Circuit board detection system and method

Abstract

The invention provides a circuit board detection system and method. The method belongs to the technical field of circuit board detection, and comprises the steps that original data of a circuit experiment board are collected through a high-precision sensing network, and the original data comprise static data and dynamic data; the collected data is preprocessed; extracting time-related features from the dynamic data, performing Fourier transform on the dynamic data, extracting frequency-related features, and extracting spatial distribution features from the static data; static and dynamic data of the circuit board are acquired through the high-precision sensing network, and the performance of the circuit board in different working states can be comprehensively captured, so that more comprehensive detection is performed.

Description

Technical Field

[0001] The present invention provides a circuit board detection system and method, belonging to the technical field of circuit board detection. Background Art

[0002] Traditional circuit board detection methods, such as visual inspection, electrical testing, etc., although ensuring product quality to a certain extent, are difficult to meet the industry's requirements in terms of detection accuracy and efficiency in the face of increasingly complex circuit board designs and material applications. Especially for problems such as micro defects, material aging, and performance fluctuations, traditional methods are often inadequate. Therefore, it is particularly urgent to develop a new method that can comprehensively, accurately, and efficiently detect the various performances of circuit boards. Summary of the Invention

[0003] The present invention provides a circuit board detection system and method to solve the problems mentioned in the above background art: A circuit board detection method proposed by the present invention, the method includes: S1. Collect the original data of the circuit experimental board through a high-precision sensing network, where the original data includes static data and dynamic data; and preprocess the collected data; S2. Extract time-related features from the dynamic data, perform Fourier transform on the dynamic data to extract frequency-related features, and extract spatially distributed features from the static data; S3. Based on machine learning algorithms, construct a circuit board state recognition model, use circuit board data with known states to train and verify the model, and apply the trained model to perform anomaly detection on real-time collected data to identify potential defects or abnormal states; S4. Based on time series analysis algorithms, model the historical data to predict the future performance change trend of the circuit board; according to the prediction results, combined with an expert system and a knowledge graph, automatically generate targeted optimization suggestions; S5. Display the relevant information in a visual manner and automatically generate a detection report.

[0004] A circuit board detection system proposed by the present invention, the system includes: Data acquisition module: Collect the original data of the circuit experimental board through a high-precision sensing network, where the original data includes static data and dynamic data; and preprocess the collected data; Feature extraction module: Extract time-related features from the dynamic data, perform Fourier transform on the dynamic data to extract frequency-related features, and extract spatially distributed features from the static data; Model construction module: Based on machine learning algorithms, construct a circuit board status recognition model, train and validate the model using circuit board data with known statuses, and apply the trained model to perform anomaly detection on real-time collected data to identify potential defects or abnormal statuses; Suggestion generation module: Based on time series analysis algorithms, model historical data to predict the future performance change trends of the circuit board; according to the prediction results, combine an expert system and a knowledge graph to automatically generate targeted optimization suggestions; Information display module: Display relevant information in a visual manner and automatically generate a detection report.

[0005] Advantages of the present invention: By collecting static and dynamic data of the circuit board through a high-precision sensing network, it is possible to comprehensively capture the performance of the circuit board under different working conditions, thereby enabling more comprehensive detection; extracting features of the circuit board from multiple dimensions such as time domain, frequency domain, and spatial distribution helps to deeply understand the working status and potential problems of the circuit board; using machine learning algorithms to construct a status recognition model can automatically learn and identify the normal and abnormal statuses of the circuit board, and at the same time predict future performance change trends based on historical data; combining an expert system and a knowledge graph, automatically generate targeted optimization suggestions according to the analysis results to help the maintenance team solve problems more effectively; display the detection results and relevant information through a visual interface and automatically generate a detection report, facilitating relevant personnel to quickly understand the status of the circuit board and the measures to be taken; the automated data processing and analysis process reduces manual intervention, improves the efficiency and accuracy of detection; through the prediction and anomaly detection of the circuit board performance, preventive maintenance can be achieved, reducing the occurrence of sudden failures and extending the service life of the circuit board; the provided optimization suggestions and risk assessments can help the design team, manufacturing team, and maintenance team make more informed decisions; this method can be adjusted according to different circuit board types and working environments, having strong adaptability and flexibility; by continuously collecting new data and feedback, the model and algorithms can be continuously optimized to improve the accuracy and reliability of detection. Description of the Drawings

