Teaching system based on embedded artificial intelligence practical training platform

Abstract

The invention discloses a teaching system based on an embedded artificial intelligence practical training platform, and relates to the technical field of embedded artificial intelligence. The system comprises a processing unit, a database, a timing module, a switch and n learning terminals, the output end of the processing unit is connected with the database and the input end of the switch, the output end of the timing module is connected with the input end of the processing unit, and the output end of the database is connected with the input end of the switch. The output end of the switch is connected with the input end of the timing module, a port of the switch is in data connection with a port of each learning terminal, and the processing unit endows each learning terminal with an independent serial number. Through student answering time and correct rate feedback, the playing speed of the learning terminal is automatically adjusted, so that the teaching content can well adapt to the learning progress of all students.

Description

Technical Field

[0001] The present invention relates to the field of embedded artificial intelligence technology, and in particular to a teaching system based on an embedded artificial intelligence training platform. Background Art

[0002] As the performance of embedded processors continues to improve, including processor computing speed, storage capacity, parallel processing capabilities, etc., embedded processors at this stage can support complex neural network calculations, providing a hardware foundation for the operation of artificial intelligence algorithms. Implanting artificial intelligence technology into embedded systems can improve data collection and preprocessing capabilities.

[0003] Traditional teaching systems cannot well integrate artificial intelligence technology into the teaching system of embedded platforms, do not highlight learning points, and cannot systematically learn embedded artificial intelligence courses. At the same time, each student has different mastery of knowledge points, and unified teaching cannot adapt to the learning progress of all students. Therefore, highlighting the embedded artificial intelligence learning points and intelligently adapting the learning progress of all students are technical problems that technical personnel in this field need to solve. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a teaching system based on an embedded artificial intelligence training platform, which solves the problems raised in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a teaching system based on an embedded artificial intelligence training platform, comprising a processing unit, a database, a timing module, a switch and n learning terminals, wherein the output end of the processing unit is connected to the input end of the database and the switch respectively, the output end of the timing module is connected to the input end of the processing unit, the output end of the database is connected to the input end of the switch, the output end of the switch is connected to the input end of the timing module, and a port of the switch establishes a data connection with a port of each learning terminal; The processing unit assigns an independent number to each learning terminal, and the independent number is used to distinguish each learning terminal. Different learning terminals have different playback speeds, which can adapt to the learning speed of each student. The database transmits the video courseware to the switch, and the switch distributes the embedded artificial intelligence video courseware to each learning terminal. Each student watches the embedded artificial intelligence video courseware through the learning terminal. The student manually adjusts the playback speed s1 and playback time p1 of the video courseware through the learning terminal. When the playback speed s1 of the video courseware is different from the default speed s0, the processing unit binds the manually adjusted playback speed s1 to the learning terminal number as the designated playback speed s1 of the student. 1. The processing unit sends instructions to the database, and the database randomly sends the questions in the question bank to the learning terminal for students to answer. Each learning terminal transmits its own playback speed s1, playback time p1 and answer content a1 to the processing unit through the switch. When the student starts to input the answer content a1 through the learning terminal, the timing module counts the answer time t1 of each learning terminal and forwards it to the processing unit. The answer time t1 is the answer time of a single question. The processing unit counts the number of playbacks q1 and the cumulative playback time t2 of the playback time p1. The processing unit counts the cumulative answer time t3 of each learning terminal. The cumulative answer time t3 is the answer time accumulated by the answer time t1 of all questions; The processing unit pre-scans the video courseware and question bank in the database. The processing unit uses the Whisper speech recognition engine to extract the voice text in the video courseware. The Whisper speech recognition engine is an open source technology that converts the voice in the video courseware into text information. The processing unit performs a word co-occurrence analysis on the extracted voice text and the text content in the question bank to obtain keywords. The processing unit marks the keywords as learning points and records the timestamp p2 when the learning point appears in the video courseware. The processing unit presets the playback speed s2 and the playback speed s3 of the video courseware, the playback speed s2 is lower than the playback speed s1 and the playback speed s3, and the playback speed s3 is within the range of the playback speed s1±20%. The processing unit sends instructions to each learning terminal. When the learning terminal encounters a learning point corresponding to the timestamp p2 while playing the video courseware, the processing unit reduces the playback speed of the video courseware from s1 to s2 and increases the audio output volume of the video courseware, so as to help students deepen their learning memory. The processing unit automatically adjusts the playback speed s3 of each learning terminal to adapt to the different learning speeds of each student.

