Intelligent classroom-based real-time learning condition feedback method and system for intelligent classroom

Abstract

This invention provides a real-time learning feedback method and system for smart classrooms based on smart classrooms, including: Step S1: collecting and preprocessing data; Step S2: selecting learning characteristic variables based on the data to obtain training variables; Step S3: training a model based on the training variables; Step S4: determining whether the fitting result of the model's input data and output data is greater than a preset value; if the result is yes, deploying the model; if the result is no, repeating step S3; Step S5: outputting the student's academic performance through the model and providing feedback to the target personnel. This invention integrates learning display and the classroom central control system into a learning feedback screen. Through the learning feedback screen display scheme, the interface for teachers to view learning data and initiate teaching activities is separated from the traditional podium screen, avoiding cumbersome screen switching and protecting the privacy of teaching information.

Description

Technical Field

[0001] This invention belongs to the field of internet education technology, specifically relating to a method and system for real-time learning feedback in smart classrooms based on smart classrooms. Background Technology

[0002] With the rapid development of information technology, smart classrooms have become an important part of the modern teaching environment. By integrating smart terminals, network devices, and various teaching software, smart classrooms achieve efficient information interaction and real-time data processing.

[0003] One key technology in smart classrooms is the real-time learning feedback system. Utilizing instant feedback systems (IRS) such as HiTeach, and combining features like instant question-and-answer, real-time priority assignment, and intelligent student selection, it enables students to provide immediate feedback to teachers. This system uses students' hands-on instant feedback devices to statistically analyze and display the responses as visual charts in real time, allowing teachers to quickly understand students' learning progress and make targeted adjustments to their teaching. Furthermore, the learning analysis system also enriches the sources of learning data by linking with the academic affairs management system and student affairs system to obtain personal information from teachers and students.

[0004] While real-time learning feedback systems in smart classrooms offer numerous conveniences, they still face several challenges in practical application. First, teachers need to cumbersomely switch between multiple systems, reducing teaching efficiency and potentially exposing student learning information during screen transitions. Second, traditional real-time learning feedback only provides statistical analysis of ad-hoc teaching activities during class, lacking a comprehensive analysis of the entire teaching process and failing to accurately grasp students' mastery of knowledge points. Finally, data security issues cannot be ignored; large amounts of student personal information and teaching data require proper protection.

[0005] Given the challenges faced by real-time learning feedback systems in smart classrooms during practical applications, the urgency of finding solutions is self-evident. On the one hand, improving teaching efficiency through various technological means and display solutions reduces teachers' psychological fear of operating equipment, and making the systems simple and easy to use increases teachers' willingness to engage in in-class interaction. On the other hand, enriching teaching content and methods, increasing the analysis of learning progress throughout the entire process, comprehensively analyzing students' dynamics and learning mastery, and strengthening data security measures are all key to improving the effectiveness of smart classroom applications.

[0006] Patent document CN111768128A discloses an intelligent digital ideological and political education teaching evaluation system, and also discloses a lectern. The lectern includes an evaluation system, which in turn includes a processor. This invention, through the coordinated arrangement of the evaluation system, processor, data storage module, comparison module, in-class test score comparison, attendance rate comparison, answer accuracy rate comparison, fingerprint attendance device, display module, operation module, alarm module, data storage card, monitor, operation panel, light control buttons, buzzer, red flashlight, and green flashlight, achieves the advantage of allowing ideological and political education teachers to scientifically assess students' classroom learning effectiveness, the results of their absorbing classroom knowledge, and attendance. This enables teachers to scientifically and effectively evaluate students' classroom learning effectiveness, understand students' absorbing classroom knowledge, and know their attendance rates. However, this solution cannot improve the effective utilization of the lectern screen display space, cannot enhance the data collection dimensions, and cannot achieve one-stop data display.

[0007] Therefore, developing a more user-friendly, efficient, and secure real-time learning feedback system for smart classrooms is of great significance for promoting the informatization of education and improving teaching quality. This problem urgently needs to be solved. Summary of the Invention

[0008] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a method and system for real-time learning feedback in smart classrooms based on smart classrooms.