[0006] Figure 1 It is a flowchart of the method described in the present invention; Figure 2 It is a block diagram of the system described in the present invention. Detailed Embodiment

[0007] The following describes the preferred embodiments of the present invention with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only for the purpose of illustrating and explaining the present invention, and are not used to limit the present invention.

[0008] One embodiment of the present invention, as Figure 1 shown, a circuit board detection method, the method includes: S1. Collect the original data of the circuit experiment board through a high-precision sensing network. The original data includes static data and dynamic data. The static data includes the output data of each sensor, including static information such as surface topography, material composition, and electrical characteristics. The dynamic data includes the dynamic data of the sensors during the operation of the circuit board, such as temperature changes and electrical parameter fluctuations, and preprocess the collected data. S2. Extract time-related features from the dynamic data, perform Fourier transform on the dynamic data to extract frequency-related features, and extract spatially distributed features from the static data. S3. Based on machine learning algorithms, construct a circuit board state recognition model, use the circuit board data with known states to train and validate the model, and apply the trained model to perform anomaly detection on the real-time collected data to identify potential defects or abnormal states. S4. Based on time series analysis algorithms, model the historical data to predict the future performance change trend of the circuit board. According to the prediction results, combined with an expert system and a knowledge graph, automatically generate targeted optimization suggestions. S5. Display the relevant information in a visual way. The relevant information includes detection results, anomaly information, performance prediction, and optimization suggestions. And automatically generate a detection report. The monitoring report includes the detection time, detection conditions, detection results, anomaly information, and optimization suggestions.

[0009] The working principle of the above technical solution is as follows: A high-precision sensing network is used to collect comprehensive data from the circuit experiment board, including static data and dynamic data. Static data mainly reflects the physical and chemical properties of the circuit board, such as surface topography, material composition, and electrical characteristics, etc.; dynamic data focuses on the real-time changes during the operation of the circuit board, such as temperature fluctuations, electrical parameter changes, etc. Key features are extracted from the preprocessed data for subsequent model training and anomaly detection. For dynamic data, time-related features (such as mean, variance, peak value, etc.) are extracted to capture the time-varying trend of the data; at the same time, Fourier transform is used to extract frequency-related features to analyze the periodic components of the data. For static data, spatially distributed features, such as surface roughness, material distribution, etc., are extracted to reflect the physical structure characteristics of the circuit board; based on machine learning algorithms, a circuit board state recognition model is constructed. The model is trained using circuit board data with known states so that it can accurately identify different states of the circuit board. After training, the model is applied to perform anomaly detection on the real-time collected data. By comparing the differences between the real-time data and the normal state data, potential defects or abnormal states are identified; based on time series analysis algorithms, historical data is modeled to predict the future performance change trend of the circuit board. By analyzing the time series features in the historical data, the law of the circuit board performance change over time can be revealed. According to the prediction results, combined with the expert system and knowledge graph, targeted optimization suggestions are automatically generated. These suggestions are aimed at guiding the design and manufacturing process of the circuit board to improve the reliability and performance of the circuit board; relevant information such as detection results, anomaly information, performance predictions, and optimization suggestions is displayed in a visual manner to facilitate users to intuitively understand the state and potential problems of the circuit board. At the same time, a detection report is automatically generated, which details the detection time, detection conditions, detection results, anomaly information, and optimization suggestions, etc., providing an important reference for the subsequent processing and maintenance of the circuit board.

