Emotion recognition and multidimensional teaching quality evaluation system and method based on artificial intelligence

Abstract

The present invention discloses an emotion recognition and multi-dimensional teaching quality evaluation system and method based on artificial intelligence, which relates to the field of artificial intelligence and new teaching technologies. The system includes an emotion recognition model that can efficiently extract and fuse emotion feature information in multi-dimensional peripheral physiological signals, thereby achieving accurate emotion classification. Based on these analysis results, the classroom teacher-student status analysis and teaching quality evaluation system constructs a classroom status analysis framework by quantifying student emotions, teacher emotions and classroom interaction characteristics, providing educators with rich, quantitative classroom status data, and generating multi-dimensional teaching quality evaluation indicators. The present invention can dynamically generate and execute teaching adjustment strategies based on the evaluation results, and generate a structured teaching classroom status report at the same time, which can effectively solve the core problems of student distraction and insufficient learning results in current teaching scenarios.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and new teaching technologies, and in particular to an artificial intelligence-based emotion recognition and multi-dimensional teaching quality evaluation system and method. Background Art

[0002] With the development of educational informatization, improving the quality of classroom teaching in the information age has become a key goal in the field of teaching. However, existing classroom assessment methods mostly rely on simple statistical analysis and fail to fully utilize the emotional state information of students and teachers for comprehensive evaluation. Traditional emotion classification methods typically use machine learning algorithms, but these methods often require manual feature design and rely on domain expert knowledge, which has limitations in scalability and adaptability. These methods rarely involve identifying emotion data through peripheral physiological signals, which have the advantage of being easily acquired through wearable devices and can objectively reflect the human body's emotional state.

[0003] In addition, the existing emotion classification model based on peripheral physiological signals has deficiencies in extracting and fusing complex features between different physiological signals, resulting in the accuracy of emotion classification unable to meet the requirements of classroom teacher-student status analysis and teaching quality evaluation. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an emotion recognition and multi-dimensional teaching quality evaluation system and method based on artificial intelligence. The system includes an emotion recognition model that can efficiently extract and fuse emotion feature information in multi-dimensional peripheral physiological signals, thereby achieving accurate emotion classification. Based on these analysis results, the classroom teacher-student status analysis and teaching quality evaluation system constructs a classroom status analysis framework by quantifying student emotions, teacher emotions and classroom interaction characteristics, providing educators with rich, quantitative classroom status data, and generating multi-dimensional teaching quality evaluation indicators. The present invention can dynamically generate and execute teaching adjustment strategies based on the evaluation results, and generate a structured teaching classroom status report, which can effectively solve the core problems of student distraction and insufficient learning outcomes in current teaching scenarios. It provides a scientific basis for teaching evaluation and empowers teachers to optimize teaching strategies, with significant practical application value.

[0005] The present invention adopts the following technical solutions to achieve the invention objectives:

[0006] The emotion recognition and multi-dimensional teaching quality evaluation system based on artificial intelligence is characterized by including:

[0007] The physiological signal perception and fusion module, the emotion feature engineering and preprocessing module, the emotion classification module, the emotion-driven classroom state analysis module, the dynamic teaching intervention module, the teaching quality evaluation and optimization module and the intelligent analysis report generation and feedback module are connected in sequence; the physiological signal perception and fusion module includes a multimodal physiological sensor integration unit and a peripheral physiological signal data fusion unit; the emotion feature engineering and preprocessing module includes a signal preprocessing unit, a data segmentation and labeling unit and a label encoding and feature vector quantization unit; the emotion classification module includes a multi-scale feature extraction layer, a spatial attention optimization layer, a channel global information interaction layer, a parameter-free feature weighting layer and an adaptive pooling and classification layer; the emotion-driven classroom state analysis module includes a student emotion dimension analysis unit, a teacher emotion dimension analysis unit and a classroom interaction dimension analysis unit; the dynamic teaching intervention module includes a real-time monitoring and early warning unit, an intelligent strategy generation unit, a multimodal intervention execution unit and an intervention effect evaluation unit; the teaching quality evaluation and optimization module includes a data integration and analysis unit, a multi-dimensional evaluation generation unit and an optimization suggestion generation unit; the intelligent analysis report generation and feedback module includes a data integration and visualization unit, an interactive analysis unit and an intelligent distribution and feedback unit.

[0008] An evaluation method for emotion recognition and multidimensional teaching quality evaluation system based on artificial intelligence includes the following steps:

[0009] S01: Real-time acquisition and integration of human peripheral physiological signal data;

[0010] S02: Data preprocessing of peripheral physiological signals;

[0011] The signal preprocessing unit performs preliminary processing on the peripheral physiological signals of teachers and students through a noise filter and a Min-Max normalization processor to improve the accuracy and efficiency of subsequent analysis. The noise filter uses a Butterworth low-pass filter to remove sensor jitter or electromagnetic interference, and uses a Butterworth high-pass filter to remove baseline drift interference. The transfer function formula of the Butterworth low-pass filter is shown in formula (1), and the transfer function formula of the Butterworth high-pass filter is shown in formula (2):

[0012] (1);

[0013] Where: N represents the filter order; ω c represents the cutoff frequency; s The complex frequency variable representing the input signal;

[0014] (2);

[0015] Where: D0 represents the cutoff frequency; Represents the midpoint of the image frequency domain distance to the center of the spectrum; is the abscissa of the midpoint in the frequency domain; is the ordinate of the midpoint in the frequency domain;

[0016] The Min-Max normalization processor uses the Min-Max method to unify all physiological signals into the interval [0, 1]. Suppose the input sequence is ,in, n Represents the length of the sequence. The output sequence after Min-Max normalization is , as shown in formula (3):

[0017] (3);

[0018] in: Indicates the first values, and are the maximum and minimum values ​​of all samples in the input sequence x respectively;

[0019] S03: student and teacher sentiment classification;

[0020] S04: Judgment of classroom status by students and teachers;

[0021] S05: Adjustment of classroom status between teachers and students;

[0022] S06: Classroom teaching quality assessment;

[0023] S07: Output of teaching classroom status report.

[0024] As a further limitation of this technical solution, the specific steps of S03 are:

[0025] S031: Obtain first emotion data information using a multi-scale feature extraction layer;

[0026] S032: Obtain second emotion data information using the channel global information interaction layer;

[0027] S033: Use the spatial attention optimization layer to obtain the third emotion data information;

[0028] S034: Obtain fourth emotion data information using a parameter-free feature weighting layer;

[0029] S035: Use the residual connection layer to obtain the fifth emotion data information;

[0030] S036: Use adaptive pooling and classification layers to get sentiment classification output.

[0031] As a further limitation of this technical solution, the specific steps of S04 are:

[0032] S041: Analyze students’ learning emotions using the student emotion dimension analysis unit;

[0033] The student emotion dimension analysis unit focuses on the emotional state of individual and group students. It quantifies the attention, emotional stability, positive and negative learning emotions during the learning process through the positive learning emotion intensity index, negative learning emotion intensity index, learning motivation index and mood fluctuation index, and assists in identifying potential problems in the teaching content or environment. The learning focus weight is a parameter used to quantify the external factors that affect students' concentration. The calculation formula of the positive learning emotion intensity index is shown in formula (4):

[0034] (4);

[0035] in: n Indicates the number of time segments into which class time can be divided; Indicates the i In the time segment, k Class actively learns the confidence of emotions; Indicates the i In the time segment, k Confidence of negative learning emotions; Indicates the i In the time segment, k confidence of class-neutral learning emotions; Indicates the k The class actively learns the weight of the sentiment; Indicates the k The weight of the negative learning sentiment, Indicates the k weights of class-neutral learning emotions;

[0036] Indicates the i The external active learning weights in the time slices are used to quantize the i The influence of external factors on the positive learning emotion intensity index at each time point is calculated by weighting formula (5):

[0037] (5);

[0038] in: Indicates the i The difficulty of the teaching content at a certain time point, represents the quality of the teaching method at the i-th time point, Indicates the i Adaptability of classroom sessions at different time points, 、 and are the weight coefficients of these three external influencing factors respectively;

[0039] The negative learning emotion intensity index is used to quantify the intensity of students' negative emotions in the classroom, helping teachers identify potential problems in the teaching content or environment and adjust teaching strategies in a timely manner. The external negative learning weight is used to quantify the weight of the impact of students' negative external environment on the classroom effect. The calculation formula of the negative learning emotion intensity index is shown in formula (6):

[0040] (6);

[0041] in: Indicates the i The external negative learning weights in the time slices are used to quantize the i The influence of external factors on the negative learning emotion intensity index at each time point is calculated by weighting formula (7):

[0042] (7);

[0043] in: Indicates the i The difficulty of the teaching content at a certain time point, Indicates the i Disruptive factors in the classroom environment at each time point, Indicates the i The negativity of teacher feedback at each time point, 、 and are the weight coefficients of these three external influencing factors respectively;

[0044] The learning motivation index is used to evaluate students' learning enthusiasm driven by positive emotions, exclude the interference of neutral emotions, and quantify the motivational effect of classroom activities on students. The learning motivation weight is used to quantify the parameters of external or internal factors that affect students' learning motivation. The formula of the learning motivation index is shown in formula (8):

[0045] (8);

[0046] in: Indicates the i The learning momentum weight in the time segment is used to quantify the i The influence of external factors on the learning motivation index at each time point is calculated by weighted formula (9):

[0047] (9);

[0048] in: Indicates the i The students' interest in the subject at a certain point in time, Indicates the i Teacher incentives at each time point, Indicates the i Family expectations and support at each time point, , and are the weight coefficients of these three external influencing factors respectively;

[0049] The emotional fluctuation index is used to analyze the stability of students' emotional state. The impact of classroom rhythm or external interference on students' emotions is judged by the fluctuation amplitude of the emotional value. For each time point, the square of the difference between the emotional value and the average emotional value is calculated, multiplied by the weight, summed, averaged, and squared to obtain the emotional fluctuation index. The formula of the emotional fluctuation index is shown in formula (10):

[0050] (10);

[0051] in: Indicates the t The weight coefficient in each time segment; Indicates the t The student's emotional value at a time point; Indicates the total class time; Represents the weighted average sentiment value; calculated by formula (11):

[0052] (11);

[0053] S042: Analyze teachers’ teaching emotions using the teacher emotion dimension analysis unit;

[0054] The positive teaching emotion intensity index is used to quantify the positive emotional expression of teachers in the classroom, reflecting their teaching input and appeal. The external teaching positive weight is used to quantify the influence of teachers' positive emotions on the classroom atmosphere. The calculation formula of the positive teaching emotion intensity index is shown in formula (12):

[0055] (12);

[0056] in: Indicates the i In the time segment, k confidence level of positive teaching emotions; Indicates the i In the time segment, k Confidence of negative teaching emotions; Indicates the i In the time segment, k Confidence of neutral teaching emotions; Indicates the k the weight of positive teaching emotions; Indicates the k the weight of negative teaching emotions; Indicates the k the weight of class-neutral teaching emotions;

[0057] Indicates the i The external active teaching weight in the time segment is used to quantify the i The influence of external factors on the positive teaching emotion intensity index at each time point is calculated by weighting formula (13):

[0058] (13);

[0059] in: Indicates the i The vividness of the teaching language at each time point; Indicates the i The enthusiasm of classroom interaction at each time point; and are the weight coefficients of these two external influencing factors respectively;

[0060] The negative teaching emotion intensity index is used to evaluate teachers' negative emotions caused by classroom challenges and assist teachers in adjusting their teaching strategies. The external negative teaching weight is used to quantify the impact of teachers' negative emotions on teaching effectiveness. The calculation formula of the negative teaching emotion intensity index is shown in formula (14):