[0009] A real-time learning feedback method for smart classrooms based on smart classrooms, provided by the present invention, includes:

[0010] Step S1: Collect and preprocess data;

[0011] Step S2: Based on the data, select the learning characteristics variables to obtain the training variables;

[0012] Step S3: Train the model based on the training variables;

[0013] Step S4: Determine whether the fitting result of the input data and output data of the model is greater than a preset value; if the result is yes, then deploy the model; if the result is no, then repeat step S3.

[0014] Step S5: Output the student's academic performance through the model and provide feedback to the target personnel.

[0015] Preferably, in step S1, the preprocessing involves deleting duplicate or invalid data; the data is retrieved through an API gateway; the data includes basic data and learning characteristic variables.

[0016] The basic data includes: class size, course information, classroom environment, attendance and interaction data; the student learning characteristic variables include: student grade, age, gender, exam scores, attendance, classroom head-up rate, classroom participation, classroom interaction rate, and discussion and posting activity.

[0017] Preferably, in step S1, the data includes real-time data and non-real-time data;

[0018] The non-real-time data is synchronized daily according to a preset time.

[0019] The real-time data is generated using the RabbitMQ message queue mechanism.

[0020] In step S2, the training variables are obtained based on the correlation coefficient;

[0021] The mathematical expression for the correlation coefficient is:

[0022]

[0023] Where r is the correlation coefficient; x i Let y represent the i-th learning characteristic variable. i Let i be the i-th academic performance variable; The mean of the student learning characteristic variables; The mean of the academic performance variable;

[0024] Step S2 includes:

[0025] Step S2.1: Select a feature quantity of the learning characteristic variable and its corresponding academic performance quantity, and calculate the correlation coefficient;

[0026] Step S2.2: Determine whether the correlation coefficient is greater than or equal to a preset threshold; if the result is yes, select one feature quantity and academic performance quantity as training variables, and repeat step S2.1 until all training variables are selected; if the result is no, do not process, and repeat step S2.1 until all training variables are selected.

[0027] In step S3, the model is a large AI model.

[0028] Preferably, step S4 includes:

[0029] Step S4.1: Label the learning situation feature variables as input data, input them into the AI ​​large model, return a feature coefficient corresponding to each learning situation feature, and obtain the learning situation result output data;

[0030] Step S4.2: Fit the learning performance value corresponding to the input data with the output data to obtain the fitting result;

[0031] Step S4.3: Determine whether the fitting result is greater than a preset value; if the result is yes, deploy the AI ​​large model; if the result is no, repeat step S3; the preset value is 90%.

[0032] In step S5, the academic performance is expressed mathematically as follows:

[0033] y = β0 + β1x1 + β2x2 + ... + β n x n +∈

[0034] Where y represents academic performance, β0 is the value of y when all inputs are zero, x1 is the first learning performance feature, x2 is the second learning performance feature, and x... n Let β1 be the nth learning characteristic; β2 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β3 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β4 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β5 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β6 be the characteristic coefficient corresponding to the first learning characteristic. n The coefficients are the characteristic coefficients corresponding to the nth learning characteristic after being determined by the model; ∈ represents the error term.

[0035] Specifically, β1......β n The following are the characteristic coefficients corresponding to each learning characteristic variable after being judged by the model; the characteristic coefficients represent the change in academic performance when a certain learning characteristic increases by one unit.

[0036] According to the present invention, a real-time learning feedback system for smart classrooms based on smart classrooms includes:

[0037] Module M1: Acquires and preprocesses data;

[0038] Module M2: Based on the data, select learning characteristic variables to obtain training variables;

[0039] Module M3: Train the model based on the aforementioned training variables;

[0040] Module M4: Determines whether the fitting result of the input data and output data of the model is greater than a preset value; if the result is yes, the model is deployed; if the result is no, module M3 is triggered again.

[0041] Module M5: Outputs students' academic performance through the model and provides feedback to the target personnel.