[0006] Furthermore, the word co-occurrence analysis method includes the following steps: The processing unit marks the text content in the question bank as tk, and the processing unit marks the text information converted from speech as yy; The text is preprocessed. The processing unit deletes spaces, punctuation marks and special symbols in the text content tk and the text information yy, and only retains the Chinese characters. The processing unit sorts and classifies the text content tk and the text information yy according to the pronunciation of pinyin. If the text content tk and the text information yy of the same category have the same pronunciation but inconsistent text contents, the processing unit uses the text content tk as a reference standard to correct the text information yy. Word segmentation: The processing unit uses the jieba word segmentation tool to semantically segment the text information converted from speech and the text in the question bank to obtain several phrases. For example, "SSH remote connection" can be semantically segmented into the phrase "SHH" and the phrase "remote connection". The jieba word segmentation tool is an existing Python Chinese word segmentation tool open source technology; The processing unit presets a co-occurrence window, the size of which is 10 Chinese characters. The processing unit slides the co-occurrence window from the starting point to the end in the text content tk and the text information yy. Each time the co-occurrence window slides, the processing unit counts the number of co-occurrences of each phrase. The processing unit marks the phrases that co-occur more than twice as related phrases. The processing unit marks the associated phrases with the same content in the text content tk and the text information yy as keywords.

[0007] Furthermore, the playback speed s2 is obtained by analyzing in the following way: The processing unit counts the number of times the learning points in the video courseware appear repeatedly, and arranges the number of times the learning points in the video courseware appear repeatedly in descending order to obtain a first sequence, and the processing unit records the ranking q2 of each learning point in the first sequence; The processing unit presets a time window, and the size of the time window is , where kj is the duration of each course in the video courseware, akj is the average duration of each course in the video courseware, and the duration kj of each course is different, and the corresponding time window size of each course is also different. In order to avoid a course duration that is too short and causes the time window to be too small, the processing unit presets a basic duration of 10 seconds as a supplement; The processing unit uses the time window to scan the video courseware from the beginning to the end in sequence. Every time the time window moves backward by 1 second on the time axis of the video courseware, the processing unit scans the video courseware content within the time window once. The processing unit counts the number of repetitions of the learning point in the time window, arranges the number of repetitions of the learning point in the time window in descending order to obtain a second sequence, and records the ranking q3 of each learning point in the second sequence; Processing units are grouped according to formulas Calculate the playback speed s2, where w1 and w2 are weight coefficients, and the specific values ​​are determined according to the number of sections in the video courseware, and w1 and w2 satisfy the condition w1+w2=1, and f is the adjustment ratio of the playback speed s2.