[0061] (14);

[0062] in: Indicates the i The external negative teaching weight in the time segment is used to quantify the i The influence of external factors on the negative learning emotion intensity index at each time point is calculated by weighting formula (15):

[0063] (15);

[0064] in: Indicates the i The situation of class schedule pressure at a certain time point, Indicates the i Insufficient preparation of teaching content at a certain time point, Indicates the i The external interference at a certain time point, , and are the weight coefficients of these three external influencing factors respectively;

[0065] S043: Analyze classroom interaction using the classroom interaction dimension analysis unit;

[0066] The teacher-student emotional synchronization index is used to analyze the matching degree of the emotional state of teachers and students, reflecting the consistency of emotional communication and the quality of interaction in the classroom. The teacher's emotional sequence is multiplied by the synchronization weight and the correlation is calculated with the student's emotional sequence to obtain the teacher-student emotional synchronization index. The calculation formula of the teacher-student emotional synchronization index is shown in formula (16):

[0067] (16);

[0068] in: Indicates that the teacher t The sentiment value at a time point; Indicates that students t The sentiment value at a time point; Indicates the t The weight of a time point reflects the importance of that moment; represents the weighted average of teachers’ sentiment values, represents the weighted average of students’ sentiment values;

[0069] The interaction frequency index is used to count the intensity of teacher-student interaction within a unit of time and to evaluate whether the classroom activity design effectively promotes student participation. The calculation formula of the interaction frequency index is shown in formula (17):

[0070] (17);

[0071] in: Indicates the t The number of teacher-student interactions at each time point;

[0072] The interaction quality index is based on the proportion of interactions marked with positive emotions, quantifying the positive effects of classroom interaction. The calculation formula of the interaction quality index is shown in formula (18):

[0073] (18);

[0074] in: Indicates the t The number of teacher-student interactions based on positive emotions at each time point.

[0075] As a further limitation of this technical solution, the specific steps of S06 are:

[0076] The classroom engagement index reflects the activeness and emotional synchronization of teacher-student interaction through the teacher-student emotional synchronization index, interaction frequency index, and interaction quality index, and measures the quality of classroom participation. The calculation formula of the classroom engagement index is shown in formula (19):

[0077] (19);

[0078] in: , and are the weights of the teacher-student emotional synchronization index, interaction frequency index, and interaction quality index respectively;

[0079] The teaching efficiency score measures the achievement of teaching objectives and the learning effect driven by emotions through the positive learning emotion intensity index, the negative learning emotion intensity index, the positive teaching emotion intensity index, and the negative teaching emotion intensity index. The calculation formula of the teaching efficiency score is shown in formula (20):

[0080] (20);

[0081] in: and are the weights of the positive learning emotion intensity index and the negative learning emotion intensity index, and the positive teaching emotion intensity index and the negative teaching emotion intensity index respectively;

[0082] The teaching motivation index measures the driving effect of students' learning motivation and teachers' emotions in the classroom through the learning motivation index, positive learning emotion intensity index and positive teaching emotion intensity index. The calculation formula of the teaching motivation index is shown in formula (21):

[0083] (twenty one);

[0084] in: , and are the weights of the learning motivation index, the positive learning emotion intensity index, and the positive teaching emotion intensity index, respectively;

[0085] The interaction-driven emotional stability index combines the interaction frequency index and the interaction quality index to evaluate the inhibitory effect of classroom interaction on the emotional fluctuation index and quantify the ability of teacher-student interaction to stabilize the emotional state of the classroom during teaching activities. The calculation formula of the interaction-driven emotional stability index is shown in formula (22):

[0086] (twenty two);

[0087] in: , and are the weights of interaction frequency index, interaction quality index and mood fluctuation index respectively.

[0088] Compared with existing technologies, the advantages and positive effects of the present invention are as follows: The present invention designs a neural network model that utilizes multiple peripheral physiological information for emotion classification, achieving efficient and accurate emotion classification, providing real-time, comprehensive teacher-student emotional data for classroom teaching. The present invention also designs a classroom teacher-student state analysis and teaching quality assessment system. This system dynamically monitors and quantitatively assesses the emotional state of students and teachers in the classroom environment, as well as the teaching and learning outcomes, based on the emotion data identified from the neural network model. This assessment system provides real-time feedback, enabling timely adjustments to teachers' teaching strategies and students' learning states, optimizing the classroom atmosphere and thereby enhancing teaching effectiveness and students' learning experience. Simultaneously, each student is individually assessed, monitoring their classroom performance and providing personalized learning optimization plans. A structured classroom status report can also be generated for the entire class. This innovative approach not only provides a more comprehensive understanding of emotional interactions in the classroom but also effectively promotes the continuous improvement of teaching quality. The present invention deeply integrates the collection of student and teacher peripheral physiological data, emotion classification and recognition, classroom teacher-student state analysis, and teaching quality assessment systems with the classroom teaching process, proposing an artificial intelligence-based emotion recognition and multi-dimensional teaching quality assessment system and method. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] Figure 1 This is a system framework diagram of the present invention.

[0090] Figure 2 This is an evaluation system diagram of the present invention.

[0091] Figure 3 It is the workflow diagram of the present invention. DETAILED DESCRIPTION

[0092] A specific embodiment of the present invention is described in detail below with reference to the accompanying drawings, but it should be understood that the protection scope of the present invention is not limited by the specific embodiment.

[0093] The present invention includes a physiological signal perception and fusion module, an emotion feature engineering and preprocessing module, an emotion classification module, an emotion-driven classroom state analysis module, a dynamic teaching intervention module, a teaching quality evaluation and optimization module, and an intelligent analysis report generation and feedback module, which are connected in sequence.

[0094] The physiological signal perception and fusion module includes a multimodal physiological sensor integration unit and a peripheral physiological signal data fusion unit; the physiological signal perception and fusion module collects the peripheral physiological signals of teachers and students in real time and performs data fusion; a customized peripheral physiological signal bracelet is used to collect the peripheral physiological signals of teachers and students in the classroom environment in real time and perform data fusion.

[0095] The multimodal physiological sensing integrated unit uses a customized peripheral physiological signal bracelet to collect human peripheral physiological signals of teachers and students in a classroom environment in real time. The customized peripheral physiological signal bracelet consists of a skin temperature sensor, a pulse wave sensor, a skin resistance sensor, and a charging electrode / data transmission port, wherein: the sampling frequency of the skin temperature sensor is 1Hz, and the measurement range is 0 to 50 degrees Celsius; the sampling frequency of the pulse wave sensor is 100Hz; the sampling frequency of the skin resistance sensor is 4Hz, and the AC excitation source frequency is 24Hz; the charging electrode / data transmission port is used for charging and transmitting data. The human peripheral physiological signal data includes one-axis data of skin resistance, one-axis data of pulse wave, and one-axis data of skin temperature. The one-axis data of heart rate can be calculated from the one-axis data of pulse wave. The peripheral physiological signal data fusion unit adjusts the peripheral physiological signals of teachers and students through an interpolation algorithm, and converts the sensor data of different sampling frequencies into a unified time series format to ensure that the data of different sensors have consistent time resolution. The peripheral physiological signal data fusion unit combines the peripheral physiological signals of teachers and students. The combined data is in a two-dimensional array format, with the pulse wave signal, skin resistance signal, heart rate signal and skin temperature signal arranged in order horizontally and arranged in time order vertically. The frequency of sensor data acquisition is f It can be set to an appropriate value to facilitate the subsequent data preprocessing process.

[0096] The emotion feature engineering and preprocessing module includes a signal preprocessing unit, a data segmentation and labeling unit, and a label encoding and feature quantization unit. The emotion feature engineering and preprocessing module preprocesses peripheral physiological signals, generating labeled emotion data segments and converting them into a form that facilitates feature extraction by the neural network model. The signal preprocessing unit performs preliminary processing on the teacher and student's peripheral physiological signals to improve the accuracy and efficiency of subsequent analysis. The signal preprocessing unit includes a noise filter and a minimum-maximum (Min-Max) normalization processor. The noise filter uses a low-pass filter to remove high-frequency noise and a high-pass filter to remove baseline drift and other low-frequency interference. The Min-Max normalization processor normalizes all physiological signals to the [0, 1] interval, ensuring numerical uniformity across different signal types. The data segmentation and labeling unit segments continuous physiological signals into fixed-length data segments, ensuring that each data segment reflects the complete emotional state over a period of time. The label encoding and feature quantization unit labels the segmented data segments containing peripheral physiological signal feature information with corresponding emotion labels. Each data segment is then converted into a two-dimensional vector by directly expanding its dimensionality, facilitating feature extraction by the neural network model.

[0097] The emotion classification module includes a multi-scale feature extraction layer, a spatial attention optimization layer, a channel global information interaction layer, a parameter-free feature weighting layer and an adaptive pooling and classification layer; the emotion classification module inputs the peripheral physiological signal feature information into the emotion classification model in batches, constructs the emotion classification model and determines the corresponding model parameters, and obtains the human emotion classification recognition output after training; the emotion classification module extracts multi-scale spatiotemporal features from the peripheral physiological signal feature information fragments through parallel convolution kernels through the multi-scale feature extraction layer, enhances the diversity of feature expression, and obtains the first emotion data information; the first emotion data information is grouped and dynamically convolved through the channel global information interaction layer to enhance the global information interaction between channels and obtain the second emotion data information; the second emotion data information is optimized through the spatial attention layer, and the lightweight channel attention mechanism is used to reduce parameter dependence and improve feature selectivity to obtain the third emotion data information; the third emotion data information is optimized through the parameter-free feature weighting layer and adaptive weights are generated by dynamically calculating the spatial distribution characteristics of the features to enhance the discriminability of the channel response to obtain the fourth emotion data information; the residual connection layer uses the residual connection to retain the original feature information while fusing the peripheral physiological signal features with the fourth emotion data information to reduce information loss and enhance the complementarity of multi-source features to obtain the fifth emotion data information; the fifth emotion data information is compressed through the adaptive pooling and classification layer through adaptive pooling to achieve emotion classification output.

[0098] The multi-scale feature extraction layer includes a parallel convolution kernel unit and an adaptive attention weighting unit. The two units use a selective kernel residual network to process the feature information fragments of the peripheral physiological signal, and extract multi-scale spatiotemporal features through parallel convolution kernels to enhance the diversity of feature expression and generate the first emotional data information. Among them, the parallel convolution kernel unit uses four different sizes of convolution kernels (3×1), (5×1), (7×1) and (9×1). Each convolution kernel independently processes the input feature map to capture different scale features from local to global. Each convolution kernel is followed by batch normalization to ensure the consistency and stability of features at each scale. The adaptive attention weighting unit obtains the first emotional data information by summing the feature maps output by the parallel convolution kernels and performing global average pooling, and then generating channel attention weights through a fully connected layer.

