An electronic information engineering teaching interactive system and method based on artificial intelligence

Abstract

This invention provides an artificial intelligence-based interactive teaching system and method for electronic information engineering. The system includes: a data acquisition module connected to a teaching experimental platform to collect multi-dimensional experimental data from students in hardware circuit experiments in real time; an intelligent diagnostic module based on a deep learning model to analyze the current experimental status corresponding to the multi-dimensional experimental data, compare it with standard statuses in an expert knowledge base, and automatically locate and output the fault type and location; and a teacher-side management module to receive the multi-dimensional experimental data and analysis results output by the data acquisition module and the intelligent diagnostic module forwarded by the network communication module, display student experimental progress, error distribution heatmap, and fault details in real time on the teacher's terminal, and issue warnings or push lists of students requiring special attention according to preset rules. This invention solves the problems of teachers being unable to monitor student operation status in real time and the lag in fault feedback in existing teaching aids, effectively improving experimental efficiency and teaching effectiveness.

Description

Technical Field

[0001] This invention relates to the field of intelligent teaching system technology, and in particular to an interactive teaching system and method for electronic information engineering based on artificial intelligence. Background Technology

[0002] As a highly practical discipline, Electronic Information Engineering relies heavily on experimental courses (such as circuit analysis, signals and systems, and embedded system design) to cultivate students' hands-on skills and ability to solve complex engineering problems.

[0003] In current experimental teaching, students usually need to use hardware devices such as breadboards, resistors, capacitors, oscilloscopes, signal generators, and logic analyzers to build and debug complex circuits. However, due to the complexity of hardware connections and the large number of parameter settings, students are prone to problems such as incorrect circuit connections, damaged components, and abnormal signal measurements during the experiment.

[0004] Currently, hardware malfunctions and operational errors in experiments are mainly addressed through the following methods: first, teachers observe and correct them; second, students raise their hands for help; and third, students consult the experimental manual or search online. However, these methods have significant limitations. The most critical technical issue is that teachers cannot monitor each student's operational status and circuit setup in real time and comprehensively, leading to a serious lag in fault detection and feedback. Students are thus bogged down in a long-term, aimless trial-and-error process, severely hindering experimental efficiency and teaching effectiveness.

[0005] The specific manifestations of the aforementioned core problems are as follows: First, a teacher needs to guide dozens of students at the same time, making it difficult to detect abnormalities at each experimental station in a timely manner; second, students lack effective self-troubleshooting tools and spend a lot of time on problems such as cold solder joints, incorrect resistor values, and improper selection of oscilloscope probes; third, teachers cannot keep track of each student's mastery of knowledge points in real time and can only adopt a uniform pace and uniform explanation model, making it difficult to achieve personalized tutoring. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides an interactive teaching system and method for electronic information engineering based on artificial intelligence. It solves the problem that existing teaching aids cannot monitor students' operation status and circuit construction in real time, resulting in delayed fault feedback and severely restricting experimental efficiency and teaching effectiveness.

[0007] According to an embodiment of the present invention, an interactive teaching system for electronic information engineering based on artificial intelligence is proposed, comprising: The data acquisition module is connected to at least one teaching experimental platform and is used to collect multi-dimensional experimental data in real time when students conduct hardware circuit experiments on the teaching experimental platform. The multi-dimensional experimental data includes at least the circuit physical connection topology diagram, circuit electrical signal parameters, and student operation interaction data. The intelligent diagnostic module is communicatively connected to the data acquisition module. It is used to analyze the current experimental state based on the multidimensional experimental data output by the data acquisition module and a pre-trained deep learning model. It is also used to compare the current experimental state with the standard expected state in the built-in expert knowledge base, and automatically locate and output the fault type and location when the detected deviation exceeds a preset threshold. The teacher-side management module is connected to the data acquisition module and the intelligent diagnosis module via a network communication module. It is used to receive multidimensional experimental data output by the data acquisition module and analysis results output by the intelligent diagnosis module, which are forwarded by the network communication module. It is also used to display the experimental progress of all students, error distribution heatmap, and specific fault details of each student on the teacher terminal in real time, and to issue warnings or push a list of students who need special attention according to preset rules based on the specific fault details of the students.