[0008] Furthermore, the playback speed s3 is obtained by analyzing in the following way: The processing unit classifies the playback times q1 and the playback statistical time t2. When the learning terminal uses the default speed s0 for playback, when the student uses the learning terminal for playback, the processing unit marks the playback times q1 of the default speed s0 as the default times q2, and marks the playback cumulative time t2 of the default speed s0 as the default time t4. The student uses the learning terminal to manually adjust to the playback speed s1. The playback speed s1 is not a fixed value and is determined by the student's manual adjustment. The playback speed s1 of each learning terminal is different, and the speed of the playback speed s1 is lower or higher than the default speed s0. When the learning terminal uses the playback speed s1 for playback, when the student uses the learning terminal for playback, the processing unit marks the playback times q1 of the playback speed s1 as the adjustment times q3, and marks the playback cumulative time t2 of the playback speed s1 as the adjustment time t5. The processing unit presets a qualified threshold value, and the qualified threshold value includes a speed qualified rate c2 and a correct qualified rate c3. The speed qualified rate c2 is, and the correct qualified rate is 90%; When s1-s0 is a positive value, it means that the student thinks that the default speed s0 is too slow, and increases the playback speed of the learning terminal. After the first section of the video course is played, the processing unit executes the first review process; When the first review process is executed, the processing unit retrieves the corresponding questions from the question bank and sends them to the learning terminal for the student to answer. The timing module counts the cumulative answering time t3 of the learning terminal and transmits it to the processing unit. The processing unit retrieves the standard answer of the question bank from the database. The student inputs the answer content a1 of the question bank through the learning terminal. The processing unit compares the answer content a1 with the standard answer and calculates the correctness c1. The processing unit executes the first speed adjustment process; When the value of the accuracy rate c1 in the first review process is lower than the accuracy pass rate c3, it indicates that the student has not fully mastered the learning point, and the processing unit reduces the original playback speed s1 by 10% to replay the content of the video courseware. After the playback of the content of the video courseware is completed, the first review process is re-executed. When the value of the accuracy rate c1 in the first review process is higher than or equal to the accuracy pass rate c3, the learning terminal plays the next content of the video courseware; When the first speed regulation process is executed, the processing unit uses the formula Calculate the playback speed s3 of the learning terminal, where T is the time taken by the video courseware to play the entire content, w3, w4 and w5 are weight coefficients, and the specific values ​​are determined according to the number of questions in the question bank, and w3, w4 and w5 meet the condition w3+w4=w5. The processing unit sends the playback speed s3 to the corresponding learning terminal, and the learning terminal replaces the original playback speed s1 with the playback speed s3 when playing the next section of the video courseware; When s1-s0 is a negative value, it means that the student thinks that the default speed s0 is too fast and slows down the playback speed of the learning terminal. After the first section of the video course is played, the processing unit executes the second review process; When the second review process is executed, the processing unit retrieves the corresponding questions from the question bank and sends them to the learning terminal for the student to answer. The timing module counts the cumulative answering time t6 of the learning terminal and transmits it to the processing unit. The processing unit retrieves the standard answer of the question bank from the database. The student inputs the answer content a2 of the question bank through the learning terminal. The processing unit compares the answer content a1 with the standard answer and calculates the correctness c4. The processing unit executes the second speed adjustment process. When the value of the accuracy rate c2 in the second review process is lower than the accuracy pass rate c3, it indicates that the student has not fully mastered the learning point, and the processing unit reduces the original playback speed s1 by 5% to replay the content of the video courseware. After the playback of the content of the video courseware is completed, the second review process is re-executed. When the value of the accuracy rate c4 in the second review process is higher than or equal to the accuracy pass rate c3, the learning terminal plays the next content of the video courseware; When the second speed control process is executed, the processing unit uses the formula Calculate the playback speed s3 of the learning terminal, where T is the time taken by the video courseware to play the entire content, w3, w4 and w5 are weight coefficients, and the specific values ​​are determined according to the number of questions in the question bank, and w3, w4 and w5 meet the condition w3+w4=w5. The processing unit sends the playback speed s3 to the corresponding learning terminal, and the learning terminal replaces the original playback speed s1 with the playback speed s3 when playing the next section of the video courseware; If the student does not manually adjust the playback speed s1 of the learning terminal subsequently, the playback speed s3 remains constant. If the student manually modifies the playback speed s1 of the learning terminal subsequently, the processing unit executes the first review process before the next section of the video courseware is played, and the processing unit recalculates the value of the playback speed s3.

[0009] Furthermore, the database stores computer instructions or programs for execution by the processing unit, and the database also stores video courseware and question banks of embedded artificial intelligence.

[0010] Furthermore, the learning terminal is composed of a display, a keyboard, a mouse and an audio output. The display is used to display the video courseware and question bank of the embedded artificial intelligence, and the students input the answers to the question bank through the keyboard and mouse.