[0099] The channel global information interaction layer comprises a dual-pooling feature extraction unit, a receptive field feature generation unit, and a receptive field feature fusion unit. These three units process the first emotion data in real time through a dynamic multi-branch receptive field attention module and enhance global information interaction between channels through channel grouping and dynamic convolution to generate the second emotion data. The dual-pooling feature extraction unit extracts average and key features of the first emotion data through an average pooling path and a maximum pooling path, respectively. The average pooling path performs average pooling on the first emotion data and then expands the channel dimension through 1×1 grouped convolution to generate global average features. The maximum pooling path captures local key features through maximum pooling, followed by 1×1 grouped convolution to expand the channel dimension and generate local key features. The weights generated by the two paths are then weighted fused. The receptive field feature generation unit spatially expands and nonlinearly activates the weighted fused input features through convolution. Specifically, after convolution, the input features are normalized and nonlinearly mapped using batch normalization and the rectified linear unit (ReLU) activation function. The receptive field feature fusion unit first multiplies the feature data output by the receptive field feature generation unit by the dynamic weights generated by the dual-pooling feature extraction unit element by element to generate weighted features. Subsequently, the receptive field feature fusion unit expands the spatial dimension of the weighted feature through tensor reconstruction to restore the original input data format. This reconstruction operation ensures that cross-channel feature information can be subsequently interacted and fused in the original input data format by rearranging the spatial dimensions of the features. Finally, the receptive field feature fusion unit compresses the channel dimension to the target output channel number through the convolution layer, and generates the second emotion data information through batch normalization and ReLU activation function processing.

[0100] The spatial attention optimization layer includes a global spatial information aggregation unit and an adaptive channel weighting unit. These two units process the global spatial information of the second emotion data in real time through lightweight parameter design, and generate channel attention weights through one-dimensional convolution, enhancing the adaptive feature weighting between channels to generate the third emotion data. The global spatial information aggregation unit compresses the number of channels of the second emotion data through adaptive average pooling, eliminating spatial dimensionality differences and retaining the global statistical information of the channel features. The adaptive channel weighting unit dynamically generates channel attention weights by performing a one-dimensional convolution operation on the global descriptor, combining symmetric padding and a sigmoid activation function. The weights are then expanded to the original spatial dimensions through a broadcast mechanism and element-wise multiplied with the input feature map, dynamically strengthening key channel features while suppressing redundant information. The spatial attention optimization layer replaces the fully connected layer with a parameter-efficient one-dimensional convolution to achieve adaptive attention control between channels and generate the third emotion data.

[0101] The parameter-free feature weighting layer includes a spatial distribution quantization unit and a statistically driven weight generation unit. The two units process the spatial distribution characteristics of the third emotional data information in real time through parameter-free design, generate adaptive weights by dynamically calculating the spatial discreteness of the features, enhance the feature response of the spatially significant area, and generate the fourth emotional data information. Among them, the spatial distribution quantization unit extracts the spatial distribution characteristics of the features through global mean elimination and variance normalization, first calculates the spatial mean of each channel, and then calculates the square of the difference between the feature and the mean to quantify the spatial discreteness of the feature. The statistically driven weight generation unit generates channel weights by normalizing the spatial statistical results. After the weights are normalized to the range of [0, 1] by the Sigmoid activation function, they are multiplied element by element with the feature map of the third emotional data information to suppress low variance areas. The parameter-free feature weighting layer realizes spatial adaptive weighting through zero-parameter calculation to generate the fourth emotional data information.

[0102] The residual connection layer includes a multi-scale feature processing unit and a dynamic dimension matching unit. The two units retain the original feature information through residual connection while fusing the peripheral physiological signal features with the fourth emotional data information, reducing information loss and enhancing the complementarity of multi-source features. Among them, the multi-scale feature processing unit uses a convolution kernel with a variable receptive field to perform a nonlinear transformation on the input feature map, and strengthens the nonlinear expression ability of the features through batch normalization and activation function. The dynamic dimension matching unit dynamically matches the feature dimension according to the difference between the number of input channels and the number of output channels: if the input and output dimensions are inconsistent or the stride is not 1, the input features are linearly transformed through the convolution kernel and batch normalization operations to align them with the output dimensions of the multi-scale feature processing unit; if the dimensions are consistent and the stride is 1, the original input features are directly passed. The residual connection layer adds the output of the multi-scale feature processing unit and the output of the dynamic dimension matching unit element by element, and processes them through the ReLU activation function to finally generate the fifth emotional data information.

[0103] The adaptive pooling and classification layer includes a dynamic spatial compression unit and a fully connected classification unit. The two units adaptively compress the feature dimensions of the fifth emotion data information through adaptive pooling and classification layer to achieve emotion classification output. The dynamic spatial compression unit dynamically adjusts the spatial dimension of the input feature map through adaptive average pooling, compresses the height of the feature map to 1, and maintains the width as the original time step of the input sequence. The fully connected classification unit flattens the compressed features and maps them to a preset number of categories through a linear transformation layer, and finally generates the emotion classification output.

[0104] The emotion classification module can use students' peripheral physiological signals to identify and classify three positive learning emotions, three negative learning emotions, and two neutral learning emotions; the emotion classification module can use teachers' peripheral physiological signals to identify and classify three positive teaching emotions, three negative teaching emotions, and two neutral teaching emotions. It should be noted that this setting is not fixed. In actual application, the number and specific dimensions of emotion classification can be flexibly adjusted according to the course type, subject characteristics, or teaching needs.

[0105] The emotion-driven classroom state analysis module includes a student emotion dimension analysis unit, a teacher emotion dimension analysis unit, and a classroom interaction dimension analysis unit. After receiving the teacher and student emotion classification results from the emotion classification module, the emotion-driven classroom state analysis module calculates and quantifies student emotion, teacher emotion, and classroom interaction characteristics to construct a classroom state analysis framework, providing educators with rich, quantitative classroom state data. After receiving the teacher and student emotion classification results from the emotion classification module, the three units combine some statistical information and analyze to derive nine indicators of student emotion dimension, teacher emotion dimension, and classroom interaction dimension during classroom teaching. The student emotion dimension analysis unit focuses on the emotional state of individual and group students, quantifying attention, emotional stability, and the impact of positive and negative learning emotions during the learning process through the positive learning emotion intensity index, negative learning emotion intensity index, learning motivation index, and emotion fluctuation index, assisting in identifying potential problems in the teaching content or environment. The teacher emotion dimension analysis unit is used to evaluate teachers' emotional state and teaching adaptability in the classroom using the positive teaching emotion intensity index and the negative teaching emotion intensity index, providing a reference for optimizing teachers' teaching behavior and professional development. The classroom interaction dimension analysis unit is used to evaluate the quality of teacher-student emotional interaction and the effectiveness of classroom activity design using the teacher-student emotional synchronization index, interaction frequency index and interaction quality index, providing a basis for improving two-way communication in teaching.

[0106] The student emotion dimension analysis unit focuses on the emotional states of individual and group students. Using the Positive Emotion Learning Index (PELI), Negative Emotion Learning Index (NELI), Learning Motivation Index (LMI), and Emotion Fluctuation Index (EFI), it quantifies attention, emotional stability, and the impact of positive and negative learning emotions during learning, helping to identify potential issues within the teaching content or environment. The PELI measures the degree to which students focus their attention in class, driven by positive emotions, reflecting the appeal of the teaching content and environment. At each time point, the student's positive emotion value is multiplied by the corresponding external positive learning weight, the sum is divided by the sum of all emotions at all time points, and finally multiplied by 100% to obtain the percentage result. The external positive learning weight is a parameter used to quantify external factors that affect student focus. The Negative Learning Affect Intensity Index quantifies the intensity of students' negative emotions in the classroom, helping teachers identify potential issues within the teaching content or environment and adjust teaching strategies accordingly. At each time point, the student's negative affect score was multiplied by the external negative learning weight, the sum was divided by the total affect score across all time points, and finally multiplied by 100% to obtain a percentage. The external negative learning weight quantifies the impact of a student's negative external environment on classroom effectiveness. The learning motivation index assesses students' learning enthusiasm driven by positive affect, excluding the interference of neutral affect, and quantifies the motivational effect of classroom activities on students. At each time point, the positive affect score was multiplied by the learning motivation weight, the sum was divided by the total affect score across all time points excluding neutral affect, and finally multiplied by 100% to obtain a percentage. The learning motivation weight quantifies the parameters of external or internal factors that influence students' learning motivation. The mood fluctuation index analyzes the stability of students' emotional state and determines the impact of classroom rhythm or external interference on students' emotions by the amplitude of affect fluctuations. At each time point, the square of the difference between the affect score and the average affect score was calculated, multiplied by the weight, summed, averaged, and squared to obtain the mood fluctuation index.

[0107] The teacher emotion dimension analysis unit uses the Positive Emotion Teaching Index (PETI) and the Negative Emotion Teaching Index (NETI) to assess teachers' emotional state and teaching adaptability in the classroom. The PETI quantifies teachers' positive emotional expressions in the classroom, reflecting their teaching engagement and appeal. At each time point, the teacher's positive emotion value is multiplied by the external positive teaching weight, the sum is divided by the sum of teacher emotions across all time points, and the final result is multiplied by 100% to obtain the percentage result. The external positive teaching weight quantifies the impact of teachers' positive emotions on the classroom atmosphere. The Negative Emotion Teaching Index is used to assess teachers' negative emotions caused by classroom challenges and to assist teachers in adjusting their teaching strategies. At each time point, the teacher's negative emotion value is multiplied by the external negative teaching weight, the sum is divided by the sum of teacher emotions across all time points, and the final result is multiplied by 100% to obtain the percentage result. The external negative teaching weight quantifies the impact of teachers' negative emotions on teaching effectiveness.

[0108] The classroom interaction dimension analysis unit uses the Teacher-Student Emotional Synchronization Index (TSESI), Interaction Frequency Index (IFI), and Interaction Quality Index (IQI) to assess the quality of teacher-student emotional interactions and the effectiveness of classroom activity design, providing a basis for improving teaching quality. The Teacher-Student Emotional Synchronization Index analyzes the degree of matching between teacher and student emotional states, reflecting the consistency of emotional communication and the quality of interaction within the classroom. The Teacher-Student Emotional Synchronization Index is calculated by multiplying the teacher's emotional sequence by the synchronization weight and calculating the correlation with the student's emotional sequence. The Interaction Frequency Index measures the intensity of teacher-student interaction per unit time and assesses whether classroom activity design effectively promotes student engagement. At each time point, the Interaction Frequency Index is calculated by summing the number of interactions and dividing it by the total class duration. The Interaction Quality Index quantifies the positive impact of classroom interaction based on the proportion of interactions marked by positive emotions. At each time point, the number of interactions marked by positive emotions is multiplied by the Interaction Quality Weight, the sum is divided by the total number of interactions, and the final result is multiplied by 100% to obtain the percentage result.

[0109] The dynamic teaching intervention module includes a real-time monitoring and early warning unit, an intelligent strategy generation unit, a multimodal intervention execution unit, and an intervention effect evaluation unit. The dynamic teaching intervention module dynamically generates and executes teaching adjustment strategies by receiving the nine categories of classroom status analysis indicators output by the emotion-driven classroom status analysis module in real time and combining them with the real-time classification results of the emotion classification model. These four units dynamically generate and execute teaching adjustment strategies by receiving the indicators output by the emotion-driven classroom status analysis module in real time and combining them with the real-time classification results of the emotion classification model. This converts the emotion analysis results into precise intervention actions to directly optimize the classroom status. The real-time monitoring and early warning unit continuously receives the indicators output by the emotion-driven classroom status analysis module and combines them with the real-time classification results of the emotion classification model. Using preset threshold rules, it identifies key classroom issues in real time and immediately sends intervention signals to subsequent units. The intelligent strategy generation unit dynamically generates differentiated teaching adjustment strategies based on the triggered warning type and real-time data. The adjustment strategy generation process comprehensively considers multiple parameters to ensure that the strategy is accurately matched to the current classroom context. The multimodal intervention execution unit converts the generated strategy into multi-channel execution actions. The teacher's device receives intervention suggestions in real time and displays them in order of urgency. Students receive personalized feedback or learning resource support through their smart devices. In terms of environmental linkage, the system automatically controls classroom equipment to achieve coordinated adjustments. The intervention effect evaluation unit verifies the effectiveness of the strategy by recalculating relevant indicators after the intervention and comparing them with pre-intervention data.