[0042] Preferably, in module M1, the preprocessing involves deleting duplicate or invalid data; the data is retrieved through an API gateway; the data includes basic data and learning characteristic variables.

[0043] The basic data includes: class size, course information, classroom environment, attendance and interaction data; the student learning characteristic variables include: student grade, age, gender, exam scores, attendance, classroom head-up rate, classroom participation, classroom interaction rate, and discussion and posting activity.

[0044] Preferably, in module M1, the data includes real-time data and non-real-time data;

[0045] The non-real-time data is synchronized daily according to a preset time.

[0046] The real-time data is generated using the RabbitMQ message queue mechanism.

[0047] In module M2, the training variables are obtained based on correlation coefficients;

[0048] The mathematical expression for the correlation coefficient is:

[0049]

[0050] Where r is the correlation coefficient; x i Let y represent the i-th learning characteristic variable. i Let i be the i-th academic performance variable; The mean of the student learning characteristic variables; The mean of the academic performance variable;

[0051] The module M2 includes:

[0052] Module M2.1: Select a feature quantity of the learning characteristic variable and its corresponding academic performance quantity, and calculate the correlation coefficient;

[0053] Module M2.2: Determine whether the correlation coefficient is greater than or equal to a preset threshold; if the result is yes, select one feature quantity and academic performance quantity as training variables, and re-trigger module M2.1 until all training variables are selected; if the result is no, do not process, and re-trigger module M2.1 until all training variables are selected.

[0054] In module M3, the model is a large AI model.

[0055] Preferably, the module M4 includes:

[0056] Module M4.1: Label the learning characteristics variables as input data, input them into the AI ​​large model, and obtain the feature coefficient β. n and output data;

[0057] Module M4.2: Fits the learning variable values ​​corresponding to the input data with the output data to obtain the fitting result;

[0058] Module M4.3: Determines whether the fitting result is greater than a preset value; if the result is yes, deploys the AI ​​large model; if the result is no, re-triggers module M3; the preset value is 90%.

[0059] In module M5, the academic performance is expressed mathematically as follows:

[0060] y = β0 + β1x1 + β2x2 + ... + β n x n +∈

[0061] Where y represents academic performance, β0 is the value of y when all inputs are zero, x1 is the first learning performance feature, x2 is the second learning performance feature, and x... n Let β1 be the nth learning characteristic; β2 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β3 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β4 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β5 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β6 be the characteristic coefficient corresponding to the first learning characteristic. n The coefficients are the characteristic coefficients corresponding to the nth learning characteristic after being determined by the model; ∈ represents the error term.

[0062] Specifically, β1......β n The following are the characteristic coefficients corresponding to each learning characteristic variable after being judged by the model; the characteristic coefficients represent the change in academic performance when a certain learning characteristic increases by one unit.

[0063] Preferably, it also includes a central control screen; the central control screen can display the basic data and the academic performance.

[0064] According to the present invention, a computer-readable storage medium storing a computer program is provided, wherein when the computer program is executed by a processor, the steps of the real-time learning feedback method for smart classrooms based on smart classrooms are implemented.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] 1. The data platform and pipeline technology provided by this invention aggregates multi-dimensional learning data, presents it in real time, and comprehensively integrates the key functions of various teaching business modules. In addition to aggregating various teaching data in the classroom, it also adds various data on students' life and learning outside the classroom, improving the dimensions of data aggregation and realizing one-stop data display, reducing the cumbersome switching between multiple systems for teachers. In other words, this invention helps teachers accurately grasp various information about students throughout the learning process, understand students' learning dynamics, and give special attention to students who are lagging behind academically, effectively improving classroom teaching efficiency.