[0011] The present invention has the following beneficial effects: 1. Extracting learning points from courseware through word co-occurrence analysis can help students deepen their understanding of the learning content, integrate artificial intelligence technology into the teaching system of the embedded platform, and enable students to systematically learn embedded artificial intelligence courses.

[0012] 2. Based on the students' answering time and accuracy feedback, the playback speed of the learning terminal is automatically adjusted, so that the teaching content can be well adapted to the learning progress of all students and improve the students' learning efficiency.

[0013] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0015] Figure 1 This is a block diagram of the teaching system based on the embedded artificial intelligence training platform of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] See also Figure 1 The present invention provides a technical solution: a teaching system based on an embedded artificial intelligence training platform, comprising a processing unit, a database, a timing module, a switch and n learning terminals, wherein the output end of the processing unit is connected to the input end of the database and the switch respectively, the output end of the timing module is connected to the input end of the processing unit, the output end of the database is connected to the input end of the switch, the output end of the switch is connected to the input end of the timing module, and a port of the switch is connected to a port of each learning terminal to establish a data connection; The processing unit assigns an independent number to each learning terminal, and the independent number is used to distinguish each learning terminal. Different learning terminals have different playback speeds, which can adapt to the learning speed of each student. The database transmits the video courseware to the switch, and the switch distributes the embedded artificial intelligence video courseware to each learning terminal. Each student watches the embedded artificial intelligence video courseware through the learning terminal. The student manually adjusts the playback speed s1 and playback time p1 of the video courseware through the learning terminal. When the playback speed s1 of the video courseware is different from the default speed s0, the processing unit binds the manually adjusted playback speed s1 to the learning terminal number as the designated playback speed s1 of the student. The processing unit sends instructions to the database, and the database randomly sends the questions in the question bank to the learning terminal for display and answering by the students. Each learning terminal transmits its own playback speed s1, playback time p1 and answer content a1 to the processing unit through the switch. When the student starts to input the answer content a1 through the learning terminal, the timing module counts the answer time t1 of each learning terminal and forwards it to the processing unit. The answer time t1 is the answer time of a single question. The processing unit counts the number of playbacks q1 and the cumulative playback time t2 of the playback time p1. The processing unit counts the cumulative answer time t3 of each learning terminal. The cumulative answer time t3 is the answer time accumulated by the answer time t1 of all questions. The video courseware content is mainly divided into the following 8 teaching sections: 1. Understand the basic concepts and development process of embedded artificial intelligence 2. Understand the basic concepts of machine learning and deep learning; 3. Master the basic operations of the deep learning framework TensorFlow; 4. Understand the implementation principles of deep learning neural networks; 5. Master the implementation methods of linear regression and nonlinear regression of deep learning neural networks; 6. Understand the basic concepts and implementation principles of deep learning convolutional neural networks, fully connected neural networks, and recurrent neural networks; 7. Master the deep learning convolutional neural network image classification model training and recurrent neural network text classification and text generation model.

[0018] 8. Master the construction of embedded artificial intelligence application development environment and the use of software.

[0019] The practical training content of the question bank in the database is mainly divided into the following 7 training contents: 1. Introduction to embedded artificial intelligence; 2. Configuration of embedded artificial intelligence development environment; 3. Basics of embedded artificial intelligence operations; 4. Basics of image processing; 5. Fully connected neural network; 6. Convolutional Neural Networks; 7. Recurrent Neural Network.

[0020] The practical training content of the question bank in the database is mainly divided into the following practical training cases: 1. Supermarket smart retail system; 2. Epidemic prevention and control security inspection system; 3. Facial emotion recognition atmosphere light control system; 4. Passenger flow statistics analysis system; 5. Human body posture and action recognition system; 6. License plate recognition gate control system; 7. Face recognition intelligent access control system; 8. OCR character recognition system; 9. Intelligent voice interactive dialogue system.