[0008] According to another embodiment of the present invention, an interactive teaching method for electronic information engineering based on artificial intelligence is proposed, comprising the following steps: The system collects multi-dimensional experimental data in real time when students conduct hardware circuit experiments on the teaching experimental platform. The multi-dimensional experimental data includes at least the circuit physical connection topology diagram, circuit electrical signal parameters, and student operation interaction data. The multidimensional experimental data is input into a pre-trained deep learning model to analyze the current experimental state. The current experimental state is compared with the standard expected state in the built-in expert knowledge base. When the deviation exceeds a preset threshold, the fault type and location are automatically located and output. Based on the multidimensional experimental data and analysis results, the teacher's terminal displays the experimental progress of all students, the error distribution heatmap, and the specific fault details of each student in real time. Based on the specific fault details of the students, the teacher issues warnings or pushes a list of students who need special attention according to preset rules.

[0009] Compared with the prior art, the present invention has the following beneficial effects: The data acquisition module is connected to the teaching experimental platform to collect multi-dimensional experimental data from students conducting hardware circuit experiments on the platform in real time. Compared with the black box problem and data silo problem in traditional experimental teaching, the data acquisition module transforms the originally difficult-to-quantify and dynamic experimental process into a storable, analyzable, and traceable digital data stream by synchronously collecting and correlating three types of experimental data: circuit topology, electrical signals, and operational behavior. This achieves full-process digitalization of the experimental process and transparent monitoring of students' experimental activities.

[0010] The intelligent diagnostic module uses a pre-trained deep learning model to comprehensively analyze the multi-dimensional experimental data transmitted by the data acquisition module, intelligently determine the current experimental state, and compare the current experimental state with the standard expected state in the built-in expert knowledge base. When a deviation (such as wiring error or improper parameter setting) is detected and exceeds a preset threshold, the module automatically locates and outputs the type and specific location of the fault. Compared with the serious lag in fault feedback in traditional experimental teaching, this module can respond and locate the fault immediately once a student makes a mistake during operation, greatly improving experimental efficiency and preventing students from wasting a lot of time on the wrong path.

[0011] The teacher-side management module uses an intuitive graphical interface to analyze and present massive amounts of data from the data acquisition and intelligent diagnostic modules. It can automatically issue warnings based on preset rules according to student fault details, or push lists of students requiring special attention to the teacher. Compared to traditional experimental teaching where teachers cannot understand the overall student situation in real time, the teacher-side management module provides a global, visualized teaching data dashboard, allowing teachers to grasp the entire class's teaching and experimental situation at a glance and provide differentiated instruction based on students' weaknesses, thereby significantly improving teaching effectiveness. Attached Figure Description

[0012] Figure 1 This is a control principle diagram of an interactive teaching system for electronic information engineering based on artificial intelligence, according to Embodiment 1 of the present invention.

[0013] Figure 2 This is a control principle diagram of an interactive teaching system for electronic information engineering based on artificial intelligence, according to Embodiment 2 of the present invention.

[0014] Figure 3 This is a flowchart illustrating the steps of an interactive teaching method for electronic information engineering based on artificial intelligence, as described in Embodiment 3 of the present invention.

[0015] In the above attached diagram: 1. Data acquisition module; 2. Intelligent diagnosis module; 3. Teacher-side management module; 4. Teaching experimental platform; 5. Network communication module; 6. Intelligent interactive guidance module. Detailed Implementation

[0016] The technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0017] Example 1 like Figure 1 As shown in the figure, this invention proposes an interactive teaching system for electronic information engineering based on artificial intelligence, comprising: The data acquisition module 1 is connected to at least one teaching experimental platform 4 and is used to collect multi-dimensional experimental data in real time when students conduct hardware circuit experiments on the teaching experimental platform 4. The multi-dimensional experimental data includes at least the circuit physical connection topology diagram, circuit electrical signal parameters, and student operation interaction data. The intelligent diagnostic module 2 is communicatively connected to the data acquisition module 1. It is used to analyze the current experimental state based on the multidimensional experimental data output by the data acquisition module 1 and a pre-trained deep learning model. It is also used to compare the current experimental state with the standard expected state in the built-in expert knowledge base, and automatically locate and output the fault type and location when the detected deviation exceeds a preset threshold. The teacher-side management module 3 is connected to the data acquisition module 1 and the intelligent diagnosis module 2 via a network communication module 5. It is used to receive the multidimensional experimental data output by the data acquisition module 1 and the analysis results output by the intelligent diagnosis module 2, which are forwarded by the network communication module 5. It is also used to display the experimental progress of all students, the error distribution heatmap, and the specific fault details of each student on the teacher terminal in real time, and to issue warnings or push the list of students who need to be focused on according to preset rules based on the specific fault details of the students.