[0010] The effects of the above technical solutions are as follows: By collecting the original data of the circuit experiment board through a high-precision sensing network, more accurate and comprehensive data can be obtained, including static data and dynamic data. These data provide a solid foundation for subsequent anomaly detection and performance prediction; preprocessing the collected data, such as denoising and standardization, can further improve the data quality, reduce the impact of noise and outliers on subsequent analysis, and thus enhance the accuracy and reliability of detection; extracting time-related features and frequency-related features from dynamic data, as well as extracting spatially distributed features from static data, these features can comprehensively reflect the state and performance of the circuit board. The efficiency of feature extraction helps to speed up the subsequent model training and anomaly detection; constructing a circuit board state recognition model based on machine learning algorithms and using the circuit board data with known states for training and verification. This intelligent method can automatically identify potential defects or abnormal states, improving the detection efficiency and accuracy; modeling the historical data based on time series analysis algorithms can predict the future performance change trend of the circuit board. This forward-looking prediction helps to timely discover potential problems and take corresponding measures to avoid or reduce the occurrence of faults; combining the expert system with the knowledge graph to automatically generate targeted optimization suggestions. These suggestions can guide the design and manufacturing process of the circuit board, improve the reliability and performance of the circuit board, and reduce the maintenance cost; displaying the relevant information in a visual way, making the detection results, anomaly information, performance prediction, optimization suggestions, etc. more intuitive and easy to understand. This helps users quickly understand the state and potential problems of the circuit board; automatically generating a detection report, which details the detection time, detection conditions, detection results, anomaly information, and optimization suggestions, etc. This standardized report helps subsequent analysis and traceability, improving the work efficiency and accuracy.

[0011] In one embodiment of the present invention, S1 includes: S11. Select sensors according to the structural characteristics, working environment, and detection requirements of the circuit board. The sensors include an infrared temperature sensor, a high-precision resistance measurement sensor, a capacitance measurement sensor, and an inductance measurement sensor; S12. Connect the selected sensors to the data acquisition system by wired or wireless means to form a sensing network; and calibrate the sensing network; S13. When the circuit board is in a non-working state or a stable state, start the sensing network to collect the static data of each sensor, including surface topography data (such as roughness, flatness), material composition data (such as element composition, alloy ratio), electrical characteristic data (such as resistance value, capacitance value, inductance value), etc.; and during the operation of the circuit board, monitor and record the dynamic data of the sensors in real time, such as temperature change data, electrical parameter fluctuation data (such as voltage fluctuation, current fluctuation), mechanical vibration data, etc.; S14. Clean the collected raw data to remove invalid data, abnormal data or duplicate data, and adopt filtering algorithms or denoising techniques to eliminate the influence of environmental noise or instrument errors on the data; S15. Normalize the data to convert it into a unified dimension and range, and standardize the data to eliminate measurement deviations between different sensors.

[0012] The working principle of the above technical solution is as follows: According to the structural characteristics of the circuit board (such as hierarchical structure, component layout), working environment (such as temperature, humidity, electromagnetic interference, etc.) and specific detection requirements (such as fault prediction, performance evaluation, etc.), suitable sensors are carefully selected. For example, to monitor the temperature change of the circuit board, an infrared temperature sensor is selected; to accurately measure the resistance, capacitance and inductance values, high-precision resistance measurement sensors, capacitance measurement sensors and inductance measurement sensors are respectively configured; the selected sensors are connected to the data acquisition system by wired (such as Ethernet, RS485, etc.) or wireless (such as Wi-Fi, Bluetooth, Zigbee, etc.) methods to build a comprehensive sensing network. To ensure the accuracy of the data, the sensing network is then strictly calibrated, including zero calibration, range calibration and cross-sensitivity calibration, etc., to eliminate the errors of the sensors themselves; when the circuit board is in a non-working state or a stable state, the sensing network is started to comprehensively collect the static data of each sensor, and these data comprehensively reflect the physical and chemical characteristics of the circuit board, such as surface topography, material composition and electrical characteristics, etc. When the circuit board starts to work, the sensing network enters the real-time monitoring mode, recording dynamic data such as temperature, electrical parameters (voltage, current) and mechanical vibration, etc., and these data help to capture the real-time state changes of the circuit board during operation; the raw data may contain invalid data (such as invalid readings caused by sensor failures), abnormal data (such as extreme values or mutation points) and duplicate data. Through the data cleaning step, these bad data can be effectively removed. At the same time, filtering algorithms (such as low-pass filtering, high-pass filtering, etc.) or denoising techniques (such as wavelet denoising, Kalman filtering, etc.) are adopted to further eliminate the influence of environmental noise or instrument errors on the data quality, ensuring the accuracy and reliability of the data; since different sensors may use different dimensions and measurement ranges, it is necessary to normalize the data to convert it into a unified dimension and range. In addition, to eliminate the measurement deviations between different sensors, standardization processing is also required to make the data comparable and consistent. This step is crucial for subsequent machine learning model training and anomaly detection, and can significantly improve the accuracy and robustness of the model.

