain the characteristic signals corresponding to the emotions of different frequency components in each intrinsic mode function, and obtain the target EEG characteristic signal.

[0085] In this scheme, the target EEG characteristic signals are obtained through time-frequency analysis, including multi-scale and multi-band global EEG signals and emotion corresponding characteristic signals of different frequency components. By decomposing and processing the EEG signals, the emotion-related features in brain activities can be captured more accurately, thereby realizing the prediction of emotions. It can provide more accurate and reliable data support for emotion recognition and emotion monitoring, and help to further study the relationship between emotions and EEG signals.

[0086] During the specific implementation process, the analysis unit includes a first generation module and a second generation module. The first generation module is used to generate a time-frequency spectrum diagram according to the target EEG characteristic signal; the second generation module is used to use a convolutional neural network to extract features from the time-frequency spectrum diagram to generate multi-dimensional EEG features.

[0087] In this scheme, by extracting EEG features of different frequency bands, generating time-frequency spectrograms, and using convolutional neural networks to extract features from the time-frequency spectrograms, multi-dimensional EEG features are obtained, which can more comprehensively capture the information in EEG signals and improve the accuracy and reliability of emotion prediction. Using multi-dimensional EEG features for emotion prediction can more accurately analyze the emotional state of an individual and help understand the emotion regulation mechanism and the occurrence mechanism of emotional disorders. At the same time, through the application of deep learning algorithms, a large amount of EEG signal data can be effectively processed to improve the efficiency and accuracy of emotion prediction.

[0088] During the specific implementation process, the above-mentioned device also includes a second acquisition unit, a second analysis unit and a first determination unit. The second acquisition unit is used to calibrate the multi-dimensional EEG features according to the resting emotional state baseline, and obtain resting EEG signal data when the target user is in a resting state before obtaining the calibrated EEG features. The resting state is the brain activity state of the target user when not affected by emotional fluctuations or external stimuli; the second analysis unit is used to analyze the resting EEG signal data to extract the characteristic frequency band corresponding to the resting EEG signal data; the first determination unit is used to determine the resting emotional state baseline of the target user based on the characteristic frequency band.

[0089] In this solution, obtaining the target user's EEG signal data in a resting state can help determine the user's brain activity state when not affected by emotional fluctuations or external stimuli. By analyzing the resting EEG signal data and extracting the corresponding characteristic frequency bands, the user's resting emotional state baseline can be determined. Based on the baseline, the user's emotional state can be calibrated more accurately, thereby improving the accuracy and reliability of emotional prediction, and better understanding and analyzing the user's brain activity, providing a stronger basis and support for emotional prediction.

[0090] During the specific implementation process, the calibration unit includes a processing module and a comparison module. The processing module is used to segment the multi-dimensional EEG features using a sliding window method and extract the first EEG feature from each sliding window; the comparison module is used to compare the first EEG feature with the resting emotional state baseline, and perform weighted update on the resting emotional state baseline to obtain a calibrated emotional state baseline.

[0091] In this scheme, the accuracy and stability of emotion prediction can be improved by calibrating the emotional state baseline. By utilizing multi-dimensional EEG features for segmented processing and extracting the first EEG feature, and comparing and weighted updating it with the resting emotional state baseline, it can better reflect the changes in the individual's emotional state, help improve the accuracy of emotion prediction, and make the prediction results more reliable and effective.

[0092] In the specific implementation process, the above-mentioned device also includes a second determination unit, a construction unit and a training unit. The second determination unit is used to determine the model structure of the emotion prediction model before inputting the calibrated EEG features into the emotion prediction model to use the emotion prediction model to perform emotion prediction and output the emotion prediction result. The model structure includes an input layer, an intermediate convolution layer and an output layer. The intermediate convolution layer includes a convolutional neural network layer, a long short-term memory network layer, a convolution layer and a loop layer. The convolutional neural network layer is used to extract EEG feature signals related to emotions. The long short-term memory network layer is used to capture the changes of EEG feature signals in time series. The convolution layer is used to receive the EEG feature signals of the convolutional neural network layer and perform feature processing on the EEG feature signals to obtain emotion features. The loop layer receives the changes of the EEG feature signals output by the long short-term memory network layer in time series and identifies the changes of emotion features in time series. The construction unit is used to construct an initial emotion prediction model according to the model structure; the training unit is used to iteratively train the initial emotion prediction model using a neural network using multiple sets of training data to obtain an emotion prediction model.

[0093] In this scheme, the model structure of the emotion prediction model is determined, including the input layer, the intermediate convolution layer and the output layer. The intermediate convolution layer includes the convolutional neural network layer, the long short-term memory network layer, the convolution layer and the loop layer, which can extract the EEG characteristic signals related to emotions and capture the changes of the EEG characteristic signals in the time series, and finally obtain the emotion characteristics. Through iterative training, an optimized emotion prediction model is obtained, which can more accurately predict the emotional state of an individual, help researchers and clinicians better understand and predict the emotional state of an individual, and provide an important reference for emotion management and treatment.