[0018] In this embodiment, the data acquisition module 1 connects to several teaching experimental platforms 4 to collect multi-dimensional experimental data in real time when students conduct hardware circuit experiments on the teaching experimental platforms 4. The multi-dimensional experimental data includes at least the circuit physical connection topology diagram identified and collected by the camera, the circuit electrical signal parameters (voltage, current, waveform) collected by the oscilloscope and multimeter, and the student operation interaction data collected by mouse clicks, keyboard acquisition, knob adjustment, panel operation, etc. Compared with the black box problem and data island problem in traditional experimental teaching, the data acquisition module 1 transforms the originally difficult-to-quantify and dynamic experimental process into a digital data stream that can be stored, analyzed, and traced by synchronously collecting and associating three types of experimental data: circuit topology, electrical signals, and operation behavior. This realizes the full-process digitalization of the experimental process and transparent monitoring of students' experimental activities.

[0019] The intelligent diagnostic module 2 receives multidimensional experimental data transmitted by the data acquisition module 1, and then uses a pre-trained deep learning model to comprehensively analyze the multidimensional experimental data transmitted by the data acquisition module 1. It intelligently judges the current experimental state and compares the current experimental state with the standard expected state in the built-in expert knowledge base. When a deviation (such as wiring error or improper parameter setting) is detected and exceeds the preset threshold, it automatically locates and outputs the type and specific location of the fault, realizing a complete closed loop from "data acquisition → feature extraction → model reasoning → decision output". Compared with the serious lag in fault feedback in traditional experimental teaching, it can respond and locate the fault immediately once the student makes a mistake during operation. It can shorten the fault root cause diagnosis time from several hours to minutes, significantly reduce the false alarm rate, greatly improve experimental efficiency, and avoid students wasting a lot of time on the wrong path.

[0020] The teacher-side management module 3 uses an intuitive graphical interface to analyze and present massive amounts of data from the data acquisition module 1 and the intelligent diagnostic module 2. It can automatically issue warnings based on preset rules according to student fault details, or push lists of students requiring special attention to the teacher. Compared to traditional experimental teaching where teachers cannot understand the overall student situation in real time, the teacher-side management module 3 provides a global, visualized teaching data dashboard, allowing teachers to grasp the entire class's teaching and experimental situation at a glance and provide differentiated instruction based on students' weaknesses, thereby significantly improving teaching effectiveness.

[0021] In summary, the AI-based interactive teaching system for electronic information engineering adopted in this invention has the advantages of faster response speed and higher diagnostic accuracy compared with existing technologies, as shown in the table below.

[0022]

[0023] Preferably, the deep learning model includes a first feature extraction model and a second feature extraction model. The first feature extraction model is used to process the physical connection topology diagram of the circuit captured by the camera to extract spatial structure features; the second feature extraction model is used to process time-series data composed of student operation interaction data to extract time-series change features.

[0024] More preferably, the first feature extraction model is a graph neural network, which is used to compare the physical circuit topology diagram actually connected by the student with the preset standard experimental circuit topology diagram and output the connection error node information.

[0025] Specifically, the Graph Neural Network (GNN) is a deep learning model specifically designed for processing non-Euclidean data (such as graph-structured data). Its core structure revolves around a graph (composed of nodes and edges) and learns the representation of nodes and the entire graph through a unique message passing mechanism. In this invention, the GNN accurately locates the specific nodes with incorrect connections by comparing the physical circuit topology diagram actually connected by the student with the preset standard experimental circuit topology diagram in real time. The specific steps include: (1) First, model the circuit topology graph: convert both the physical circuit and the standard circuit into a data structure (graph) that can be processed by the graph neural network. In this graph, nodes represent components (such as transistors, resistors, capacitors, diodes) or key connection points (such as pins, network nodes) in the circuit, and edges represent physical connections (wires) between components. (2) Secondly, feature extraction and embedding are performed: through the message passing mechanism, each node aggregates the information of its neighboring nodes to update its own feature representation. That is, each node is first given initial features, and then the information of its multi-hop neighbors is aggregated for each node through a multi-layer graph convolutional network (GCN). When the number of layers in the graph neural network is too deep, the features of all nodes tend to converge to the same value, which is the "oversmoothing" phenomenon. This will lose the distinguishability of the nodes. Therefore, the model design needs to balance the receptive field and the oversmoothing problem, such as through regularization or adaptive graph structure to alleviate it.