[0021] The processing unit pre-scans the video courseware and question bank in the database. The processing unit uses the Whisper speech recognition engine to extract the voice text in the video courseware. The Whisper speech recognition engine is an open source technology that converts the voice in the video courseware into text information. The processing unit performs a word co-occurrence analysis on the extracted voice text and the text content in the question bank to obtain keywords. The processing unit marks the keywords as learning points and records the timestamp p2 when the learning point appears in the video courseware. The processing unit presets the playback speed s2 and the playback speed s3 of the video courseware, the playback speed s2 is lower than the playback speed s1 and the playback speed s3, and the playback speed s3 is within the range of the playback speed s1±20%. The processing unit sends instructions to each learning terminal. When the learning terminal encounters a learning point corresponding to the timestamp p2 while playing the video courseware, the processing unit reduces the playback speed of the video courseware from s1 to s2 and increases the audio output volume of the video courseware, so as to help students deepen their learning memory. The processing unit automatically adjusts the playback speed s3 of each learning terminal to adapt to the different learning speeds of each student.

[0022] The word co-occurrence analysis method includes the following steps: The processing unit marks the text content in the question bank as tk, and the processing unit marks the text information converted from speech as yy; The text is preprocessed. The processing unit deletes spaces, punctuation marks and special symbols in the text content tk and the text information yy, and only retains the Chinese characters. The processing unit sorts and classifies the text content tk and the text information yy according to the pronunciation of pinyin. If the text content tk and the text information yy of the same category have the same pronunciation but inconsistent text contents, the processing unit uses the text content tk as a reference standard to correct the text information yy. Word segmentation: The processing unit uses the jieba word segmentation tool to semantically segment the text information converted from speech and the text in the question bank to obtain several phrases. For example, "SSH remote connection" can be semantically segmented into the phrase "SHH" and the phrase "remote connection". The jieba word segmentation tool is an existing Python Chinese word segmentation tool open source technology; The processing unit presets a co-occurrence window, the size of which is 10 Chinese characters. The processing unit slides the co-occurrence window from the starting point to the end in the text content tk and the text information yy. Each time the co-occurrence window slides, the processing unit counts the number of co-occurrences of each phrase. The processing unit marks the phrases that co-occur more than twice as related phrases. The processing unit marks the associated phrases with the same content in the text content tk and the text information yy as keywords.

[0023] The playback speed s2 is obtained by analyzing in the following way: The processing unit counts the number of times the learning points in the video courseware appear repeatedly, and arranges the number of times the learning points in the video courseware appear repeatedly in descending order to obtain a first sequence, and the processing unit records the ranking q2 of each learning point in the first sequence; The processing unit presets a time window, and the size of the time window is , where kj is the duration of each course in the video courseware, akj is the average duration of each course in the video courseware, and the duration kj of each course is different, and the corresponding time window size of each course is also different. In order to avoid a course duration that is too short and causes the time window to be too small, the processing unit presets a basic duration of 10 seconds as a supplement; The processing unit uses the time window to scan the video courseware from the beginning to the end in sequence. Every time the time window moves backward by 1 second on the time axis of the video courseware, the processing unit scans the video courseware content within the time window once. The processing unit counts the number of repetitions of the learning point in the time window, arranges the number of repetitions of the learning point in the time window in descending order to obtain a second sequence, and records the ranking q3 of each learning point in the second sequence; Processing units are grouped according to formulas Calculate the playback speed s2, where w1 and w2 are weight coefficients, and the specific values ​​are determined according to the number of sections in the video courseware, and w1 and w2 satisfy the condition w1+w2=1, and f is the adjustment ratio of the playback speed s2.