[0094] In the specific implementation process, the monitoring feedback unit includes a third generation module, which is used to generate a feedback signal based on the emotion prediction result in the closed-loop feedback system, so as to record the change history of the target user's emotional state according to the feedback signal.

[0095] In this solution, it can help monitor the user's emotional state in real time, and generate feedback signals based on the prediction results to help the user adjust his or her emotions. By recording the user's emotional state change history, it can better understand the user's emotional change patterns and help users better manage their emotions. It can help users more effectively cope with emotional fluctuations and enhance their emotional management capabilities, thereby improving the user's quality of life and work efficiency.

[0096] The above-mentioned emotion prediction device based on EEG signals includes a processor and a memory. The above-mentioned acquisition unit, analysis unit, calibration unit, input and output unit, monitoring feedback unit, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize the corresponding functions. The above-mentioned modules are all located in the same processor; or, the above-mentioned modules are located in different processors in any combination.

[0097] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the problem of low emotion recognition accuracy caused by ignoring individual differences and non-stationary features of data in the emotion recognition process in the prior art can be solved by adjusting kernel parameters.

[0098] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0099] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the emotion prediction method based on EEG signals.

[0100] Specifically, the emotion prediction method based on EEG signals includes:

[0101] Step S201, obtaining the EEG signal of the target user.

[0102] Specifically, by analyzing the EEG signals of the target user, we can understand the patterns and characteristics of their brain activity and thus infer their emotional state. EEG signals can provide objective data about the user's emotions, which is more accurate and objective than traditional self-reporting methods. By analyzing EEG signals, we can identify brain activity patterns associated with different emotional states. For example, emotional states of pleasure, stress, and anxiety will show different characteristics in the brain, thereby more accurately predicting the emotional state of the target user.

[0103] Step S202, performing time-frequency analysis on the EEG signal to obtain a target EEG characteristic signal, and extracting EEG features of different frequency bands from the target EEG characteristic signal to obtain a multi-dimensional EEG feature, wherein the target EEG characteristic signal is an EEG signal related to emotion.

[0104] Specifically, by performing time-frequency analysis on EEG signals, extracting target EEG feature signals and further extracting EEG features in different frequency bands, multi-dimensional EEG features are obtained, thereby realizing the extraction and analysis of EEG signals related to emotions. This can effectively capture emotion-related features in EEG signals, and provide more accurate and comprehensive information for subsequent emotion predictions. By analyzing multi-dimensional EEG features, we can have a more comprehensive understanding of the emotional state of the subjects, thereby realizing accurate prediction of emotions. Through the analysis of EEG signals, we can also explore the relationship between emotions and brain activity.

[0105] Step S203, calibrate the resting emotional state baseline according to the multi-dimensional EEG features to obtain the calibrated emotional state baseline, and extract the features corresponding to the calibrated emotional state baseline to obtain the calibrated EEG features. The resting emotional state baseline is a baseline formed by the features of the target user's brain activity state when he is not affected by emotional fluctuations or external stimuli.

[0106] Specifically, based on the extracted multi-dimensional EEG features, individual emotional state baseline calibration is performed. By calibrating the resting emotional state baseline, the emotional characteristics of the target user can be more accurately extracted, thereby improving the accuracy and reliability of emotion prediction. The calibrated EEG features obtained by calibrating the emotional state baseline can better reflect the emotional state of the target user, thereby improving the accuracy and stability of emotion prediction. By comparing the difference between the calibrated emotional state baseline and the actual emotional state, the emotional state of the target user can be more accurately identified and predicted, providing a more effective basis for further emotional intervention and regulation.

[0107] Step S204, input the calibrated EEG features into the emotion prediction model to use the emotion prediction model to perform emotion prediction, and output the emotion prediction result. The emotion prediction model is a convergent model obtained by iteratively training a predetermined model using a neural network using multiple sets of training data. Each set of training data includes sample EEG features and the emotional state corresponding to the sample EEG features.

[0108] Specifically, the calibrated EEG features are input into the emotion prediction model for emotion prediction. The training data is iteratively trained through a neural network to obtain a convergence model, and an emotion prediction model based on a neural network is constructed to achieve emotion prediction. This can more accurately predict the emotional state of an individual, help individuals better understand and manage their emotions in practical applications, and improve the effect of emotion regulation.

[0109] Step S205: performing real-time emotion monitoring and feedback based on the emotion prediction result.