[0026] (3) Secondly, real-time comparison and error handling are performed: Since the computational load of directly performing precise matching on the entire large-scale circuit diagram (such as the VF2 algorithm) is huge and the time is unacceptable, the model usually adopts a two-stage strategy, in which: The first stage involves fuzzy matching (this stage is fast and can efficiently filter out a few candidate regions that may be similar to the standard circuit structure): The model traverses every key node in the actual physical circuit (such as the pins of a transistor). For each node, it extracts its k-hop neighborhood (i.e., a local subgraph). This local subgraph is compared with the corresponding region in the standard circuit (or the overall query circuit) (the comparison method is to compare their embedding vectors generated in the graph neural network). If the embedding vectors of the two subgraphs are "close" enough in the embedding space (i.e., they meet specific constraints, such as the embedding vector of the query graph being less than or equal to the embedding vector of the target graph in all dimensions), then they are marked as potential matching regions.

[0027] The second stage involves precise matching (error node location): For candidate regions filtered by fuzzy matching, precise subgraph isomorphism algorithms (such as VF2, VF3, Ullmann, etc.) are used for verification. That is, the algorithm attempts to find a one-to-one mapping for each node in the query graph (standard circuit) in the candidate subgraph (actual physical circuit) through depth-first search and backtracking. It checks whether the node types match and whether the connection relationships are completely consistent. When precise matching fails, the algorithm backtracks and records the first node or edge that cannot be matched. This position is the specific node where the student made a connection error. For example, in the standard circuit, node A should connect to B and C, but in the actual circuit, node A is connected to B and D. Then, when trying to find a mapping for C, it will fail, thus indicating "node C connection error" or "edge connected to node C is missing / incorrect".

[0028] The graph neural network combines fuzzy matching (fast filtering) with precise matching (precise positioning). While ensuring accuracy through precise matching, the fuzzy matching stage greatly narrows the scope of precise search, thereby significantly reducing time complexity and improving overall speed.

[0029] More preferably, the second feature extraction model is a temporal convolutional network, which is used to analyze the temporal data to identify abnormal operation processes in the student operation sequence.

[0030] Specifically, the Temporal Convolutional Network (TCN) is a convolutional neural network architecture used for time series modeling. Unlike traditional recurrent neural networks (RNN / LSTM), the Temporal Convolutional Network processes sequential data through clever convolution design. It treats a student's series of operations (such as clicking, inputting, browsing, etc.) as an "operation flow" (i.e., behavior sequence) with a time order. By learning the patterns and rules of normal operations, it can discover "abnormal" operations that do not conform to the expected patterns.

[0031] In this invention, the temporal convolutional network analyzes student operation sequences based on the temporal data, predicts their intentions, and identifies non-compliant operation processes, specifically including the following steps: (1) First, construct an analyzable operation flow. Since the original operation log data is fragmented, it needs to be transformed into structured time-series data that the model can understand. Therefore, it is necessary to collect historical operation data of students in specific tasks or experimental environments and integrate these operations that occur in chronological order (such as "click button A", "enter value X", "enter interface B") into an operation behavior sequence. In addition, in order to eliminate the differences in operation speed and different index value ranges among different students, the data needs to be standardized, specifically including: (a) Timing Alignment: Mapping each operation in a sequence of operations to a uniform time step or relative time interval; (b) Mean / variance standardization: Eliminate the differences in "water level" and "shaking amplitude" caused by different time periods or different students' individual operating habits, so that the model focuses on the pattern of operation itself rather than absolute values; (c) Feature encoding: Convert non-numerical operations (such as "click", "enter", "exit") into vectors that the model can process, for example, by using one-hot encoding or embedding techniques.