[0013] The effects of the above technical solution are as follows: According to the structural characteristics, working environment and detection requirements of the circuit board, sensors are carefully selected. This targeted selection can ensure that the selected sensors can accurately and effectively monitor the key parameters of the circuit board, such as temperature, resistance, capacitance and inductance, etc.; By selecting a variety of sensors including infrared temperature sensors, high-precision resistance measurement sensors, capacitance measurement sensors and inductance measurement sensors, comprehensive monitoring of the circuit board status is achieved. This comprehensive monitoring helps to capture various subtle changes in the circuit board during operation, providing rich data support for subsequent anomaly detection and performance prediction; The selected sensors are connected to the data acquisition system by wired or wireless means to form a sensing network. This efficient construction method can ensure the real-time transmission and sharing of data, improving the efficiency and accuracy of data processing; The sensing network is strictly calibrated to ensure the accuracy and reliability of the data. This calibration can eliminate the errors of the sensors themselves, improve the accuracy and consistency of the data, and provide a solid foundation for subsequent data analysis and processing; When the circuit board is in a non-working state or a stable state, the static data of each sensor are collected. These data can reflect the physical and chemical characteristics of the circuit board, such as surface topography, material composition and electrical characteristics, etc., providing important basis for subsequent performance evaluation and fault prediction; During the operation of the circuit board, the dynamic data of the sensors are monitored and recorded in real time. These data can capture the real-time state changes of the circuit board during operation, such as temperature changes, electrical parameter fluctuations and mechanical vibrations, etc., providing key information for subsequent anomaly detection and performance optimization; Through the data cleaning step, invalid data, abnormal data or duplicate data are effectively removed to ensure the accuracy and reliability of the data; Filtering algorithms or denoising techniques are used to eliminate the influence of environmental noise or instrument errors on the data. This precise data processing can further improve the accuracy and consistency of the data, providing high-quality data support for subsequent data analysis and processing; The data are converted into a unified dimension and range, so that the data collected by different sensors are comparable and consistent. This normalization process can simplify the subsequent data processing and analysis process, improving the efficiency and accuracy of data processing; The measurement deviations between different sensors are eliminated, making the data more accurate and reliable. This standardization process can further improve the quality and consistency of the data, providing a solid foundation for subsequent data analysis and processing.

[0014] In an embodiment of the present invention, the S2 includes: S21. Extract time-related features from the dynamic data. The time-related features include mean value, variance, peak value, valley value, extreme value and change trend; and analyze the variation law of the time-related features with time to obtain the dynamic behavior information during the operation of the circuit board; S22. Perform Fourier transform or wavelet transform on the dynamic data to extract frequency-related features, where the frequency-related features include the main frequency, secondary frequency, spectral distribution, and power spectral density; and analyze the distribution law of the frequency-related features in the frequency domain to obtain the vibration and noise characteristics of the circuit board; S23. Extract spatially distributed features from the static data, where the spatially distributed features include surface roughness distribution, material composition distribution, and electrical property distribution, etc.; and analyze the distribution law of the spatially distributed features in space to obtain the microstructure and material properties of the circuit board; S24. Perform image processing on the surface topography data to extract corresponding features, where the corresponding features include edge features, texture features, and shape features; and analyze the manifestation of the corresponding features in the image to obtain the surface topography and structural characteristics of the circuit board.