[0067] 2. This invention utilizes a loosely coupled solution to allow the learning feedback system and the central control system to operate on the same screen. Compared to traditional methods where teachers rely on the podium screen for interaction, attendance tracking, or access to learning data, impacting screen display space and potentially compromising privacy, this invention integrates the learning data display and classroom central control system within the learning feedback screen. This solution separates the interface for teachers to view learning data and initiate teaching activities from the traditional podium screen, avoiding cumbersome screen switching and protecting the privacy of teaching information. Specifically, compared to traditional display solutions, this invention increases the effective utilization rate of the podium screen display space to 99%, enabling independent operation of initiating teaching activities, collecting learning data, and managing the central control platform. Furthermore, the loosely coupled solution enhances the flexibility and security of the teaching process.

[0068] 3. This invention uses a private AI big data model of academic performance to mine deep learning data. Traditional learning data analysis often records mostly objective data. This invention, based on objective data, uses Tencent's Hunyuan big data model combined with sample data to develop a private AI big data model of academic performance. This model scores students' academic performance, represents past learning status and predicts future results, and serves as a reference for teachers to adjust their teaching strategies in a timely manner.

[0069] 4. This invention can further improve the optimization parameters based on teachers' actual teaching feedback, so as to conform to teaching practice and realize the continuous updating and iteration of the academic evaluation model. Attached Figure Description

[0070] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0071] Figure 1 This invention provides a schematic diagram of the data source and processing flow.

[0072] Figure 2 A schematic diagram of the interface of the smart classroom real-time learning feedback system provided by the present invention;

[0073] Figure 3 A schematic diagram of the learning progress display interface provided by this invention;

[0074] Figure 4 This is a schematic diagram of the screen position provided by the present invention. Detailed Implementation

[0075] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0076] This invention is based on a smart classroom, integrating central control functions and data from multiple teaching business systems to provide teachers with real-time feedback on student learning and data references for precise classroom teaching.

[0077] A real-time learning feedback system for smart classrooms based on smart classrooms, provided by the present invention, includes the following functions:

[0078] System Login. Before class, teachers log in using unified authentication to access the class page. The system automatically associates with the courses the teacher teaches, retrieving course, selection, and scheduling information from the teaching data base, eliminating the need for teachers to manually create courses. Basic course information, such as course name, class time, and location, will be displayed on the student feedback screen.

[0079] Real-time electronic class roster display. This module displays the list of all students in the class currently taught by the teacher. Teachers initiate attendance checks on the LMS learning system, and students scan a QR code. The student's attendance information is displayed in real-time on the learning feedback screen, including the online student list, allowing teachers to understand real-time class attendance data.

[0080] This module displays students' recent academic performance in real time. It shows students' daily learning data and academic performance in the teaching class. It retrieves students' full-process learning data from the teaching data base and displays objective statistical data on daily learning activities such as homework, discussions, quizzes, and access to course resources on the learning feedback screen. A private AI model for academic performance scores students and displays information on outstanding or lagging students in a preset proportion. The list of students with good academic performance is available for teachers to view on the learning feedback screen while protecting student privacy.

[0081] Real-time display of classroom interaction statistics. Teachers initiate interactive quizzes in class, and students participate by scanning a QR code. This module retrieves real-time classroom interaction data through a teaching data platform and displays the quiz statistics in the form of visual charts, including: the number of students who answered, the distribution of answers, accuracy rate, etc. It further analyzes the correlation between the quiz results and the achievement of knowledge points using a private domain AI model, similar to the model training steps mentioned above, allowing teachers to understand students' mastery of knowledge points in a timely manner.

[0082] The system displays student messages in real time. During the lesson, if a student sends a message to the teacher, it can be displayed on the secondary screen. The teacher can view these messages during class to understand students' questions and adjust the teaching content accordingly.

[0083] This module displays the live stream in real time, allowing teachers to confirm whether the live stream is running normally. It also displays live stream information to keep teachers informed about the online live stream status.

[0084] According to the present invention, a real-time learning feedback method for smart classrooms based on smart classrooms includes the following information flow process in its software system:

[0085] Step 1: The teaching data base provides many application integration data call APIs to meet different business needs. The smart classroom real-time learning feedback system retrieves various data for learning analysis through the API gateway.