[0032] (2) Secondly, capture the temporal dependencies and intentions between operations, specifically including: (a) Multi-scale feature extraction: By using convolutional kernels of different sizes, the model can simultaneously extract features of different scales in the operation flow, such as "quick operations in a short period of time" (e.g., continuous mouse clicks) and "macro steps over a long period of time" (e.g., from "starting the experiment" to "finally drawing a conclusion"). This multi-scale capability enables the model to identify both point anomalies (e.g., a sudden click) and collective anomalies (e.g., the wrong order of the entire operation steps). (b) Intent decoding: The model identifies the intent to "plagiarize" from the student's sequence of actions (e.g., "login to the system → click on the assignment → find the answer → close the page"). This is usually achieved through an attention mechanism, where the model "focuses" on the key behaviors in the sequence that are most important to the current intent (e.g., "add to cart" has a much higher weight than "browse product B").

[0033] (c) Prediction and Reconstruction: The model has two working modes: one is the prediction mode, which predicts the most likely next operation that the student will perform based on the current number of steps. For example, after learning a large number of correct operations such as "mixing is usually done after measuring liquid", the model can predict that the next step for the student after measuring hydrochloric acid should be "pour hydrochloric acid into a beaker". The other is the reconstruction mode, which inputs the entire operation sequence into the model, and the model "compresses" and then "reconstructs" it, trying to output an "ideal operation sequence" that is as similar as possible to the input.

[0034] (3) Again, identify non-standard operating procedures. After the model learns the normal operating mode, it determines the abnormality by calculating the "difference". Among them, the abnormality detection method based on prediction error is: if the model predicts the next operation of the student in the prediction mode and there is a huge deviation between the prediction and the actual operation (prediction error), then the operation at this time point and the vicinity may be marked as abnormal. For example, the model predicts that the student should "turn off the instrument" next, but the student actually performs "adjust parameters". This deviation may indicate that the operation is not standardized. The abnormality detection method based on reconstruction error is: if the model reconstructs the "ideal operation sequence" in the reconstruction mode and there is a huge difference between the actual operation sequence of the student (reconstruction error), it means that the student's operation process does not conform to the conventional standard mode. The flexible detection method combined with time step is: the model can introduce a "time step" parameter to switch the detection type. When the time step is short, the model focuses on abnormalities, such as detecting "the intensity value of operation X is suddenly too high". When the time step is long, the model focuses on collective abnormalities, such as detecting "although each operation seems correct, the execution order of the three steps A, B and C is completely reversed, which is a process error".

[0035] In summary, the temporal convolutional network operation flow anomaly detection model effectively identifies non-standard behaviors in complex operation processes by transforming scattered operations into structured temporal data, capturing temporal dependencies and user intent using multi-scale convolution and attention mechanisms, and finally quantifying the degree of anomalies through prediction or reconstruction errors.

[0036] Preferably, the expert knowledge base is constructed by structurally modeling the circuit schematics, expected waveform parameters, and operating steps in the standard experimental manual; the intelligent diagnostic module 2 performs graph matching and / or parameter threshold comparison between the multidimensional experimental data and the expert knowledge base to determine the fault point.

[0037] Specifically, the expert knowledge base is constructed by structurally modeling the circuit schematics, expected waveform parameters, and operating procedures in the standard experimental manual. This transforms the originally static manual into a multi-layered, computable expert knowledge base containing a knowledge element base, an association base, an assertion base, and a rule base. This knowledge base is not only a collection of data but also an explicit expression of the tacit knowledge of domain experts.

[0038] More preferably, the intelligent diagnostic module 2 performs graph matching and parameter threshold comparison with the expert knowledge base to determine the fault point. The parameter threshold comparison compares the real-time collected "multi-dimensional experimental data" (such as voltage, current, temperature, etc.) with predefined assertions or rules in the expert knowledge base. Graph matching models the entire experimental circuit (including all component parameters, connection relationships, and waveform measurement points) into an "actual operating graph" and matches it with "standard circuit diagrams" or "typical fault diagrams" in the knowledge base. The combination of these two approaches forms a complete diagnostic logic: first, preliminary diagnosis is performed through parameter threshold comparison to quickly filter out obviously abnormal measurement points, forming a "suspected fault point list"; then, deep diagnosis is performed through graph matching to conduct correlation analysis on the "suspected fault point list," and graph matching algorithms (such as shortest path, subgraph isomorphism, graph neural networks, etc.) are used to accurately locate the root cause component and specific fault mode of the fault. This significantly improves the location efficiency while ensuring the accuracy of fault location.