[0015] The working principle of the above technical solution is as follows: Extract time-related features from the dynamic data, and these features can reflect the dynamic behavior information during the operation of the circuit board. The features include the mean value (reflecting the average level of the data), variance (measuring the degree of dispersion of the data), peak value (the maximum value in the data), valley value (the minimum value in the data), extreme value (collectively referring to the peak value and valley value), and change trend (the change of the data over time). By analyzing the variation law of these features over time, the dynamic performance of the circuit board during operation, such as stability and volatility, can be deeply understood; Perform Fourier transform or wavelet transform on the dynamic data to extract frequency-related features. These features can reflect the vibration and noise characteristics of the circuit board. The features include the main frequency (the strongest frequency component in the signal), secondary frequency (other frequency components except the main frequency), frequency spectrum distribution (the distribution of the signal in the frequency domain), and power spectral density (the density function describing the change of the signal power with frequency). By analyzing the distribution law of the frequency-related features in the frequency domain, the vibration mode and noise source in the circuit board can be identified, and then its vibration and noise performance can be evaluated; Extract spatially distributed features from the static data, and these features can reflect the microscopic structure and material properties of the circuit board. The features include surface roughness distribution (describing the distribution of minute unevenness on the surface of the circuit board), material composition distribution (describing the distribution of different material components in the circuit board), and electrical property distribution (describing the distribution of electrical parameters such as resistance, capacitance, and inductance in the circuit board). By analyzing the distribution law of these features in space, the microscopic structure and material composition of the circuit board can be deeply understood, and then its electrical performance and reliability can be evaluated; Perform image processing on the surface topography data to extract corresponding features. These features can reflect the surface topography and structural characteristics of the circuit board. The features include edge features (describing the shape and contour of the surface edge of the circuit board), texture features (describing the thickness, direction, density, etc. of the surface texture of the circuit board), and shape features (describing the characteristics of the overall or local shape of the surface of the circuit board). By analyzing the manifestation of these features in the image, the surface topography and structural characteristics of the circuit board can be intuitively understood, and then its manufacturing quality and appearance performance can be evaluated.

[0016] The effects of the above technical solutions are as follows: By extracting time-related features from dynamic data and analyzing the variation patterns of these features over time, the dynamic behavior information of the circuit board during operation can be deeply revealed. This helps to understand the stability, response speed, and possible abnormal fluctuations of the circuit board, providing an important basis for the performance evaluation and fault prediction of the circuit board; By performing Fourier transform or wavelet transform on the dynamic data, extracting frequency-related features, and analyzing the distribution patterns of these features in the frequency domain, the vibration and noise characteristics of the circuit board can be accurately evaluated. This helps to identify the vibration modes and noise sources in the circuit board, and then take corresponding vibration and noise reduction measures to improve the reliability and service life of the circuit board; By extracting spatially distributed features from static data and analyzing the distribution patterns of these features in space, the microscopic structure and material properties of the circuit board can be comprehensively understood. This helps to understand the material composition, microscopic structure, and possible defects of the circuit board, providing an important reference basis for the design and manufacturing of the circuit board; By performing image processing on the surface topography data, extracting edge features, texture features, shape features, etc., and analyzing the manifestations of these features in the image, the surface topography and structural characteristics of the circuit board can be intuitively displayed. This helps to understand the manufacturing quality, appearance performance, and possible surface defects of the circuit board, providing important support for the quality control and improvement of the circuit board. By extracting and analyzing time-related, frequency-related, spatially distributed, and surface topography features, the comprehensive monitoring and in-depth analysis of the circuit board state can be achieved. This helps to improve the accuracy of circuit board state monitoring and fault diagnosis, providing a scientific basis for the maintenance and management of the circuit board, and then ensuring the reliable operation and extending the service life of the circuit board.