[0086] Step 2: For basic data display, it is directly displayed on the central control screen;

[0087] The basic data includes: class size, course information, classroom environment, attendance and interaction data;

[0088] Step 3: Input attendance, online time on the learning platform, number of interactions, homework submission, homework scores and evaluations, test results, and student engagement during class into the private domain AI model for academic performance. Based on the algorithm described above, the model will output a score for the student's academic performance.

[0089] Step 4: Feedback on academic performance scores to teachers via the central control screen to facilitate precise teaching during class.

[0090] Specifically, in step 1, the data sources are divided into non-real-time and real-time data;

[0091] For non-real-time data, a unified data synchronization is performed during off-peak hours each day. In other words, data synchronization is performed uniformly according to a preset time.

[0092] For real-time data, in order to meet the requirements of high concurrency and low latency, the RabbitMQ message queue mechanism is adopted. This mechanism supports clustering and high scalability, achieves high availability, and ensures the stable operation of the message queue service.

[0093] Both types of data are stored in a data pool and used as an API gateway for invocation.

[0094] Specifically, in step 2, for static data display, i.e., basic data display, the smart classroom real-time learning feedback system retrieves data from the API gateway and displays it on the central control screen. After setting the data refresh rate, the latest data will be automatically displayed.

[0095] Specifically, in step 3, objective data from teaching is mainly used, such as the number of students who answered, the distribution of answers, accuracy rate, homework submissions, test results, and project performance. This data is then comprehensively analyzed using a private domain AI model to assess students' academic performance and provide a score, which serves as a reference for teaching. This result is only a suggestion and is not used as the basis for the student's final assessment.

[0096] Specifically, in step 4, the feedback method includes: the central control display screen showing the student's academic performance score during class and the display of the learning situation analysis module on the LMS learning platform.

[0097] A training method for the academic performance private domain AI model provided by the present invention includes:

[0098] Step A1: Collect and preprocess the data;

[0099] Step A2: Based on the data obtained in Step 1, select the learning characteristics variables;

[0100] Step A3: Train a large AI model based on the aforementioned learning characteristics variables;

[0101] Step A4: Determine whether the fitting result of the input data and output data of the AI ​​large model is greater than the target value; if the result is yes, then deploy the AI ​​large model; if the result is no, then repeat step A3; the target value, i.e., the preset value, is 90%;

[0102] In step A1, the API gateway originating from the teaching data base performs data preprocessing to delete duplicate or invalid data; in other words, the preprocessed data is the deletion of duplicate or invalid data.

[0103] In step A2, the student learning characteristics variables are selected based on Bloom's Taxonomy of Educational Objectives and constructivist theory; the student learning characteristics variables include: student grade, age, gender, test scores, quiz scores, attendance, classroom engagement, classroom participation, classroom interaction rate, and discussion / posting activity.

[0104] The learning characteristics variables are identified through correlation coefficients;

[0105] The mathematical expression for the correlation coefficient is:

[0106]

[0107] Where r is the correlation coefficient, greater than 0 indicates a positive correlation, used to identify which learning variables are significantly correlated with academic performance; x i y represents a certain learning characteristic variable. i For academic performance variables; The mean of the student learning characteristic variables; The mean of the academic performance variable;

[0108] As shown in Table 1, the following data is available:

[0109] student Attendance rate x Exam score y Zhang San 80％ 90 Li Si 100％ 88 Wang Wu 60％ 70

[0110] Table 1

[0111] Based on the calculation results, and substituting them into the above formula, the correlation coefficient r is obtained, thus analyzing the correlation coefficient between attendance rate and exam scores. The calculation of other features follows the same principle; finally, the features with significant influence are selected.

[0112] In step A3, model training and optimization are performed. Specifically, the Tencent Hunyuan large model is used by default. Based on the large model, the aforementioned feature data is used for training and data alignment is performed: the learning characteristic variables are used as input, and after the large model makes its judgment, the feature coefficient β corresponding to each learning characteristic variable is given. n This forms a private domain model, or large model, used only for the analysis and evaluation of teaching and learning within the school to predict student performance.