[0110] Specifically, by monitoring the EEG signals of the test subjects, the system can capture their emotional state in real time. Based on the prediction results, the system can provide corresponding feedback, such as reminding the test subjects to relax when their emotions fluctuate greatly, or providing encouragement and support when they are depressed. Real-time emotional monitoring and feedback can help the test subjects better understand their emotional changes and take timely measures to regulate their emotions, thereby improving their emotional management ability and mental health level, helping the test subjects to better understand and control their emotions, and improve their quality of life and work efficiency.

[0111] Optionally, a time-frequency analysis is performed on the EEG signal to obtain a target EEG characteristic signal, including: using wavelet transform to decompose the EEG signal to generate a first EEG characteristic signal, the first EEG characteristic signal being a multi-scale and multi-band global EEG signal; using an empirical mode decomposition method to decompose the first EEG characteristic signal into multiple groups of intrinsic mode functions to obtain characteristic signals corresponding to emotions of different frequency components in each intrinsic mode function, thereby obtaining a target EEG characteristic signal.

[0112] Optionally, EEG features of different frequency bands are extracted from the target EEG feature signal to obtain multi-dimensional EEG features, including: generating a time-frequency spectrum diagram according to the target EEG feature signal; and extracting features from the time-frequency spectrum diagram using a convolutional neural network to generate multi-dimensional EEG features.

[0113] Optionally, before calibrating the multi-dimensional EEG features according to the resting emotional state baseline to obtain the calibrated EEG features, the method also includes: acquiring resting EEG signal data when the target user is in a resting state, the resting state being the brain activity state of the target user when not affected by emotional fluctuations or external stimuli; extracting the characteristic frequency band corresponding to the resting EEG signal data by analyzing the resting EEG signal data; and determining the resting emotional state baseline of the target user based on the characteristic frequency band.

[0114] Optionally, the resting emotional state baseline is calibrated according to the multi-dimensional EEG features to obtain a calibrated emotional state baseline, including: using a sliding window method to segment the multi-dimensional EEG features and extracting a first EEG feature from each sliding window; and comparing the first EEG feature with the resting emotional state baseline, and weighted updating the resting emotional state baseline to obtain a calibrated emotional state baseline.

[0115] Optionally, before inputting the calibrated EEG features into the emotion prediction model to use the emotion prediction model to perform emotion prediction and outputting the emotion prediction results, the method also includes: determining a model structure of the emotion prediction model, the model structure including an input layer, an intermediate convolutional layer and an output layer, the intermediate convolutional layer including a convolutional neural network layer, a long short-term memory network layer, a convolutional layer and a recurrent layer, the convolutional neural network layer is used to extract EEG feature signals related to emotions, the long short-term memory network layer is used to capture changes in EEG feature signals in a time series, the convolutional layer is used to receive the EEG feature signals of the convolutional neural network layer and perform feature processing on the EEG feature signals to obtain emotion features, the recurrent layer receives changes in the EEG feature signals output by the long short-term memory network layer in a time series, and identifies changes in emotion features in the time series; constructing an initial emotion prediction model according to the model structure; and iteratively training the initial emotion prediction model using a neural network using multiple sets of training data to obtain an emotion prediction model.

[0116] Optionally, based on the emotion prediction results, real-time emotion monitoring and feedback are performed, including: generating a feedback signal based on the emotion prediction results in a closed-loop feedback system, and recording the target user's emotional state change history according to the feedback signal.

[0117] An embodiment of the present invention provides an emotion prediction system, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing any one of the above-mentioned EEG signal-based emotion prediction methods.

[0118] An embodiment of the present invention provides a processor, which is used to run a program, wherein the program executes the above-mentioned emotion prediction method based on EEG signals when running.

[0119] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0120] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0121] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0122] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0123] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0124] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0125] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0126] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are 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 memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0127] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0128] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for predicting emotions based on electroencephalogram signals, characterized in that: include: Obtain the target user's EEG signal; Performing time-frequency analysis on the EEG signal to obtain a target EEG characteristic signal, and extracting EEG characteristics of different frequency bands from the target EEG characteristic signal to obtain a multi-dimensional EEG feature, wherein the target EEG characteristic signal is an EEG signal related to emotion; Calibrate the resting emotional state baseline according to the multi-dimensional EEG feature to obtain a calibrated emotional state baseline, and extract features corresponding to the calibrated emotional state baseline to obtain a calibrated EEG feature, wherein the resting emotional state baseline is a baseline formed by features of the target user's brain activity state when the target user is not subject to emotional fluctuations or external stimuli; Inputting the calibrated EEG features into an emotion prediction model to perform emotion prediction using the emotion prediction model, and outputting an emotion prediction result, wherein the emotion prediction model is a convergent model obtained by iteratively training a predetermined model using a neural network using multiple sets of training data, and each set of the training data includes sample EEG features and an emotional state corresponding to the sample EEG features; Based on the emotion prediction results, real-time emotion monitoring and feedback are performed.