[0017] In one embodiment of the present invention, step S3 includes: S31. Select the features useful for circuit board state recognition from the extracted features, and eliminate redundant or irrelevant features; Use principal component analysis to reduce the feature dimension; S32. Select a machine learning algorithm to train the model using circuit board data with known states, and adjust the model parameters. S33. Apply the trained machine learning model to perform anomaly detection on the real-time collected data to identify potential defects or abnormal states; S34. Set an anomaly threshold. When the detected anomaly degree exceeds the threshold, trigger an alarm mechanism and output anomaly information; Classify and identify the detected anomalies to determine the types of anomalies (such as overheating, short circuit, open circuit, etc.); The anomaly degree is calculated by the following formula:

[0018] where AD represents the quantization value of the anomaly degree; n represents the number of features. denotes the weight of the \(i\)-th feature, reflecting the importance of this feature in anomaly detection; denotes the \(i\)-th feature value collected in real time; denotes the average value of the \(i\)-th feature in the normal state; denotes the standard deviation of the \(i\)-th feature in the normal state; \(f\) represents a mapping function that normalizes the deviation between the feature value and the normal value to a value between 0 and 1, used to reflect the severity of the anomaly. And \(f\) is expressed in the following form:

[0019] where, and denote the set thresholds, used to determine whether a feature value is significantly abnormal.

[0020] S35. Locate the specific location where the anomaly occurs according to the spatial distribution and time series information of the anomaly features.

[0021] The working principle of the above technical solution is as follows: Select those features that are most useful for circuit board state recognition from the numerous extracted features. For example, in circuit board state recognition, the stability of voltage and current, the uniformity of frequency distribution, and the consistency of surface roughness and material composition may be selected as key features. At the same time, eliminate those redundant or irrelevant features to reduce the complexity of data processing and improve the accuracy of recognition; Use PCA technology to reduce the dimension of the selected feature dimensions. PCA projects the original high-dimensional data onto these principal components by finding the main components in the data (i.e., the directions with the largest variance), thereby achieving dimensionality reduction. This can not only reduce the storage and computational amount of data, but also remove noise and redundant information in the data, improving the performance and accuracy of subsequent machine learning models; Select machine learning algorithms suitable for circuit board state recognition, such as support vector machine (SVM), random forest, neural network, etc. These algorithms can learn the mapping relationship between features and target states from circuit board data with known states; Use circuit board data with known states to train the selected machine learning model. During the training process, the model will continuously adjust its internal parameters to minimize the prediction error and improve the recognition accuracy; Optimize the parameters of the model through techniques such as cross-validation and grid search to find the best parameter combination and improve the generalization ability and recognition accuracy of the model; Real-time collect the working data of the circuit board through devices such as sensors, including voltage, current, frequency, temperature, etc.; Apply the trained machine learning model to perform anomaly detection on the real-time collected data. The model will judge whether the current data is abnormal according to the mapping relationship between the learned features and the target state; When abnormal data is detected, the model will further analyze the abnormal features and patterns to identify potential defects or abnormal states, such as overheating, short circuit, open circuit, etc.; Set reasonable anomaly thresholds according to the working characteristics and historical data of the circuit board. When the detected anomaly degree exceeds the threshold, trigger the alarm mechanism; When the anomaly degree exceeds the threshold, the system will automatically output anomaly information and trigger the alarm mechanism, such as sound and light alarm, SMS notification, etc. At the same time, classify and identify the detected anomalies to determine the type of anomaly; Analyze the possible causes and locations of the anomaly according to the spatial distribution and time series information of the anomaly features; Combine the structure and working principle of the circuit board to locate the specific location where the anomaly occurs, such as a certain component, a certain line or a certain module; According to the positioning result, take corresponding measures for fault troubleshooting and repair to ensure the normal operation of the circuit board.