[0110] The teaching quality assessment and optimization module includes a data integration and analysis unit, a multi-dimensional assessment generation unit, and an optimization suggestion generation unit. The teaching quality assessment and optimization module generates multi-dimensional teaching quality assessment indicators by integrating the classroom status analysis indicators output by the emotion-driven classroom status analysis module with the intervention strategy execution records and effect data of the dynamic teaching intervention module, combined with the classification results of the emotion classification module, and then proposes optimization suggestions based on the data-driven analysis results. The three units generate multi-dimensional teaching quality assessment indicators by integrating the output results of the classroom status judgment module and the dynamic teaching intervention module, combined with the classification results of the emotion classification module, and then propose optimization suggestions based on the data-driven analysis results. The data integration and analysis unit receives and processes the nine indicator results output by the classroom status judgment module, the intervention strategy execution records and effect data of the dynamic teaching intervention module, and the real-time classification results of the emotion classification module. Through data cleaning, feature fusion, and multi-source information alignment, the data integration and analysis unit converts the scattered raw data into a structured assessment dataset, providing a unified data foundation for subsequent analysis. The multidimensional evaluation generation unit quantitatively generates multidimensional teaching quality indicators based on the integrated data, and quantifies the classroom teaching effect through the classroom participation index, teaching efficiency score, teaching motivation index and interaction-driven emotional stability index.

[0111] The data integration and analysis unit receives and processes the nine indicators of student and teacher classroom status output by the classroom status assessment module, the intervention strategy execution records and effectiveness data from the dynamic teaching intervention module, and the real-time classification results from the sentiment classification module. Through data cleaning, feature fusion, and multi-source information alignment, the data integration and analysis unit transforms the scattered raw data into a structured evaluation dataset, providing a unified data foundation for subsequent analysis. The multidimensional evaluation generation unit quantifies and generates multidimensional teaching quality indicators based on the integrated data. The Class Participation Index (CPI) measures the activeness and emotional synchronization of teacher-student interactions through the teacher-student emotional synchronization index, interaction frequency index, and interaction quality index, reflecting the level of activeness and emotional synchronization of teacher-student interactions and measuring classroom participation quality. The Teaching Efficiency Score (TES) measures the achievement of teaching objectives and the effectiveness of emotion-driven learning through the positive learning emotion intensity index, negative learning emotion intensity index, positive teaching emotion intensity index, and negative teaching emotion intensity index. The Teaching Momentum Index (TMI) measures the driving effect of student learning motivation and teacher emotions in the classroom through the Learning Motivation Index, the Positive Learning Affect Intensity Index, and the Positive Teaching Affect Intensity Index. The Interaction-Driven Stability Index (IDSI) combines the Interaction Frequency Index and the Interaction Quality Index to assess the effectiveness of classroom interaction in suppressing emotional fluctuations, quantifying the ability of teacher-student interaction to stabilize classroom emotions during teaching activities.

[0112] The optimization suggestion generating unit further analyzes abnormal or inefficient links of the evaluation indicators, combines educational theories with historical teaching data, and generates targeted optimization suggestions.

[0113] The intelligent analysis report generation and feedback module includes a data integration and visualization unit, an interactive analysis unit, and an intelligent distribution and feedback unit. The intelligent analysis report generation and feedback module generates a structured visualization report by integrating the output results of the emotion-driven classroom state analysis module, the dynamic teaching intervention module, and the teaching quality assessment and optimization module, combined with the classification results of the emotion classification module. These three units generate a structured visualization report by integrating the output results of the classroom state judgment module, the dynamic teaching intervention module, and the teaching quality assessment and optimization module, combined with the classification results of the emotion classification module. The data integration and visualization unit receives and processes multi-source data from the classroom state judgment module, the dynamic teaching intervention module, and the teaching quality assessment and optimization module, including nine categories of indicators for student and teacher classroom state judgment, intervention strategy execution records and effectiveness data, multi-dimensional teaching quality indicators and optimization suggestions, and generates a structured visualization report based on the classification results of the emotion classification module. This unit transforms the scattered raw data into structured visualization elements, generating core report components that present the spatiotemporal distribution of emotion states, display the dynamic trends of key indicators through a timeline, and present optimization suggestions in a tabular format. The interactive analysis unit allows users to deeply explore data relevance through interactive operations such as filtering and comparison. This interactive design shifts data presentation from static reports to dynamic analysis tools, helping educators quickly identify the root causes of problems and verify the effectiveness of interventions. The intelligent distribution and feedback unit accurately pushes reports to educators, administrators, or parents through multiple channels. Simultaneously, this unit collects user feedback on report content and integrates it into the system to optimize subsequent data processing logic and report generation strategies.

[0114] An evaluation method for emotion recognition and multidimensional teaching quality evaluation system based on artificial intelligence includes the following steps:

[0115] S01: Real-time acquisition and integration of human peripheral physiological signal data;

[0116] The real-time collection and fusion of peripheral physiological signals is carried out in the context of a classroom teaching system. Through a physiological signal perception and fusion module, a customized peripheral physiological signal bracelet is used to collect the peripheral physiological signals of teachers and students in a classroom environment in real time and perform data fusion.

[0117] The human body's peripheral physiological signal data includes one-axis data of skin resistance, one-axis data of pulse wave and one-axis data of skin temperature. The one-axis data of heart rate can be calculated through the one-axis data of pulse wave. The peripheral physiological signal data fusion unit adjusts the peripheral physiological signals of teachers and students through an interpolation algorithm, and converts the sensor data of different sampling frequencies into a unified time series format to ensure that the data of different sensors have consistent time resolution. The peripheral physiological signal data fusion unit splices and merges the peripheral physiological signals of teachers and students. The merged data is in a two-dimensional array format, with the pulse wave signal, skin resistance signal, heart rate signal and skin temperature signal in the horizontal direction, and the frequency of sensor data acquisition arranged in chronological order in the vertical direction. f It can be set to a suitable frequency, such as 100 Hz, which means that the size of data generated per second is 100 rows and 4 columns, which is convenient for subsequent data preprocessing.

[0118] S02: Data preprocessing of peripheral physiological signals;

[0119] The data preprocessing of the peripheral physiological signals is performed through the emotional feature engineering and preprocessing module to preprocess the peripheral physiological signals, generate emotional data segments with labels, and convert them into a form that is convenient for the neural network model to extract features. The signal preprocessing unit performs preliminary processing on the peripheral physiological signals of teachers and students through a noise filter and a Min-Max normalization processor to improve the accuracy and efficiency of subsequent analysis. Among them, the noise filter uses a Butterworth low-pass filter to remove high-frequency noise such as sensor jitter or electromagnetic interference, and uses a Butterworth high-pass filter to remove baseline drift and other low-frequency interference. The transfer function formula of the Butterworth low-pass filter is shown in formula (1), and the transfer function formula of the Butterworth high-pass filter is shown in formula (2):

[0120] (1);

[0121] Where: N represents the filter order, which determines the steepness of the transition band and the filtering characteristics. The higher the order, the faster the stop band attenuation, but the transition band width increases and the phase delay also increases; ω c represents the cutoff frequency; s The complex frequency variable representing the input signal;

[0122] (2);

[0123] Where: D0 represents the cutoff frequency; Represents the midpoint of the image frequency domain distance to the center of the spectrum; is the abscissa of the midpoint in the frequency domain; is the ordinate of the midpoint in the frequency domain;

[0124] The Min-Max normalization processor uses the Min-Max method to unify all physiological signals into the range [0, 1] to ensure the numerical uniformity of different types of signals, which is convenient for subsequent fusion and analysis. Suppose the input sequence is ,in, n Represents the length of the sequence. The output sequence after Min-Max normalization is , as shown in formula (3):

[0125] (3);

[0126] in: Indicates the first values, and are the maximum and minimum values ​​of all samples in the input sequence x respectively;

[0127] The data segmentation and annotation unit uses a fixed-length window to segment the continuous sensor data into fixed-length data segments for input into the recognition model. Set the sensor acquisition frequency to f =100Hz, according to the characteristics of human peripheral physiological signals, the length is selected W = 100 windows, that is, each window contains W / f = 1 second of sensor data. Sliding step size S It can be set to 50, that is, the data coverage is S / W =50%, thus dividing the data into W OK M List, M is the number of columns after all sensor data are combined. When pulse wave signals, skin resistance signals, heart rate signals, and skin temperature signals are used, the size of a data segment is 100 rows and 4 columns.

[0128] The label encoding and feature vectorization unit then labels the segmented data segments containing peripheral physiological signal feature information with corresponding emotional labels. The labels include three positive learning emotions, three negative learning emotions, and two neutral learning emotions of students, as well as three positive teaching emotions, three negative teaching emotions, and two neutral teaching emotions of teachers. Finally, each data segment is converted into a two-dimensional vector form by directly expanding the dimension, which facilitates feature extraction by the neural network model.

[0129] S03: student and teacher sentiment classification;

[0130] The emotion classification of students and teachers is to input the emotion data segments into the emotion classification model in batches through the emotion classification module, construct the emotion classification model and determine the corresponding model parameters, and obtain the human emotion classification recognition output after training; the emotion classification model uses a multi-scale feature extraction layer to extract multi-scale spatiotemporal features from the peripheral physiological signal feature information segments using parallel convolution kernels to enhance the diversity of feature expression and obtain the first emotion data information; the first emotion data information is grouped and dynamically convolved through the channel global information interaction layer to enhance the global information interaction between channels to obtain the second emotion data information; the second emotion data information is optimized through the spatial attention layer. Through the lightweight channel attention mechanism, the parameter dependence is reduced and the feature selectivity is improved to obtain the third emotional data information; through the parameter-free feature weighting layer, the third emotional data information is subjected to adaptive weights generated by dynamically calculating the spatial distribution characteristics of the features to enhance the discriminability of the channel response and obtain the fourth emotional data information; through the residual connection layer, the residual connection is used to retain the original feature information while the peripheral physiological signal features are fused with the fourth emotional data information to reduce information loss and enhance the complementarity of multi-source features to obtain the fifth emotional data information; through the adaptive pooling and classification layer, the feature dimensions of the fifth emotional data information are compressed through adaptive pooling to achieve emotional classification output.

[0131] Among them, the emotion classification module can classify and count the three positive learning emotions, three negative learning emotions and two neutral learning emotions of students through the students' peripheral physiological signals; the emotion classification module can classify and count the three positive teaching emotions, three negative teaching emotions and two neutral teaching emotions of teachers through the teachers' peripheral physiological signals. It should be noted that for the sake of convenience, the number of emotions set in this system here is three positive learning emotions, three negative learning emotions, two neutral learning emotions, three positive teaching emotions, three negative teaching emotions and two neutral teaching emotions, but this setting is not fixed. In actual application, the number and specific dimensions of emotion classification can be flexibly adjusted according to the course type, subject characteristics or teaching needs.

[0132] S04: Judgment of classroom status by students and teachers;

[0133] The student and teacher classroom status is determined through an emotion-driven classroom status analysis module. Based on the teacher and student emotion classification results of the emotion classification module in S03, the module combines some statistical information, such as classroom interaction, homework completion, and course difficulty, to analyze the classification results based on various emotion-based indicators using the student emotion dimension analysis unit, the teacher emotion dimension analysis unit, and the classroom interaction dimension analysis unit. The analysis results in nine categories of indicators, namely the student emotion dimension, the teacher emotion dimension, and the classroom interaction dimension during the classroom teaching process. These nine categories of indicators quantify the dynamic characteristics of student emotion, teacher emotion, and classroom interaction, and jointly construct a classroom emotion analysis framework to provide educators with data-based and precise teaching optimization suggestions.