[0024] The playback speed s3 is obtained by analyzing in the following way: The processing unit classifies the playback times q1 and the playback statistical time t2. When the learning terminal uses the default speed s0 for playback, when the student uses the learning terminal for playback, the processing unit marks the playback times q1 of the default speed s0 as the default times q2, and marks the playback cumulative time t2 of the default speed s0 as the default time t4. The student uses the learning terminal to manually adjust to the playback speed s1. The playback speed s1 is not a fixed value and is determined by the student's manual adjustment. The playback speed s1 of each learning terminal is different, and the speed of the playback speed s1 is lower or higher than the default speed s0. When the learning terminal uses the playback speed s1 for playback, when the student uses the learning terminal for playback, the processing unit marks the playback times q1 of the playback speed s1 as the adjustment times q3, and marks the playback cumulative time t2 of the playback speed s1 as the adjustment time t5. The processing unit presets a qualified threshold value, and the qualified threshold value includes a speed qualified rate c2 and a correct qualified rate c3. The speed qualified rate c2 is, and the correct qualified rate is 90%; When s1-s0 is a positive value, it means that the student thinks that the default speed s0 is too slow, and increases the playback speed of the learning terminal. After the first section of the video course is played, the processing unit executes the first review process; When the first review process is executed, the processing unit retrieves the corresponding questions from the question bank and sends them to the learning terminal for the student to answer. The timing module counts the cumulative answering time t3 of the learning terminal and transmits it to the processing unit. The processing unit retrieves the standard answer of the question bank from the database. The student inputs the answer content a1 of the question bank through the learning terminal. The processing unit compares the answer content a1 with the standard answer and calculates the correctness c1. The processing unit executes the first speed adjustment process; When the value of the accuracy rate c1 in the first review process is lower than the accuracy pass rate c3, it indicates that the student has not fully mastered the learning point, and the processing unit reduces the original playback speed s1 by 10% to replay the content of the video courseware. After the playback of the content of the video courseware is completed, the first review process is re-executed. When the value of the accuracy rate c1 in the first review process is higher than or equal to the accuracy pass rate c3, the learning terminal plays the next content of the video courseware; When the first speed regulation process is executed, the processing unit uses the formula Calculate the playback speed s3 of the learning terminal, where T is the time taken by the video courseware to play the entire content, w3, w4 and w5 are weight coefficients, and the specific values ​​are determined according to the number of questions in the question bank, and w3, w4 and w5 meet the condition w3+w4=w5. The processing unit sends the playback speed s3 to the corresponding learning terminal, and the learning terminal replaces the original playback speed s1 with the playback speed s3 when playing the next section of the video courseware; When s1-s0 is a negative value, it means that the student thinks that the default speed s0 is too fast and slows down the playback speed of the learning terminal. After the first section of the video course is played, the processing unit executes the second review process; When the second review process is executed, the processing unit retrieves the corresponding questions from the question bank and sends them to the learning terminal for the student to answer. The timing module counts the cumulative answering time t6 of the learning terminal and transmits it to the processing unit. The processing unit retrieves the standard answer of the question bank from the database. The student inputs the answer content a2 of the question bank through the learning terminal. The processing unit compares the answer content a1 with the standard answer and calculates the correctness c4. The processing unit executes the second speed adjustment process. When the value of the accuracy rate c2 in the second review process is lower than the accuracy pass rate c3, it indicates that the student has not fully mastered the learning point, and the processing unit reduces the original playback speed s1 by 5% to replay the content of the video courseware. After the playback of the content of the video courseware is completed, the second review process is re-executed. When the value of the accuracy rate c4 in the second review process is higher than or equal to the accuracy pass rate c3, the learning terminal plays the next content of the video courseware; When the second speed control process is executed, the processing unit uses the formula Calculate the playback speed s3 of the learning terminal, where T is the time taken by the video courseware to play the entire content, w3, w4 and w5 are weight coefficients, and the specific values ​​are determined according to the number of questions in the question bank, and w3, w4 and w5 meet the condition w3+w4=w5. The processing unit sends the playback speed s3 to the corresponding learning terminal, and the learning terminal replaces the original playback speed s1 with the playback speed s3 when playing the next section of the video courseware; If the student does not manually adjust the playback speed s1 of the learning terminal subsequently, the playback speed s3 remains constant. If the student manually modifies the playback speed s1 of the learning terminal subsequently, the processing unit executes the first review process before the next section of the video courseware is played, and the processing unit recalculates the value of the playback speed s3.

[0025] The database stores computer instructions or programs for execution by the processing unit, and also stores video courseware and question banks of embedded artificial intelligence.