[0113] In step A4, the newly labeled learning data is input into the private domain big data model. The big data model returns results, and the labeled data results are fitted with the results given by the model. When the accuracy of the academic performance prediction reaches more than 90%, the model can be adopted. In other words, it is determined whether the fitting result of the input data and output data of the AI ​​big data model is greater than the target value. If the result is yes, the AI ​​big data model is deployed; if the result is no, step A3 is repeated. The target value is 90%.

[0114] The mathematical expression for the academic performance of the aforementioned learning characteristic variables is:

[0115] y = β0 + β1x1 + β2x2 + ... + β n x n +∈

[0116] Where y represents academic performance, β0 is the value of y when all inputs are zero, x1 is the first learning performance feature, x2 is the second learning performance feature, and x... n Let β1 be the nth learning characteristic; β2 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β3 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β4 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β5 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β6 be the characteristic coefficient corresponding to the first learning characteristic. n The coefficients are the characteristic coefficients corresponding to the nth learning characteristic after being determined by the model; ∈ represents the error term.

[0117] Specifically, β1......β nThe following are the characteristic coefficients corresponding to each learning characteristic variable determined by the model; each characteristic coefficient represents the change in academic performance when a certain learning characteristic increases by one unit. The error term includes random error and other factors.

[0118]

[0119] Wherein, `result` represents the fitting result. Specifically, if the fitting result is greater than or equal to 90%, the model is used; if the fitting result is less than 90%, the model is not used. 模型预测 Indicates the model prediction result; y 标注值 This indicates the labeled value.

[0120] The specific steps for building a private domain AI model for academic performance:

[0121] First, collect and preprocess sample data: data preprocessing is performed from the API gateway of the teaching data base to remove duplicate or invalid data;

[0122] Second, the selection of student characteristics variables based on Bloom's Taxonomy of Educational Objectives and constructivist theory: the candidate variables are student grade, age, gender, test scores, quiz scores, attendance, classroom head-up rate, classroom participation, classroom interaction rate, and discussion and posting activity.

[0123]

[0124] Where r is the correlation coefficient, and a value greater than 0 indicates a positive correlation, x i y represents a certain learning characteristic variable. i For academic performance variables, the calculated value r is used to identify which learning variables are significantly related to academic performance.

[0125] Based on the calculation results, and substituting them into the above formula, the correlation coefficient r is obtained, thus analyzing the correlation coefficient between attendance rate and exam scores. The calculation of other features follows the same principle; finally, the features with significant influence are selected.

[0126] Third, AI large-scale model selection, model training and optimization: The Tencent Hunyuan large-scale model is used by default for processing. Based on this large model, the aforementioned feature data is used for training and data alignment: Learning characteristic variables are used as input, and after the large model makes its judgment, the feature coefficient β corresponding to each learning characteristic variable is given. n This forms a private domain model, or large model, used only for the analysis and evaluation of teaching and learning within the school to predict student performance.

[0127] Fourth, model evaluation: Using the new labeled learning data as input, the results of the labeled data are fitted with the results given by the model. When the accuracy of the academic performance prediction reaches more than 90%, the model can be used.

[0128] y = β0 + β1x1 + β2x2 + ... + β n x n +∈

[0129] Where y represents academic performance, β0 is the value of y when all inputs are zero, x1 is the first learning performance feature, x2 is the second learning performance feature, and x... n Let β1 be the nth learning characteristic; β2 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β3 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β4 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β5 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β6 be the characteristic coefficient corresponding to the first learning characteristic. n The coefficients are the characteristic coefficients corresponding to the nth learning characteristic after being determined by the model; ∈ represents the error term.

[0130] Specifically, β1......β n The following are the characteristic coefficients corresponding to each learning characteristic variable after being judged by the model; the characteristic coefficients represent the change in academic performance when a certain learning characteristic increases by one unit.

[0131] The error term includes random error and other factors. These other factors include individual differences in a particular learning variable, such as not submitting homework or having very low attendance, but still achieving higher grades than everyone else.