[0134] S05: Adjustment of classroom status between teachers and students;

[0135] The adjustment of the teacher-student classroom status is achieved by performing dynamic teaching intervention based on the output results in S03 and S04 through a dynamic teaching intervention module, thereby optimizing classroom teaching quality.

[0136] The real-time monitoring and early warning unit continuously receives the nine categories of indicators output by S04. Combined with the real-time classification results from S03, it uses preset threshold rules to identify key classroom issues in real time, such as student distraction, accumulated negative emotions, or a breakdown in teacher-student emotional synchronization, and immediately sends intervention signals to subsequent units. The intelligent strategy generation unit dynamically generates differentiated teaching adjustment strategies based on the triggered warning type and real-time data. The strategy generation process comprehensively considers multi-dimensional parameters to ensure that the strategy is precisely matched to the current classroom context. The multimodal intervention execution unit converts the generated strategy into multi-channel execution actions. The teacher's device receives intervention suggestions in real time and displays them in order of urgency. The student's device receives personalized feedback or learning resource support via smart devices. In terms of environmental linkage, the system automatically controls classroom equipment to achieve coordinated adjustment of the teaching environment and teaching behavior. The intervention effect evaluation unit verifies the effectiveness of the strategy by recalculating relevant indicators after the intervention and comparing them with the pre-intervention data, forming a continuous improvement cycle of "monitoring-intervention-optimization."

[0137] S06: Classroom teaching quality assessment;

[0138] The classroom teaching quality assessment is to generate multi-dimensional teaching quality assessment indicators based on the output results of S03, S04 and S05 through the teaching quality assessment and optimization module, and comprehensively evaluate the classroom teaching quality through the classroom teaching quality assessment classroom participation index, teaching efficiency score, teaching motivation index and teaching motivation index, and put forward optimization suggestions based on data-driven analysis results.

[0139] S07: Output of teaching classroom status report.

[0140] The teaching classroom status report output is generated by an intelligent analysis report generation and feedback module based on the results of S03, S04, S05 and S06 to generate a structured visual report. The intelligent analysis report generation and feedback module includes a data integration and visualization unit, an interactive analysis unit and an intelligent distribution and feedback unit.

[0141] The data integration and visualization unit receives and processes the teacher and student sentiment classification results from the sentiment classification module in S03, the nine indicators for determining student and teacher classroom status in S04, the implementation records and effectiveness data of intervention strategies for adjusting teacher and student classroom status in S05, and the multi-dimensional teaching quality indicators generated quantitatively in S06. This unit transforms the scattered raw data into structured visualization elements and generates core report components. These components display the spatiotemporal distribution of sentiment status using heat maps, the dynamic trends of key indicators using a timeline, and optimization suggestions in a list format. The interactive analysis unit allows users to deeply explore data correlations through interactive operations such as filtering and comparison. For example, teachers can click on a specific time period in the sentiment heat map to view the corresponding sentiment classification distribution and intervention strategy implementation records; or use the timeline slider to locate key events and compare indicator changes before and after the intervention. The classroom status report simultaneously displays relevant data for all students and averages all the data to accurately reflect the overall level of the class. This interactive design transforms data presentation from a static report into a dynamic analysis tool, helping educators quickly identify the root causes of problems and verify the effectiveness of interventions. The intelligent distribution and feedback unit accurately pushes reports to educators, managers, or parents through multiple channels. It also collects user feedback on the report content and integrates it into the system to optimize subsequent data processing logic and report generation strategies.

[0142] The specific steps of S03 are:

[0143] S031: Obtain first emotion data information using a multi-scale feature extraction layer;

[0144] The multi-scale feature extraction layer includes a parallel convolution kernel unit and an adaptive attention weighting unit. The two units use a selective kernel residual network to process peripheral physiological signal feature information fragments in real time, and extract multi-scale spatiotemporal features through parallel convolution kernels to enhance the diversity of feature expression and generate first emotional data information. The parallel convolution kernel unit uses four different sizes of convolution kernels: (3×1), (5×1), (7×1) and (9×1), where the padding parameters of all convolution kernels are set to (kernel size – 1) / 2, where kernel size is the convolution kernel size. Each convolution kernel independently processes the input feature map to capture different scale features from local to global. Each convolution kernel is followed by batch normalization to ensure the consistency and stability of features at each scale. The adaptive attention weighting unit splices the feature maps output by the parallel convolution kernels along the channel dimension, first sums the channel dimension of the features of each scale, fuses the multi-scale information into a single feature map, and then compresses it into a feature vector with a spatial dimension of 1 through global average pooling. Subsequently, a weight vector of [b, out_channel×4] is generated through a fully connected layer and a Sigmoid activation function, and reshaped into an attention weight tensor of [b, 4, out_channel, 1, 1], where b is the batch size and out_channel is the number of output channels. Finally, the weight tensor is element-wise multiplied with the original feature tensor and summed along the convolution kernel dimension to dynamically fuse multi-scale features, thereby generating the first emotional data information.

[0145] S032: Obtain second emotion data information using the channel global information interaction layer;

[0146] The channel global information interaction layer comprises a dual-pooling feature extraction unit, a receptive field feature generation unit, and a receptive field feature fusion unit. These three units process the first emotion data in real time using a dynamic multi-branch receptive field attention module. They enhance global information interaction between channels through channel grouping and dynamic convolution to generate the second emotion data. The dual-pooling feature extraction unit extracts average and key features from the first emotion data using an average pooling path and a maximum pooling path, respectively. The average pooling path performs spatial average pooling on the input features, then expands the channel dimension to in_channel × kernel_size² via a 1×1 grouped convolution (groups = in_channel), where groups is the number of groups and in_channel is the number of input channels, to generate global average feature weights. The maximum pooling path captures key local features through spatial max pooling, followed by a grouped convolution with the same structure to generate key local feature weights. The weights of the two paths are then weighted together (the average pooling path weight coefficients are 0.925 and the maximum pooling path weight coefficients are 0.075), achieving a dynamic balance between global and local features. The receptive field feature generation unit spatially expands the input features through depthwise separable convolution (groups = in_channel), followed by batch normalization and ReLU activation to generate an intermediate feature tensor that incorporates multi-scale spatial correlations. The receptive field feature fusion unit then performs element-wise multiplication on the weighted feature weights and the generated intermediate features, resulting in a weighted feature tensor of [b, c, kernel_size², h, w], where c is the number of channels, h is the height of the feature map, and w is the width of the feature map. The receptive field feature fusion unit performs a tensor reconstruction operation to expand the spatial dimensions from (h, w) to (h × kernel_size, w × kernel_size), restoring the spatial resolution of the original input and ensuring spatial continuity of cross-channel features. Finally, the reconstructed features are compressed to the target number of output channels through a 1×1 convolution, completing the complete feature processing flow from channel interaction to spatial reconstruction, and outputting the second sentiment data.

[0147] S033: Use the spatial attention optimization layer to obtain the third emotion data information;

[0148] The spatial attention optimization layer includes a global spatial information aggregation unit and an adaptive channel weighting unit. These two units process the global spatial information of the second emotion data information in real time through lightweight parameter design, and generate channel attention weights through one-dimensional convolution, enhance the adaptive feature weighting between channels, and generate third emotion data information. Among them, the global spatial information aggregation unit uses adaptive average pooling to compress the spatial dimension of the input feature map to 1 × 1, forming a channel-level global statistical descriptor [b, c, 1, 1]. This process retains the channel distribution information of the original features by eliminating spatial differences, while significantly reducing the amount of computation. The adaptive channel weighting unit models cross-channel interactions in the global descriptor through one-dimensional convolution: First, the pooled features [b, c, 1, 1] are converted to a channel-dimensional vector of [b, 1, c]. Subsequently, local feature interactions are performed using a convolution kernel of kernel size with symmetric padding (padding = (kernel size – 1) / 2) to generate candidate channel weights. After transposition and compression to the interval [0, 1] using a Sigmoid activation function, the weight tensor is restored to its original channel dimension of [b, c, 1, 1]. Finally, the channel weights are expanded to the original spatial dimensions [b, c, h, w] through a broadcast mechanism and element-wise multiplied with the input feature map, dynamically enhancing key channel features and suppressing redundant information. The spatial attention optimization layer replaces the fully connected layer with a parameter-efficient one-dimensional convolution to achieve adaptive attention control across channels and generate tertiary emotional data information.

[0149] S034: Obtain fourth emotion data information using a parameter-free feature weighting layer;

[0150] The parameter-free feature weighting layer includes a spatial distribution quantization unit and a statistically driven weight generation unit. These two units process the spatial distribution characteristics of the third emotional data information in real time through a parameter-free design, generate adaptive weights by dynamically calculating the spatial discreteness of the features, enhance the feature response of spatially significant areas, and generate the fourth emotional data information. Among them, the spatial distribution quantization unit first calculates the spatial mean of each channel through the global mean, and then quantifies the degree of discreteness of the features and the mean through a squared difference operation, generating an intermediate feature tensor that represents the spatial distribution discreteness. Specifically, the square of the difference between the input feature x and the global mean of the channel directly reflects the degree of deviation of the local area from the global distribution. The larger the value, the higher the spatial discreteness. The statistically driven weight generation unit generates channel weights by normalizing the spatial statistical results. After the weights are normalized to the range of [0, 1] through the Sigmoid activation function, the weights are expanded to the original feature dimension through broadcasting and multiplied element-by-element with the feature map of the third emotional data information, suppressing low-variance areas and strengthening features with significant spatial discreteness. The parameter-free feature weighting layer achieves spatial adaptive weighting through zero-parameter statistical calculations to generate the fourth emotional data information.

[0151] S035: Use the residual connection layer to obtain the fifth emotion data information;

[0152] The residual connection layer includes a multi-scale feature processing unit and a dynamic dimension matching unit. The two units retain the original feature information through residual connection while fusing the peripheral physiological signal features with the fourth emotional data information, reducing information loss and enhancing the complementarity of multi-source features. Among them, the multi-scale feature processing unit uses a convolution kernel with a variable receptive field to perform a nonlinear transformation on the input feature map, and strengthens the nonlinear expression ability of the features through batch normalization and activation function. The dynamic dimension matching unit dynamically matches the feature dimension according to the difference between the number of input channels and the number of output channels: if the input and output dimensions are inconsistent or the stride is not 1, the input features are linearly transformed through the convolution kernel and batch normalization operations to align them with the output dimensions of the multi-scale feature processing unit; if the dimensions are consistent and the stride is 1, the original input features are directly passed. The residual connection layer adds the output of the multi-scale feature processing unit and the output of the dynamic dimension matching unit element by element, and processes them through the ReLU activation function to finally generate the fifth emotional data information.

[0153] S036: Use adaptive pooling and classification layers to get sentiment classification output.

[0154] The adaptive pooling and classification layer includes a dynamic spatial compression unit and a fully connected classification unit. The two units adaptively compress the feature dimensions of the fifth emotion data information through adaptive pooling and classification layer to achieve emotion classification output. The dynamic spatial compression unit dynamically adjusts the spatial dimension of the input feature map through adaptive average pooling, compresses the height of the feature map to 1, and maintains the width as the original time step of the input sequence. The fully connected classification unit flattens the compressed features and maps them to a preset number of categories through a linear transformation layer, and finally generates the emotion classification output.