[0026] Among them, the learning terminal consists of a display, keyboard, mouse and audio output. The display is used to display embedded artificial intelligence video courseware and question banks, and students input the answers to the question banks through the keyboard and mouse.

[0027] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A teaching system based on an embedded artificial intelligence training platform, comprising a processing unit, a database, a timing module, a switch and n learning terminals, wherein the output end of the processing unit is connected to the input end of the database and the switch respectively, the output end of the timing module is connected to the input end of the processing unit, the output end of the database is connected to the input end of the switch, the output end of the switch is connected to the input end of the timing module, and a port of the switch establishes a data connection with a port of each learning terminal; characterized in that: The switch distributes the video courseware with embedded artificial intelligence to each learning terminal. The learner manually adjusts the playback speed s1 and playback time p1 of the video courseware through the learning terminal. The processing unit sends instructions to the database. Each learning terminal transmits its own playback speed s1, playback time p1 and answer content a1 to the processing unit. When the learner starts to input the answer content a1, the timing module counts the answer time t1 of each learning terminal and forwards it to the processing unit, counts the number of playbacks q1 and the cumulative playback time t2 of the playback time p1, and counts the cumulative answer time t3 of each learning terminal. The processing unit uses the speech recognition engine to extract the voice text in the video courseware, performs word co-occurrence analysis on the voice text and the text content in the question bank to obtain keywords, marks the keywords as learning points, and records the timestamp p2 of the video courseware when the learning point appears; The processing unit presets the playback speed s2 and the playback speed s3 of the video courseware, and sends instructions to each learning terminal. When the learning terminal encounters a learning point corresponding to the timestamp p2 while playing the video courseware, the learning terminal reduces the playback speed of the video courseware from s1 to s2 and increases the audio output volume of the video courseware; The processing unit automatically adjusts the playback speed s3 of each learning terminal.

2. The teaching system based on the embedded artificial intelligence training platform according to claim 1 is characterized in that: The word co-occurrence analysis method includes the following steps: The processing unit marks the text content in the question bank as tk and the text information converted from speech as yy; Preprocessing the text, the processing unit deletes spaces, punctuation marks and special symbols in the text content tk and the text information yy, sorts and classifies the text content tk and the text information yy according to the pronunciation of pinyin, and if the text content of the text content tk and the text information yy of the same category is inconsistent, the text content tk is used as a reference standard to correct the text information yy; Word segmentation: The processing unit performs semantic segmentation on the text information converted from speech and the text in the question bank to obtain several phrases; The processing unit presets a co-occurrence window, and slides the co-occurrence window from the starting point to the end in the text content tk and the text information yy. Each time the co-occurrence window slides, the number of co-occurrences of each phrase is counted, and phrases that co-occur more than twice are marked as related phrases. The processing unit marks the associated phrases with the same content in the text content tk and the text information yy as keywords.

3. The teaching system based on the embedded artificial intelligence training platform according to claim 1 is characterized in that: The playback speed s2 is obtained by analyzing in the following way: The processing unit counts the number of times the learning points in the video courseware appear repeatedly, arranges the number of times the learning points in the video courseware appear repeatedly in descending order to obtain a first sequence, and records the ranking q2 of each learning point in the first sequence; The processing unit presets a time window, and the size of the time window is , where kj is the duration of each lesson in the video courseware, akj is the average duration of each lesson in the video courseware, and the duration kj of each lesson is different; The processing unit uses the time window to scan the video courseware from the beginning to the end in sequence. Every time the time window moves backward by 1 second on the time axis of the video courseware, the video courseware content within the time window is scanned once; The processing unit counts the number of repetitions of the learning point in the time window, arranges the number of repetitions of the learning point in the time window in descending order to obtain a second sequence, and records the ranking q3 of each learning point in the second sequence; Processing units are grouped according to formulas Calculate the playback speed s2, where w1 and w2 are weight coefficients, w1 and w2 satisfy the condition w1+w2=1, and f is the adjustment ratio of the playback speed s2.