[0132] if:

[0133]

[0134] That is, this model is adopted.

[0135] Fifth, model deployment: localized deployment, providing physical machine support for up to 64GB of GPU video memory.

[0136] Sixth, continuous monitoring and updates: In the formal production application environment, based on the feedback from teachers and administrators, corrected parameters are generated as new training data for the private domain large model.

[0137] The present invention also provides a real-time learning feedback system for smart classrooms based on smart classrooms. The real-time learning feedback system for smart classrooms based on smart classrooms can be implemented by executing the process steps of the real-time learning feedback method for smart classrooms based on smart classrooms. That is, those skilled in the art can understand the real-time learning feedback method for smart classrooms based on smart classrooms as a preferred embodiment of the real-time learning feedback system for smart classrooms based on smart classrooms.

[0138] According to the present invention, a real-time learning feedback system for smart classrooms based on smart classrooms includes:

[0139] Module M1: Acquires and preprocesses data;

[0140] Module M2: Based on the data, select learning characteristic variables to obtain training variables;

[0141] Module M3: Train the model based on the aforementioned training variables;

[0142] Module M4: Determines whether the fitting result of the input data and output data of the model is greater than a preset value; if the result is yes, the model is deployed; if the result is no, module M3 is triggered again.

[0143] Module M5: Outputs students' academic performance through the model and provides feedback to the target personnel.

[0144] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0145] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A method for real-time learning feedback in a smart classroom based on a smart classroom, characterized in that, include: Step S1: Collect and preprocess data; Step S2: Based on the data, select the learning characteristic variables to obtain... Training variables; Step S3: Train the model based on the training variables; Step S4: Determine whether the fitting result of the input data and output data of the model is greater than a preset value; if the result is yes, then deploy the model; if the result is no, then repeat step S3. Step S5: Output the student's academic performance through the model and provide feedback to the target personnel.

2. The real-time learning feedback method for smart classrooms based on smart classrooms according to claim 1, characterized in that, In step S1, the preprocessing involves deleting duplicate or invalid data. The data is retrieved through an API gateway; the data includes basic data and student learning characteristic variables. The basic data includes: class size, course information, classroom environment, attendance and interaction data; the student learning characteristic variables include: student grade, age, gender, exam scores, attendance, classroom head-up rate, classroom participation, classroom interaction rate, and discussion and posting activity.

3. The real-time learning feedback method for smart classrooms based on smart classrooms according to claim 1, characterized in that, In step S1, the data includes real-time data and non-real-time data; The non-real-time data is synchronized daily according to a preset time. The real-time data is generated using the RabbitMQ message queue mechanism. In step S2, the training variables are obtained based on the correlation coefficient; The mathematical expression for the correlation coefficient is: Where r is the correlation coefficient; x i Let y represent the i-th learning characteristic variable. i Let i be the i-th academic performance variable; The mean of the characteristic variables of student learning; The mean of the academic performance variable; Step S2 includes: Step S2.1: Select a feature quantity of the learning characteristic variable and the corresponding academic performance quantity, and calculate the correlation coefficient; Step S2.2: Determine whether the correlation coefficient is greater than or equal to a preset threshold; if the result is yes, select one feature quantity and its corresponding academic performance quantity as training variables, and repeat step S2.1 until all training variables are selected; if the result is no, do not process, and repeat step S2.1 until all training variables are selected. In step S3, the model is a large AI model.

4. The real-time learning feedback method for smart classrooms based on smart classrooms according to claim 3, characterized in that, Step S4 includes: Step S4.1: Label the learning characteristic variables as input data, input them into the AI ​​large model, and obtain the feature coefficient β. n and output data; Step S4.2: Fit the academic performance data corresponding to the input data with the output data to obtain the fitting result; Step S4.3: Determine whether the fitting result is greater than a preset value; if the result is yes, deploy the AI ​​large model; if the result is no, repeat step S3; the preset value is 90%. In step S5, the academic performance is expressed mathematically as follows: y=β0+β1x1+β2x2+…+β n x n +∈ Where y represents academic performance, β0 is the value of y when all inputs are zero, x1 is the first learning performance feature, x2 is the second learning performance feature, and x... n Let β1 be the nth learning characteristic; β2 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β3 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β4 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β5 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β6 be the characteristic coefficient corresponding to the first learning characteristic. n The characteristic coefficients corresponding to the nth learning characteristic after being judged by the model; ∈ represents the error term; the model is the AI ​​large model.