[0155] Among them, the emotion classification module can classify and count the three positive learning emotions, three negative learning emotions and two neutral learning emotions of students through the students' peripheral physiological signals; the emotion classification module can classify and count the three positive teaching emotions, three negative teaching emotions and two neutral teaching emotions of teachers through the teachers' peripheral physiological signals. It should be noted that for the sake of convenience, the number of emotions set in this system here is three positive learning emotions, three negative learning emotions, two neutral learning emotions, three positive teaching emotions, three negative teaching emotions and two neutral teaching emotions, but this setting is not fixed. In actual application, the number and specific dimensions of emotion classification can be flexibly adjusted according to the course type, subject characteristics or teaching needs.

[0156] The specific steps of S04 are:

[0157] S041: Analyze students’ learning emotions using the student emotion dimension analysis unit;

[0158] The student emotion dimension analysis unit focuses on the emotional state of individual and group students. It quantifies attention, emotional stability, and the impact of positive and negative learning emotions during the learning process through the positive learning emotion intensity index, negative learning emotion intensity index, learning motivation index, and emotional fluctuation index. It helps identify potential problems in the teaching content or environment. The positive learning emotion intensity index measures the degree of student concentration driven by positive emotions in the classroom, reflecting the attractiveness of the teaching content and environment to students. At each time point, the student's positive emotion value is multiplied by the corresponding external positive learning weight, and the sum is divided by the sum of emotions at all time points. The learning focus weight is a parameter used to quantify the external factors that affect students' concentration. The calculation formula of the positive learning emotion intensity index is shown in formula (4):

[0159] (4);

[0160] in: n Indicates the number of time segments into which class time can be divided; Indicates the i In the time segment, kClass actively learns the confidence of emotions; Indicates the i In the time segment, k Confidence of negative learning emotions; Indicates the i In the time segment, k confidence of class-neutral learning emotions; Indicates the k The class actively learns the weight of the sentiment; Indicates the k The weight of the negative learning sentiment, Indicates the k weights of class-neutral learning emotions;

[0161] Indicates the i The external active learning weights in the time slices are used to quantize the i The influence of external factors on the positive learning emotion intensity index at each time point is calculated by weighting formula (5):

[0162] (5);

[0163] in: Indicates the i The difficulty of the teaching content at a certain time point, represents the quality of the teaching method at the i-th time point, Indicates the i Adaptability of classroom sessions at different time points, 、 and are the weight coefficients of these three external influencing factors respectively;

[0164] The negative learning emotion intensity index is used to quantify the intensity of students' negative emotions in the classroom, helping teachers identify potential problems in the teaching content or environment and adjust teaching strategies in a timely manner. The external negative learning weight is used to quantify the weight of the impact of students' negative external environment on the classroom effect, such as the difficulty of the teaching content exceeding the students' ability; interference factors in the classroom environment; and the negativity of teacher feedback. The calculation formula of the negative learning emotion intensity index is shown in formula (6):

[0165] (6);

[0166] in: Indicates the i The external negative learning weights in the time slices are used to quantize the i The influence of external factors on the negative learning emotion intensity index at each time point is calculated by weighting formula (7):

[0167] (7);

[0168] in: Indicates the i The difficulty of the teaching content at a certain time point, Indicates the i Disruptive factors in the classroom environment at each time point, Indicates the i The negativity of teacher feedback at each time point, 、 and are the weight coefficients of these three external influencing factors respectively;

[0169] The learning motivation index is used to assess students' learning enthusiasm driven by positive emotions, excluding the interference of neutral emotions, and quantifying the motivational effect of classroom activities on students. At each time point, the positive emotion value is multiplied by the learning motivation weight, the sum is divided by the sum of the total emotion values ​​after excluding neutral emotions at all time points, and finally multiplied by 100% to obtain the percentage result. The learning motivation weight is used to quantify the parameters of external or internal factors that affect students' learning motivation. The formula for the learning motivation index is shown in formula (8):

[0170] (8);

[0171] in: Indicates the i The learning momentum weight in the time segment is used to quantify the i The influence of external factors on the learning motivation index at each time point is calculated by weighted formula (9):

[0172] (9);

[0173] in: Indicates the i The students' interest in the subject at a certain point in time, Indicates the i Teacher incentives at each time point, Indicates the i Family expectations and support at each time point, , and are the weight coefficients of these three external influencing factors respectively;

[0174] The emotional fluctuation index is used to analyze the stability of students' emotional state. The impact of classroom rhythm or external interference on students' emotions is judged by the fluctuation amplitude of the emotional value. For each time point, the square of the difference between the emotional value and the average emotional value is calculated, multiplied by the weight, summed, averaged, and squared to obtain the emotional fluctuation index. The formula of the emotional fluctuation index is shown in formula (10):

[0175] (10);

[0176] in: Indicates the t The weight coefficient in a time segment reflects the importance of the time point and its value range is [0, 1]; Indicates the t The student's emotional value at a time point; Indicates the total class time; Represents the weighted average sentiment value, which is used as the benchmark value to calculate the sentiment fluctuation amplitude at each time point; it is calculated by formula (11):

[0177] (11);

[0178] S042: Analyze teachers’ teaching emotions using the teacher emotion dimension analysis unit;

[0179] The teacher emotion dimension analysis unit is used to evaluate the teacher's emotional state and teaching adaptability in the classroom using the positive teaching emotion intensity index and the negative teaching emotion intensity index, providing a reference for optimizing teachers' teaching behavior and professional development. The positive teaching emotion intensity index is used to quantify the teacher's positive emotional expression in the classroom, reflecting their teaching input and appeal. At each time point, the teacher's positive emotion value is multiplied by the external positive teaching weight, the sum is divided by the sum of the teacher's emotions at all time points, and finally multiplied by 100% to obtain the percentage result. The external teaching positive weight is used to quantify the weight of the teacher's positive emotion on the classroom atmosphere. The calculation formula of the positive teaching emotion intensity index is shown in formula (12):

[0180] (12);

[0181] in: Indicates the i In the time segment, k confidence level of positive teaching emotions; Indicates the i In the time segment, k Confidence of negative teaching emotions; Indicates the i In the time segment, k Confidence of neutral teaching emotions; Indicates the k the weight of positive teaching emotions; Indicates the k the weight of negative teaching emotions; Indicates the kThe weights of the three types of teaching emotions are used to comprehensively evaluate the intensity of teachers' classroom teaching emotions by quantifying the impact of different positive, negative and neutral emotions on teaching behavior;

[0182] Indicates the i The external active teaching weight in the time segment is used to quantify the i The influence of external factors on the positive teaching emotion intensity index at each time point is calculated by weighting formula (13):

[0183] (13);

[0184] in: Indicates the i The vividness of the teaching language at each time point; Indicates the i The enthusiasm of classroom interaction at each time point; and are the weight coefficients of these two external influencing factors respectively;

[0185] The negative teaching emotion intensity index is used to evaluate teachers' negative emotions caused by classroom challenges and to assist teachers in adjusting their teaching strategies. At each time point, the teacher's negative emotion value is multiplied by the teaching negativity weight, the sum is divided by the sum of the teacher's emotions at all time points, and finally multiplied by 100% to obtain the percentage result. The external negative teaching weight is used to quantify the weight of the impact of teachers' negative emotions on teaching effectiveness. The calculation formula of the negative teaching emotion intensity index is shown in formula (14):

[0186] (14);

[0187] in: Indicates the i The external negative teaching weight in the time segment is used to quantify the i The influence of external factors on the negative learning emotion intensity index at each time point is calculated by weighting formula (15):

[0188] (15);

[0189] in: Indicates the i The situation of class schedule pressure at a certain time point, Indicates the i Insufficient preparation of teaching content at a certain time point, Indicates the i The external interference at a certain time point, , and are the weight coefficients of these three external influencing factors respectively;

[0190] S043: Analyze classroom interaction using the classroom interaction dimension analysis unit;

[0191] The classroom interaction dimension analysis unit is used to evaluate the quality of teacher-student emotional interaction and the effectiveness of classroom activity design using the teacher-student emotional synchronization index, interaction frequency index, and interaction quality index, providing a basis for improving two-way communication in teaching. The teacher-student emotional synchronization index is used to analyze the degree of matching between the teacher-student emotional state and reflect the consistency of emotional communication and interaction quality in the classroom. The teacher's emotional sequence (positive, neutral, and negative time series data) is multiplied by the synchronization weight and the correlation is calculated with the student's emotional sequence to obtain the teacher-student emotional synchronization index. The calculation formula of the teacher-student emotional synchronization index is shown in formula (16):

[0192] (16);

[0193] in: Indicates that the teacher t The sentiment value at a time point; Indicates that students t The sentiment value at a time point; Indicates the t The weight of a time point reflects the importance of that moment; represents the weighted average of teachers’ sentiment values, represents the weighted average of students’ sentiment values;

[0194] The interaction frequency index is used to count the intensity of teacher-student interaction per unit time and evaluate whether the classroom activity design effectively promotes student participation. At each time point, the interaction frequency index is obtained by summing the number of interactions and dividing it by the total class time. The calculation formula of the interaction frequency index is shown in formula (17):

[0195] (17);

[0196] in: Indicates the t The number of teacher-student interactions at each time point;

[0197] The interaction quality index is based on the proportion of interactions marked by positive emotions, quantifying the positive effects of classroom interaction (such as encouragement and effective communication). At each time point, the number of interactions based on positive emotions is multiplied by the interaction quality weight, the sum is divided by the total number of interactions, and finally multiplied by 100% to obtain the percentage result. The calculation formula of the interaction quality index is shown in formula (18):

[0198] (18);

[0199] in: Indicates the t The number of teacher-student interactions based on positive emotions at each time point.

[0200] The specific steps of S06 are:

[0201] The classroom teaching quality assessment is to generate multi-dimensional teaching quality assessment indicators based on the output results of S03, S04 and S05 through the teaching quality assessment and optimization module, and to make optimization suggestions based on the data-driven analysis results.

[0202] The data integration and analysis unit receives the teacher and student sentiment classification results from the sentiment classification module in S03, the nine indicators for determining the student and teacher classroom status in S04, and the intervention strategy execution records and effect data for adjusting the teacher and student classroom status in S05. Through data cleaning, feature fusion, and multi-source information alignment, the data integration and analysis unit transforms the dispersed raw data into a structured evaluation dataset, providing a unified data foundation for subsequent analysis.

[0203] The multi-dimensional evaluation generation unit quantifies and generates multi-dimensional teaching quality indicators based on the integrated data. The classroom participation index reflects the activeness and emotional synchronization of teacher-student interaction through the teacher-student emotional synchronization index, interaction frequency index, and interaction quality index, and measures the quality of classroom participation. The calculation formula of the classroom participation index is shown in formula (19):

[0204] (19);

[0205] in: , and These are the weights of the teacher-student emotional synchronization index, interaction frequency index, and interaction quality index, which can be set according to the actual classroom situation;

[0206] The teaching efficiency score measures the degree of achievement of teaching objectives and the learning effect of emotion-driven learning through the positive learning emotion intensity index, the negative learning emotion intensity index, the positive teaching emotion intensity index, and the negative teaching emotion intensity index. The calculation formula of the teaching efficiency score is shown in formula (20):

[0207] (20);

[0208] in: and The weights of the positive learning emotion intensity index and the negative learning emotion intensity index and the positive teaching emotion intensity index and the negative teaching emotion intensity index can be set according to the actual classroom situation;

[0209] The teaching motivation index measures the driving effect of students' learning motivation and teachers' emotions in the classroom through the learning motivation index, positive learning emotion intensity index and positive teaching emotion intensity index. The calculation formula of the teaching motivation index is shown in formula (21):

[0210] (twenty one);

[0211] in: , and The weights of the learning motivation index, the positive learning emotion intensity index, and the positive teaching emotion intensity index can be set according to the actual classroom situation;

[0212] The interaction-driven emotional stability index combines the interaction frequency index and the interaction quality index to evaluate the inhibitory effect of classroom interaction on the emotional fluctuation index and quantify the ability of teacher-student interaction to stabilize the emotional state of the classroom during teaching activities. The calculation formula of the interaction-driven emotional stability index is shown in formula (22):

[0213] (twenty two);

[0214] in: , and These are the weights of the interaction frequency index, interaction quality index, and emotional fluctuation index, which can be set according to the actual classroom situation.