4. The teaching system based on the embedded artificial intelligence training platform according to claim 1 is characterized in that: The playback speed s3 is obtained by analyzing in the following way: The processing unit classifies the playback times q1 and the playback statistical time t2. When the learning terminal uses the default speed s0 for playback, when the student uses the learning terminal for playback, the playback times q1 is marked as the default times q2, and the playback cumulative time t2 is marked as the default time t4. When the student uses the learning terminal to manually adjust the playback speed to s1, when the learning terminal uses the playback speed s1 for playback, when the student uses the learning terminal for playback, the playback times q1 is marked as the adjustment times q3, and the playback cumulative time t2 is marked as the adjustment time t5. A qualified threshold is preset, and the qualified threshold includes a speed qualified rate c2 and a correct qualified rate c3; When s1-s0 is a positive value, after the first section of the video course is played, the processing unit executes the first review process; When the first review process is executed, the corresponding questions are retrieved from the question bank and sent to the learning terminal for the student to answer. The timing module counts the cumulative answering time t3 of the learning terminal and transmits it to the processing unit. The processing unit retrieves the standard answer of the question bank from the database. The student inputs the answer content a1 of the question bank through the learning terminal, compares the answer content a1 with the standard answer and calculates the correctness c1. The processing unit executes the first speed adjustment process; When the value of the accuracy rate c1 in the first review process is lower than the accuracy pass rate c3, the processing unit reduces the original playback speed s1 and replays the content of the section. After the content of the section is played, the first review process is re-executed. When the value of the accuracy rate c1 is higher than or equal to the accuracy pass rate c3, the learning terminal plays the next section of the content. When the first speed control process is executed, the formula Calculate the playback speed s3 of the learning terminal, where T is the time taken by the video courseware to play the entire content, w3, w4 and w5 are weight coefficients, w3, w4 and w5 meet the condition w3+w4=w5, and send the playback speed s3 to the corresponding learning terminal. When the learning terminal plays the next content, it will replace the original playback speed s1 with the playback speed s3 for playback; When s1-s0 is a negative value, after the first section of the video course is played, the processing unit executes the second review process; When the second review process is executed, the corresponding questions are retrieved from the question bank and sent to the learning terminal for the student to answer. The timing module counts the cumulative answering time t6 of the learning terminal and transmits it to the processing unit. The processing unit retrieves the standard answer of the question bank from the database. The student inputs the answer content a2 of the question bank through the learning terminal, compares the answer content a1 with the standard answer and calculates the correct rate c4. The processing unit executes the second speed adjustment process; When the value of the accuracy rate c2 in the second review process is lower than the accuracy pass rate c3, the processing unit reduces the original playback speed s1 and replays the content of the section. After the content of the section is played, the second review process is re-executed. When the value of the accuracy rate c4 is higher than or equal to the accuracy pass rate c3, the learning terminal plays the next section of content. When the second speed control process is executed, the formula Calculate the playback speed s3 of the learning terminal, where T is the time taken by the video courseware to play the entire content, w3, w4 and w5 are weight coefficients, w3, w4 and w5 meet the condition w3+w4=w5, and send the playback speed s3 to the corresponding learning terminal. When the learning terminal plays the next content, it will replace the original playback speed s1 with the playback speed s3 for playback; If the student does not manually adjust the playback speed s1 of the learning terminal subsequently, the playback speed s3 remains constant. If the student manually modifies the playback speed s1 of the learning terminal subsequently, the first review process is executed before the next section of the video courseware is played, and the processing unit recalculates the value of the playback speed s3.

5. The teaching system based on the embedded artificial intelligence training platform according to claim 1 is characterized in that: The database stores computer instructions or programs for execution by the processing unit, and the database also stores video courseware and a question bank.

6. The teaching system based on the embedded artificial intelligence training platform according to claim 1 is characterized in that: The learning terminal is composed of a display, a keyboard, a mouse and an audio output. The display is used to display video courseware and question banks, and the learners input the answers to the question banks through the keyboard and mouse.

Images