2. The method according to claim 1, characterized in that Performing time-frequency analysis on the EEG signal to obtain a target EEG characteristic signal includes: Decomposing the EEG signal by wavelet transform to generate a first EEG characteristic signal, wherein the first EEG characteristic signal is a global EEG signal with multiple scales and multiple frequency bands; The first EEG characteristic signal is decomposed into multiple groups of intrinsic mode functions using an empirical mode decomposition method to obtain characteristic signals corresponding to emotions of different frequency components in each of the intrinsic mode functions, thereby obtaining the target EEG characteristic signal.

3. The method according to claim 1, characterized in that Extracting EEG features of different frequency bands from the target EEG feature signal to obtain multi-dimensional EEG features, including: Generate a time-frequency spectrum diagram according to the target EEG characteristic signal; A convolutional neural network is used to extract features from the time-frequency spectrum to generate the multi-dimensional EEG features.

4. The method according to claim 1, characterized in that Before calibrating the multi-dimensional EEG feature according to the resting emotional state baseline to obtain the calibrated EEG feature, the method further includes: Acquiring resting electroencephalogram signal data when the target user is in a resting state, wherein the resting state is a brain activity state of the target user when the target user is not subject to emotional fluctuations or external stimulation; By analyzing the resting EEG signal data, a characteristic frequency band corresponding to the resting EEG signal data is extracted; A resting emotional state baseline of the target user is determined according to the characteristic frequency band.

5. The method according to claim 1, characterized in that The resting emotional state baseline is calibrated according to the multi-dimensional EEG characteristics to obtain a calibrated emotional state baseline, including: Using a sliding window method to segment the multi-dimensional EEG features, extracting a first EEG feature from each sliding window; The first EEG feature is compared with the resting emotional state baseline, and the resting emotional state baseline is weighted updated to obtain the calibrated emotional state baseline.

6. The method according to claim 1, characterized in that Before inputting the calibrated EEG features into the emotion prediction model to perform emotion prediction using the emotion prediction model and outputting the emotion prediction result, the method further includes: Determine a model structure of the emotion prediction model, the model structure comprising an input layer, an intermediate convolution layer and an output layer, the intermediate convolution layer comprising a convolutional neural network layer, a long short-term memory network layer, a convolution layer and a recurrent layer, the convolutional neural network layer is used to extract EEG characteristic signals related to emotions, the long short-term memory network layer is used to capture changes in the EEG characteristic signals in a time series, the convolution layer is used to receive the EEG characteristic signals of the convolutional neural network layer and perform feature processing on the EEG characteristic signals to obtain emotion features, the recurrent layer receives changes in the EEG characteristic signals output by the long short-term memory network layer in a time series, and identifies changes in the emotion features in the time series; Constructing an initial emotion prediction model according to the model structure; The initial emotion prediction model is iteratively trained using a neural network using multiple groups of the training data to obtain the emotion prediction model.

7. The method according to claim 1, characterized in that Based on the emotion prediction results, real-time emotion monitoring and feedback are performed, including: In a closed-loop feedback system, a feedback signal is generated based on the emotion prediction result, and a history of changes in the emotional state of the target user is recorded according to the feedback signal.

8. An emotion prediction device based on EEG signals, characterized in that: The device comprises: An acquisition unit, used for acquiring an EEG signal of a target user; An analysis unit, configured to perform time-frequency analysis on the EEG signal to obtain a target EEG characteristic signal, and extract EEG characteristics of different frequency bands from the target EEG characteristic signal to obtain a multi-dimensional EEG characteristic, wherein the target EEG characteristic signal is an EEG signal related to emotion; a calibration unit, configured to calibrate a resting emotional state baseline according to the multi-dimensional EEG feature to obtain a calibrated emotional state baseline, and extract features corresponding to the calibrated emotional state baseline to obtain a calibrated EEG feature, wherein the resting emotional state baseline is a baseline formed by features of a brain activity state of the target user when the target user is not subject to emotional fluctuations or external stimuli; An input-output unit, used for inputting the calibrated EEG features into an emotion prediction model, so as to use the emotion prediction model to perform emotion prediction, and output an emotion prediction result, wherein the emotion prediction model is a convergent model obtained by iteratively training a predetermined model using a neural network using multiple sets of training data, and each set of the training data includes sample EEG features and an emotional state corresponding to the sample EEG features; The monitoring and feedback unit is used to perform real-time emotion monitoring and feedback based on the emotion prediction result.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 7.

10. An emotion prediction system, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include methods for executing any one of claims 1 to 7.

Images