5. A real-time learning feedback system for smart classrooms based on smart classrooms, characterized in that, include: Module M1: Acquires and preprocesses data; Module M2: Based on the data, select learning characteristic variables to obtain training variables; Module M3: Train the model based on the aforementioned training variables; Module M4: Determines whether the fitting result of the input data and output data of the model is greater than a preset value; if the result is yes, the model is deployed; if the result is no, module M3 is triggered again. Module M5: Outputs students' academic performance through the model and provides feedback to the target personnel.

6. The real-time learning feedback system for smart classrooms based on smart classrooms according to claim 5, characterized in that, In module M1, the preprocessing involves deleting duplicate or invalid data. The data is retrieved through an API gateway; the data includes basic data and student learning characteristic variables. The basic data includes: class size, course information, classroom environment, attendance and interaction data; the student learning characteristic variables include: student grade, age, gender, exam scores, attendance, classroom head-up rate, classroom participation, classroom interaction rate, and discussion and posting activity.

7. The real-time learning feedback system for smart classrooms based on smart classrooms according to claim 5, characterized in that, In module M1, the data includes real-time data and non-real-time data; The non-real-time data is synchronized daily according to a preset time. The real-time data is generated using the RabbitMQ message queue mechanism. In module M2, the training variables are obtained based on correlation coefficients; The mathematical expression for the correlation coefficient is: Where r is the correlation coefficient; x i Let y represent the i-th learning characteristic variable. i Let i be the i-th academic performance variable; The mean of the characteristic variables of student learning; The mean of the academic performance variable; The module M2 includes: Module M2.1: Select a feature quantity of the learning situation characteristic variable and the corresponding learning situation quantity, and calculate the correlation coefficient; Module M2.2: Determine whether the correlation coefficient is greater than or equal to a preset threshold; if the result is yes, select one feature quantity and the corresponding academic performance quantity as training variables, and re-trigger module M2.1 until all training variables are selected; if the result is no, do not process, and re-trigger module M2.1 until all training variables are selected. In module M3, the model is a large AI model.

8. The real-time learning feedback system for smart classrooms based on smart classrooms according to claim 7, characterized in that, The module M4 includes: Module M4.1: Label the learning characteristics variables as input data, input them into the AI ​​large model, and obtain the feature coefficient β. n and output data; Module M4.2: Fits the input data and the output data to obtain the fitting result; Module M4.3: Determines whether the fitting result is greater than a preset value; if the result is yes, deploys the AI ​​large model; if the result is no, re-triggers module M3; the preset value is 90%. In module M5, the academic performance is expressed mathematically as follows: y=β0+β1x1+β2x2+…+β n x n +∈ Where y represents academic performance, β0 is the value of y when all inputs are zero, x1 is the first learning performance feature, x2 is the second learning performance feature, and x... n Let β1 be the nth learning characteristic; β2 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β3 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β4 be the characteristic coefficient corresponding to the first learning characteristic after model judgment; β5 be the characteristic coefficient corresponding to the second learning characteristic after model judgment; β6 be the characteristic coefficient corresponding to the first learning characteristic. n The characteristic coefficients corresponding to the nth learning characteristic after being judged by the model; ∈ represents the error term; the model is the AI ​​large model.

9. The real-time learning feedback system for smart classrooms based on smart classrooms according to claim 8, characterized in that, It also includes a central control screen; the central control screen can display the basic data and the academic performance.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the real-time learning feedback method for smart classrooms based on smart classrooms as described in any one of claims 1 to 4.

Images