[0039] Example 2 like Figure 2 As shown, based on Embodiment 1, the AI-based interactive teaching system for electronic information engineering further includes an intelligent interactive guidance module 6. The intelligent interactive guidance module 6 is communicatively connected to the intelligent diagnostic module 2 and is used to match corresponding solutions from a preset guidance knowledge base according to the fault type and location output by the intelligent diagnostic module 2, and provide students with real-time, step-by-step error correction or operation guidance information in the form of text, voice or augmented reality markers.

[0040] Furthermore, the intelligent interactive guidance module 6 is also used to dynamically adjust the level of detail in the guidance information based on the preset student's historical grades and current experimental performance; The level of detail in the dynamically adjusted guidance information specifically includes: providing step-by-step direct guidance for beginners and providing directional prompts for advanced students.

[0041] In this embodiment, the intelligent interactive guidance module 6 communicates with the intelligent diagnostic module 2 to obtain the type and location of faults encountered by students during the experiment in real time, and retrieves matching solutions from the preset guidance knowledge base. This realizes a closed-loop intelligent teaching assistance from "discovering problems" to "providing solutions". Afterwards, the intelligent interactive guidance module 6 presents guidance information in three forms: text, voice, or augmented reality (AR) markers, adapting to different students' learning habits and experimental scenarios, enhancing the intuitiveness and effectiveness of information transmission. In particular, augmented reality technology can directly overlay error correction steps onto real experimental equipment or circuits, significantly reducing the understanding threshold.

[0042] In addition, the intelligent interactive guidance module 6 can not only provide static solution matching, but also dynamically adjust the level of detail of the guidance content based on the student's historical grades and current experimental performance. This differentiated guidance strategy effectively takes into account the needs of students with different learning levels. For beginners, the system provides step-by-step, clear guidance to gradually guide them to complete the correction operation and reduce learning frustration. For advanced students, only directional prompts or key clues are provided to encourage their independent thinking and problem-solving abilities, avoid over-reliance on the system, and thus consolidate their advanced skills. This enables the deep integration of diagnostic results with personalized teaching strategies, improving the adaptability, practicality, and learning efficiency of the teaching system. It is especially suitable for remote or classroom interactive teaching in complex experimental scenarios in electronic engineering majors.

[0043] Example 3 like Figure 3 As shown, another embodiment of the present invention proposes an interactive teaching method for electronic information engineering based on artificial intelligence, comprising the following steps: S1. Real-time collection of multi-dimensional experimental data when students conduct hardware circuit experiments on teaching experimental platform 4. The multi-dimensional experimental data includes at least the circuit physical connection topology diagram, circuit electrical signal parameters, and student operation interaction data. S2. Input the multidimensional experimental data into a pre-trained deep learning model, analyze the current experimental state, compare the current experimental state with the standard expected state in the built-in expert knowledge base, and automatically locate and output the fault type and location when the deviation exceeds the preset threshold. S3. Based on the multidimensional experimental data and analysis results, the teacher terminal displays the experimental progress of all students, the error distribution heatmap, and the specific fault details of each student in real time. Based on the specific fault details of the students, the teacher issues warnings or pushes a list of students who need special attention according to preset rules.

[0044] S4. After outputting the fault type and location, based on the fault type and location, match the corresponding solution from the preset guidance knowledge base, and provide students with real-time, step-by-step error correction or operation guidance information in the form of text, voice or augmented reality markers. Based on the preset student's historical grades and current experimental performance, dynamically adjust the level of detail of the guidance information.