[0215] The optimization suggestion generation unit further analyzes abnormalities or inefficiencies in the evaluation indicators and generates targeted optimization suggestions based on educational theory and historical teaching data. The optimization suggestion generation unit proposes actionable improvement plans based on the teacher and student emotion classification results of the emotion classification module in S03, the nine indicators of student and teacher classroom status judgment in S04, and the execution records and effect data of the intervention strategy for adjusting the teacher and student classroom status in S05.

[0216] Since the weights are affected by different factors, each weight can be determined specifically according to the actual situation in actual application. In the following examples of this patent, each weight is temporarily set to 1.

[0217] Example 1

[0218] In the analysis of the above-mentioned emotion classification-based artificial intelligence-based emotion recognition and multi-dimensional teaching quality evaluation system, the relevant emotions are classified and identified in the classroom teaching of a certain professional course, and the classified emotion data are intelligently processed and comprehensively analyzed through the system, and finally the output of the teaching classroom status report is realized.

[0219] S01: Real-time acquisition and integration of human peripheral physiological signal data

[0220] First, during the teaching process of professional courses, customized peripheral physiological signal bracelets are used to collect real-time peripheral physiological signals of teachers and students in the classroom environment. Among them, the human peripheral physiological signal data includes one-axis data of skin resistance, one-axis data of pulse wave and one-axis data of skin temperature. The one-axis data of heart rate can be calculated through the one-axis data of pulse wave. Then, the peripheral physiological signals of teachers and students are spliced ​​and merged. The merged data is in a two-dimensional array format, with the pulse wave signal, skin resistance signal, heart rate signal and skin temperature signal in the horizontal direction, and the frequency of sensor data acquisition arranged in chronological order in the vertical direction. f The setting is 100Hz, which means that the data generated every second is 100 rows and 4 columns.

[0221] S02: Data preprocessing of peripheral physiological signals

[0222] The peripheral physiological signals of teachers and students are preliminarily processed through noise filters and Min-Max normalization processors to improve the accuracy and efficiency of subsequent analysis. Among them, the noise filter uses a Butterworth low-pass filter to remove high-frequency noise such as sensor jitter, and a Butterworth high-pass filter to remove low-frequency interference such as baseline drift. Subsequently, a fixed length of W =100, sliding step S = 50, splitting the continuous sensor data into fixed-length data with a coverage of S / W= 50% of the data segments. Finally, the data segments containing the peripheral physiological signal feature information were labeled with corresponding emotion labels. Students' positive learning emotions were labeled as: happy, excited, and satisfied; students' neutral learning emotions were labeled as: calm and indifferent; and students' positive learning emotions were labeled as: sad, anxious, and angry. Teachers' positive teaching emotions were labeled as: joyful, excited, and confident; teachers' neutral teaching emotions were labeled as: calm and neutral; and teachers' negative teaching emotions were labeled as: irritable, tired, and frustrated.

[0223] S03: Student and teacher sentiment classification

[0224] The emotional data segments after data preprocessing are input into the emotion classification model in batches, the emotion classification model is constructed and the corresponding model parameters are determined. After training, the human emotion classification recognition output is obtained, as follows:

[0225] S031: Use the multi-scale feature extraction layer to obtain the first emotional data information

[0226] The multi-scale feature extraction layer includes a parallel convolution kernel unit and an adaptive attention weighting unit. These two units use a selective kernel residual network to process peripheral physiological signal feature information fragments in real time, and extract multi-scale spatiotemporal features through parallel convolution kernels, enhancing the diversity of feature expression and generating first emotion data information.

[0227] S032: Use the channel global information interaction layer to obtain the second emotional data information

[0228] The channel global information interaction layer includes a dual-pooling feature extraction unit, a receptive field feature generation unit, and a receptive field feature fusion unit. These three units process the first emotion data information in real time through a dynamic multi-branch receptive field attention module, and enhance the global information interaction between channels through channel grouping and dynamic convolution to generate the second emotion data information.

[0229] S033: Use the spatial attention optimization layer to obtain the third emotional data information

[0230] The spatial attention optimization layer includes a global spatial information aggregation unit and an adaptive channel weighting unit. These two units process the global spatial information of the second emotion data in real time through lightweight parameter design, generate channel attention weights through one-dimensional convolution, enhance the adaptive feature weighting between channels, and generate the third emotion data.

[0231] S034: Use the parameter-free feature weighting layer to obtain the fourth emotional data information

[0232] The parameter-free feature weighting layer includes a spatial distribution quantization unit and a statistically driven weight generation unit. These two units process the spatial distribution characteristics of the third emotional data information in real time through parameter-free design, generate adaptive weights by dynamically calculating the spatial discreteness of the features, enhance the feature response of spatially significant areas, and generate the fourth emotional data information.

[0233] S035: Use the residual connection layer to obtain the fifth emotion data information

[0234] The residual connection layer includes a multi-scale feature processing unit and a dynamic dimension matching unit. These two units retain the original feature information through residual connection while fusing the peripheral physiological signal features with the fourth emotion data information, reducing information loss and enhancing the complementarity of multi-source features to generate the fifth emotion data information.

[0235] S036: Using adaptive pooling and classification layers to obtain sentiment classification output

[0236] The adaptive pooling and classification layer includes a dynamic space compression unit and a fully connected classification unit. The two units adaptively compress the feature dimensions of the fifth emotion data information through adaptive pooling and classification layer to achieve emotion output.

[0237] S04: Judgment of students and teachers’ classroom status

[0238] Teacher A's professional course lasts 40 minutes. After collecting peripheral physiological signals from teachers and students in S01 and preprocessing them in S02, the emotion classification model in S03 generates emotional data for both teachers and students. For Teacher A, the duration of pleasure was 8 minutes, excitement was 6 minutes, and confidence was 2 minutes, meaning positive teaching emotions lasted 16 minutes, accounting for 40% of the total. Calmness lasted 12 minutes, and neutrality lasted 4 minutes, meaning neutral teaching emotions lasted 16 minutes, accounting for 40% of the total. Irritation lasted 4 minutes, fatigue lasted 2 minutes, and frustration lasted 2 minutes, meaning negative teaching emotions lasted 8 minutes, accounting for 20% of the total. For the student emotion classification, Student B is used as an example; other students will also receive data in the same format. For Student B, the duration of happiness was 6 minutes, excitement was 4 minutes, and satisfaction was 4 minutes, meaning that the duration of positive learning emotions was 14 minutes, accounting for 35%; the duration of calm was 12 minutes, and the duration of indifference was 8 minutes, meaning that the duration of neutral learning emotions was 20 minutes, accounting for 50%; the duration of sadness was 2 minutes, anxiety was 3.2 minutes, and anger was 0.8 minutes, meaning that the duration of negative learning emotions was 6 minutes, accounting for 15%. Teacher A and Student B interacted 30 times (such as asking questions, answering questions, writing on the blackboard, and making eye contact), 18 of which were accompanied by positive emotions. Assuming that the confidence level of each emotion is 1, to facilitate the calculation of the emotional fluctuation index, the three positive emotions are quantized as 1, the two neutral emotions are quantized as 0, and the three negative emotions are quantized as -1.

[0239] S041: Analyze students’ learning emotions using the student emotion dimension analysis unit

[0240] The student emotion dimension analysis unit focuses on the emotional state of individual and group students. It quantifies the attention, emotional stability, and the impact of positive and negative learning emotions during the learning process through the positive learning emotion intensity index, negative learning emotion intensity index, learning motivation index, and emotional fluctuation index, and assists in identifying potential problems in the teaching content or environment. Calculated by formulas (4), (6), (8), and (10), respectively:

[0241] ;

[0242] ;

[0243] ;

[0244] ;

[0245] S042: Analyze teachers’ teaching emotions using the teacher emotion dimension analysis unit

[0246] The teacher emotion dimension analysis unit is used to evaluate the teacher's emotional state and teaching adaptability in the classroom using the positive teaching emotion intensity index and the negative teaching emotion intensity index. The following formulas (12) and (14) are used to calculate the following:

[0247] ;

[0248] ;

[0249] S043: Analyze classroom interaction using the classroom interaction dimension analysis unit

[0250] The classroom interaction dimension analysis unit is used to evaluate the quality of teacher-student emotional interaction and the effectiveness of classroom activity design using the teacher-student emotional synchronization index, interaction frequency index, and interaction quality index, providing a basis for improving two-way communication in teaching. The teacher-student emotional synchronization index is calculated using formulas (16), (17), and (18) respectively:

[0251] ;

[0252] ;

[0253] ;

[0254] S05: Adjustment of teacher and student classroom status

[0255] Based on the outputs from S03 and S04, dynamic teaching interventions are implemented to optimize classroom teaching quality. First, the nine indicators output by the emotion-driven classroom status analysis module are combined with the real-time classification results of the emotion classification model. Using pre-set threshold rules, key classroom issues are identified in real time and intervention signals are immediately sent to subsequent units. Subsequently, differentiated teaching adjustment strategies are dynamically generated based on the triggered warning types and real-time data. Finally, the effectiveness of the strategies is verified by recalculating relevant indicators after the intervention and comparing them with the pre-intervention data.

[0256] S06: Classroom Teaching Quality Assessment

[0257] Based on the output results of S03, S04 and S05, multi-dimensional teaching quality evaluation indicators are generated, and optimization suggestions are proposed based on the data-driven analysis results. First, the nine indicators of teacher and student emotion classification results based on the emotion classification module in S03, the nine indicators of student and teacher classroom status judgment in S04, and the intervention strategy execution records and effect data of teacher and student classroom status adjustment in S05 are received. Through data cleaning, feature fusion and multi-source information alignment, the data integration and analysis unit converts the scattered raw data into a structured evaluation data set, providing a unified data basis for subsequent analysis. Then, based on the integrated data, multi-dimensional teaching quality indicators are quantitatively generated through classroom participation index, teaching efficiency score, teaching motivation index and interactive driving emotional stability index, which are calculated by formulas (19), (20), (21) and (22) respectively:

[0258] ;

[0259] ;

[0260] ;

[0261] ;

[0262] Finally, we further analyze the abnormal or inefficient links in the evaluation indicators, combine educational theories with historical teaching data, and generate targeted optimization suggestions.