[0045] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An interactive teaching system for electronic information engineering based on artificial intelligence, characterized in that, include: The data acquisition module is connected to at least one teaching experimental platform and is used to collect multi-dimensional experimental data in real time when students conduct hardware circuit experiments on the teaching experimental platform. The multi-dimensional experimental data includes at least the circuit physical connection topology diagram, circuit electrical signal parameters, and student operation interaction data. The intelligent diagnostic module is communicatively connected to the data acquisition module. It is used to analyze the current experimental state based on the multidimensional experimental data output by the data acquisition module and a pre-trained deep learning model. It is also used to compare the current experimental state with the standard expected state in the built-in expert knowledge base, and automatically locate and output the fault type and location when the detected deviation exceeds a preset threshold. The teacher-side management module is connected to the data acquisition module and the intelligent diagnosis module via a network communication module. It is used to receive multidimensional experimental data output by the data acquisition module and analysis results output by the intelligent diagnosis module, which are forwarded by the network communication module. It is also used to display the experimental progress of all students, error distribution heatmap, and specific fault details of each student on the teacher terminal in real time, and to issue warnings or push a list of students who need special attention according to preset rules based on the specific fault details of the students.

2. The interactive teaching system for electronic information engineering based on artificial intelligence as described in claim 1, characterized in that, It also includes an intelligent interactive guidance module, which is communicatively connected to the intelligent diagnostic module. The intelligent interactive guidance module is used to match the corresponding solution from the preset guidance knowledge base according to the fault type and location output by the intelligent diagnostic module, and provide students with real-time, step-by-step error correction or operation guidance information in the form of text, voice or augmented reality markers.

3. The interactive teaching system for electronic information engineering based on artificial intelligence as described in claim 2, characterized in that, The intelligent interactive guidance module is also used to dynamically adjust the level of detail in the guidance information based on the student's historical grades and current experimental performance. The level of detail in the dynamically adjusted guidance information specifically includes: providing step-by-step direct guidance for beginners and providing directional prompts for advanced students.

4. The interactive teaching system for electronic information engineering based on artificial intelligence as described in claim 1, characterized in that, The deep learning model includes a first feature extraction model and a second feature extraction model. The first feature extraction model is used to process the physical connection topology diagram of the circuit captured by the camera to extract spatial structure features. The second feature extraction model is used to process time-series data composed of student operation interaction data to extract time-series change features.

5. The interactive teaching system for electronic information engineering based on artificial intelligence as described in claim 4, characterized in that, The first feature extraction model is a graph neural network, which is used to compare the physical circuit topology diagram actually connected by the student with the preset standard experimental circuit topology diagram and output the connection error node information.

6. The interactive teaching system for electronic information engineering based on artificial intelligence as described in claim 4, characterized in that, The second feature extraction model is a temporal convolutional network, which is used to analyze the temporal data to identify abnormal operation processes in the student operation sequence.

7. The interactive teaching system for electronic information engineering based on artificial intelligence as described in claim 1, characterized in that, The expert knowledge base is constructed by structurally modeling the circuit schematics, expected waveform parameters and operating steps in the standard experimental manual; The intelligent diagnostic module performs graph matching and / or parameter threshold comparison between the multidimensional experimental data and the expert knowledge base to determine the fault point.

8. An artificial intelligence-based interactive teaching method for electronic information engineering, applied to an artificial intelligence-based interactive teaching system for electronic information engineering as described in any one of claims 1-7, characterized in that, Includes the following steps: The system collects multi-dimensional experimental data in real time when students conduct hardware circuit experiments on the teaching experimental platform. The multi-dimensional experimental data includes at least the circuit physical connection topology diagram, circuit electrical signal parameters, and student operation interaction data. The multidimensional experimental data is input into a pre-trained deep learning model to analyze the current experimental state. The current experimental state is compared with the standard expected state in the built-in expert knowledge base. When the deviation exceeds a preset threshold, the fault type and location are automatically located and output. Based on the multidimensional experimental data and analysis results, the teacher's terminal displays the experimental progress of all students, the error distribution heatmap, and the specific fault details of each student in real time. Based on the specific fault details of the students, the teacher issues warnings or pushes a list of students who need special attention according to preset rules.

9. The interactive teaching method for electronic information engineering based on artificial intelligence as described in claim 8, characterized in that, Also includes: After outputting the fault type and location, the system matches the corresponding solution from the preset guidance knowledge base based on the fault type and location, and provides students with real-time, step-by-step error correction or operation guidance information in the form of text, voice or augmented reality markers. The level of detail of the guidance information is dynamically adjusted based on the student's historical grades and current experimental performance.

Images