[0263] S07: Teaching class status report output

[0264] A visualization report based on the output of S03, S04, S05, and S06. First, it receives and processes the teacher and student emotion classification results from the emotion classification module in S03, the nine indicators for determining student and teacher classroom status in S04, the implementation records and effectiveness data of intervention strategies for adjusting teacher and student classroom status in S05, and the multi-dimensional teaching quality indicators generated quantitatively in S06. Through data cleaning, standardization, and feature alignment, this unit transforms the scattered raw data into structured visualization elements and generates core report components. These components present the spatiotemporal distribution of emotion status using heat maps, the dynamic trends of key indicators using a timeline, and optimization suggestions in a tabular format. For Teacher A, the positive teaching emotion intensity index and the negative teaching emotion intensity index are 40.0% and 20.0%, respectively, with the former significantly higher than the latter, indicating a high level of teaching enthusiasm. The classroom engagement index, teaching efficiency score, and teaching motivation index are 63.3%, 80.0%, and 52.5%, respectively, indicating that Teacher A's teaching has achieved good teaching effectiveness and efficiency for Student B, but teaching motivation could be further improved. For Student B, the positive learning emotion intensity index, negative learning emotion intensity index, learning motivation index, mood swings index, and interaction-driven mood stability index were 35.0%, 15.0%, 70.0%, 6.7%, and 9.9%, respectively. This indicates that Student B's positive learning emotions far outweighed their negative ones during the course, indicating a high level of learning motivation and minimal mood swings. Furthermore, Student B's teacher-student emotional synchronization index, interaction frequency index, and interaction quality index with Teacher A were 55%, 0.75 (times / minute), and 60%, respectively. This indicates that Student B clearly perceived Teacher A's teaching style more than half the time, and that the quantity and quality of their interactions were high. The classroom status report also displays data from other students and averages all data to accurately reflect the overall level of the class.

[0265] Then, users are supported to deeply explore data relevance through interactive operations such as filtering and comparison. Finally, reports are accurately pushed to educators, managers, or parents through multiple channels, and user feedback on the report content is collected and integrated into the system to optimize subsequent data processing logic and report generation strategies.

[0266] The above disclosure is only a specific embodiment of the present invention, but the present invention is not limited thereto. Any changes that can be conceived by those skilled in the art should fall within the scope of protection of the present invention.

Claims

1. A method for emotion recognition and multi-dimensional teaching quality evaluation based on artificial intelligence, characterized by: The following steps are involved: S01: Real-time acquisition and integration of human peripheral physiological signal data; S02: Data preprocessing of peripheral physiological signals; The signal preprocessing unit performs preliminary processing on the peripheral physiological signals of teachers and students through a noise filter and a Min-Max normalization processor to improve the accuracy and efficiency of subsequent analysis. The noise filter uses a Butterworth low-pass filter to remove sensor jitter or electromagnetic interference, and uses a Butterworth high-pass filter to remove baseline drift interference. The transfer function formula of the Butterworth low-pass filter is shown in formula (1), and the transfer function formula of the Butterworth high-pass filter is shown in formula (2): （1）； Where: N represents the filter order; ω c represents the cutoff frequency; s The complex frequency variable representing the input signal; （2）； Where: D0 represents the cutoff frequency; Represents the midpoint of the image frequency domain distance to the center of the spectrum; is the abscissa of the midpoint in the frequency domain; is the ordinate of the midpoint in the frequency domain; The Min-Max normalization processor uses the Min-Max method to unify all physiological signals into the interval [0, 1]. Suppose the input sequence is ,in, n Represents the length of the sequence. The output sequence after Min-Max normalization is , as shown in formula (3): （3）； in: Indicates the first values, and are the maximum and minimum values ​​of all samples in the input sequence x respectively; S03: student and teacher sentiment classification; S04: Judgment of classroom status by students and teachers; S05: Adjustment of classroom status between teachers and students; S06: Classroom teaching quality assessment; S07: Output of teaching classroom status report; The specific steps of S03 are: S031: Obtain first emotion data information using a multi-scale feature extraction layer; S032: Obtain second emotion data information using the channel global information interaction layer; S033: Use the spatial attention optimization layer to obtain the third emotion data information; S034: Obtain fourth emotion data information using a parameter-free feature weighting layer; S035: Use the residual connection layer to obtain the fifth emotion data information; S036: Use adaptive pooling and classification layers to obtain sentiment classification output; The specific steps of S04 are: S041: Analyze students’ learning emotions using the student emotion dimension analysis unit; The student emotion dimension analysis unit focuses on the emotional state of individual and group students. It quantifies the attention, emotional stability, positive and negative learning emotions during the learning process through the positive learning emotion intensity index, negative learning emotion intensity index, learning motivation index and mood fluctuation index, and assists in identifying potential problems in the teaching content or environment. The learning focus weight is a parameter used to quantify the external factors that affect students' concentration. The calculation formula of the positive learning emotion intensity index is shown in formula (4): （4）； in: n Indicates the number of time segments into which class time can be divided; Indicates the i In the time segment, k Class actively learns the confidence of emotions; Indicates the i In the time segment, k Confidence of negative learning emotions; Indicates the i In the time segment, k confidence of class-neutral learning emotions; Indicates the k The class actively learns the weight of the sentiment; Indicates the k The weight of the negative learning sentiment, Indicates the k weights of class-neutral learning emotions; Indicates the i The external active learning weights in the time slices are used to quantize the i The influence of external factors on the positive learning emotion intensity index at each time point is calculated by weighting formula (5): （5）； in: Indicates the i The difficulty of the teaching content at a certain time point, represents the quality of teaching methods at the i-th time point, Indicates the i Adaptability of classroom sessions at different time points, 、 and are the weight coefficients of these three external influencing factors respectively; The negative learning emotion intensity index is used to quantify the intensity of students' negative emotions in the classroom, helping teachers identify potential problems in the teaching content or environment and adjust teaching strategies in a timely manner. The external negative learning weight is used to quantify the weight of the impact of students' negative external environment on the classroom effect. The calculation formula of the negative learning emotion intensity index is shown in formula (6): (6)； in: Indicates the i The external negative learning weights in the time slices are used to quantize the i The influence of external factors on the negative learning emotion intensity index at each time point is calculated by weighting formula (7): (7)； in: Indicates the i The difficulty of the teaching content at a certain time point, Indicates the i Disruptive factors in the classroom environment at each time point, Indicates the i The negativity of teacher feedback at each time point, 、 and are the weight coefficients of these three external influencing factors respectively; The learning motivation index is used to evaluate students' learning enthusiasm driven by positive emotions, exclude the interference of neutral emotions, and quantify the motivational effect of classroom activities on students. The learning motivation weight is used to quantify the parameters of external or internal factors that affect students' learning motivation. The formula of the learning motivation index is shown in formula (8): （8）； in: Indicates the i The learning momentum weight in the time segment is used to quantify the i The influence of external factors on the learning motivation index at each time point is calculated by weighted formula (9): （9）； in: Indicates the i The students’ interest in the subject at a certain point in time, Indicates the i Teacher incentives at each time point, Indicates the i Family expectations and support at each time point, , and are the weight coefficients of these three external influencing factors respectively; The emotional fluctuation index is used to analyze the stability of students' emotional state. The impact of classroom rhythm or external interference on students' emotions is judged by the fluctuation amplitude of the emotional value. For each time point, the square of the difference between the emotional value and the average emotional value is calculated, multiplied by the weight, summed, averaged, and squared to obtain the emotional fluctuation index. The formula of the emotional fluctuation index is shown in formula (10): （10）； in: Indicates the t The weight coefficient in each time segment; Indicates the t The student's emotional value at a time point; Indicates the total class time; Represents the weighted average sentiment value; calculated by formula (11): （11）； S042: Analyze teachers’ teaching emotions using the teacher emotion dimension analysis unit; The positive teaching emotion intensity index is used to quantify the positive emotional expression of teachers in the classroom, reflecting their teaching input and appeal. The external teaching positive weight is used to quantify the influence of teachers' positive emotions on the classroom atmosphere. The calculation formula of the positive teaching emotion intensity index is shown in formula (12): （12）； in: Indicates the i In the time segment, k confidence level of positive teaching emotions; Indicates the i In the time segment, k Confidence of negative teaching emotions; Indicates the i In the time segment, k Confidence of neutral teaching emotions; Indicates the k the weight of positive teaching emotions; Indicates the k the weight of negative teaching emotions; Indicates the k the weight of class-neutral teaching emotions; Indicates the i The external active teaching weight in the time segment is used to quantify the i The influence of external factors on the positive teaching emotion intensity index at each time point is calculated by weighting formula (13): （13）； in: Indicates the i The vividness of the teaching language at each time point; Indicates the i The enthusiasm of classroom interaction at each time point; and are the weight coefficients of these two external influencing factors respectively; The negative teaching emotion intensity index is used to evaluate teachers' negative emotions caused by classroom challenges and assist teachers in adjusting their teaching strategies. The external negative teaching weight is used to quantify the impact of teachers' negative emotions on teaching effectiveness. The calculation formula of the negative teaching emotion intensity index is shown in formula (14): （14）； in: Indicates the i The external negative teaching weight in the time segment is used to quantify the i The influence of external factors on the negative learning emotion intensity index at each time point is calculated by weighting formula (15): （15）； in: Indicates the i The situation of class schedule pressure at a certain time point, Indicates the i Insufficient preparation of teaching content at a certain time point, Indicates the i The external interference at a certain time point, , and are the weight coefficients of these three external influencing factors respectively; S043: Analyze classroom interaction using the classroom interaction dimension analysis unit; The teacher-student emotional synchronization index is used to analyze the matching degree of the emotional state of teachers and students, reflecting the consistency of emotional communication and the quality of interaction in the classroom. The teacher's emotional sequence is multiplied by the synchronization weight and the correlation is calculated with the student's emotional sequence to obtain the teacher-student emotional synchronization index. The calculation formula of the teacher-student emotional synchronization index is shown in formula (16): （16）； in: Indicates that the teacher t The sentiment value at a time point; Indicates that students t The sentiment value at a time point; Indicates the t The weight of a time point reflects the importance of that moment; represents the weighted average of teachers’ sentiment values, represents the weighted average of students’ sentiment values; The interaction frequency index is used to count the intensity of teacher-student interaction within a unit of time and to evaluate whether the classroom activity design effectively promotes student participation. The calculation formula of the interaction frequency index is shown in formula (17): （17）； in: Indicates the t The number of teacher-student interactions at each time point; The interaction quality index is based on the proportion of interactions marked with positive emotions, quantifying the positive effects of classroom interaction. The calculation formula of the interaction quality index is shown in formula (18): （18）； in: Indicates the t The number of teacher-student interactions based on positive emotions at each time point.

2. The evaluation method according to claim 1, wherein: The specific steps of S06 are: The classroom engagement index reflects the activeness and emotional synchronization of teacher-student interaction through the teacher-student emotional synchronization index, interaction frequency index, and interaction quality index, and measures the quality of classroom participation. The calculation formula of the classroom engagement index is shown in formula (19): （19）； in: , and are the weights of the teacher-student emotional synchronization index, interaction frequency index, and interaction quality index respectively; The teaching efficiency score measures the achievement of teaching objectives and the learning effect driven by emotions through the positive learning emotion intensity index, the negative learning emotion intensity index, the positive teaching emotion intensity index, and the negative teaching emotion intensity index. The calculation formula of the teaching efficiency score is shown in formula (20): （20）； in: and are the weights of the positive learning emotion intensity index and the negative learning emotion intensity index, and the positive teaching emotion intensity index and the negative teaching emotion intensity index respectively; The teaching motivation index measures the driving effect of students' learning motivation and teachers' emotions in the classroom through the learning motivation index, positive learning emotion intensity index and positive teaching emotion intensity index. The calculation formula of the teaching motivation index is shown in formula (21): （21）； in: , and are the weights of the learning motivation index, the positive learning emotion intensity index, and the positive teaching emotion intensity index, respectively; The interaction-driven emotional stability index combines the interaction frequency index and the interaction quality index to evaluate the inhibitory effect of classroom interaction on the emotional fluctuation index and quantify the ability of teacher-student interaction to stabilize the emotional state of the classroom during teaching activities. The calculation formula of the interaction-driven emotional stability index is shown in formula (22): （22）； in: , and are the weights of interaction frequency index, interaction quality index and mood fluctuation index respectively.

Images