[0022] The effects of the above technical solution are as follows: By accurately selecting the features crucial for circuit board state recognition from a large number of extracted features and eliminating redundant or irrelevant features simultaneously, this process significantly improves the data processing efficiency. Further, the use of principal component analysis (PCA) technology to reduce the feature dimension not only reduces the computational complexity but also helps to eliminate noise in the data, laying a solid foundation for the subsequent training of machine learning models; selecting appropriate machine learning algorithms and fully training the model with circuit board data of known states ensures that the model can accurately capture the key features of the circuit board state. By finely tuning the model parameters, the recognition accuracy and generalization ability of the model are further improved, enabling it to make accurate judgments even when facing unknown data; through the anomaly detection of real-time collected data, potential defects or abnormal states can be quickly identified. This function is of great significance for preventing circuit board failures and ensuring the stable operation of equipment. At the same time, by continuously monitoring data changes, the system can promptly detect and respond to anomalies, providing a valuable time window for maintenance personnel; the system can set reasonable anomaly thresholds and trigger an alarm mechanism according to the detected anomaly degree, outputting detailed anomaly information. More importantly, the system can also classify and identify the detected anomalies to determine the specific types of anomalies (such as overheating, short circuit, open circuit, etc.), which helps maintenance personnel quickly locate the root cause of the problem and take targeted repair measures. In addition, based on the spatial distribution and time series information of anomaly features, the system can further locate the specific location where the anomaly occurs, providing precise guidance for maintenance work; through a series of measures such as feature optimization, machine learning model training, real-time anomaly detection, anomaly classification, and precise positioning, the accuracy and efficiency of circuit board state monitoring are significantly improved. This not only helps to reduce equipment failure rates, extend service life, but also reduces maintenance costs and time, creating greater economic benefits for enterprises. At the same time, this technical solution also has high flexibility and scalability, can adapt to the monitoring requirements of different scales and types of circuit boards, and has broad application prospects. The above formula for calculating the anomaly degree (AD) can comprehensively consider the importance of different features in anomaly detection by introducing feature weights. This means that if a certain feature is particularly crucial for the recognition of the circuit board state, its contribution in calculating the anomaly degree will be greater; the mapping function in the formula normalizes the deviation between the feature value and the normal value to a value between 0 and 1, which helps to unify the dimensions between different features, enabling the anomaly degrees of different features to be compared and summarized; the threshold setting in the mapping function enables the system to flexibly handle anomalies of different degrees.When the eigenvalue is between these two thresholds, the degree of abnormality will increase linearly with the increase of the deviation; when the eigenvalue exceeds these two thresholds, the degree of abnormality will be regarded as the highest or the lowest, which helps the system to quickly identify significantly abnormal features; by comprehensively considering multiple features and their weights, as well as standardizing the feature deviation, this formula can more accurately evaluate the degree of abnormality of the circuit board state. This helps to reduce false alarms and missed detections, and improve the accuracy and reliability of anomaly detection; the quantified value of the degree of abnormality provides a basis for subsequent anomaly classification and localization. When the degree of abnormality exceeds the set threshold, the system can trigger an alarm mechanism and locate the specific location where the anomaly occurs based on the spatial distribution and time series information of the abnormal features. This helps maintenance personnel quickly locate the problem and take corresponding repair measures; this formula can adapt to different types of circuit boards and different working environments. By adjusting parameters such as feature weights and thresholds, the system can be customized and optimized according to the actual situation to meet the requirements of different scenarios.

[0023] In one embodiment of the present invention, the S33 includes: S331. Preprocess the circuit board state data collected in real time from sensors or other data sources; according to the typical time scale of the circuit board state change, divide the continuous data stream into multiple time windows, and the data within each window is used as a sample for one model input; S332. Construct a prediction model through various machine learning algorithms (such as random forest, support vector machine, neural network, etc.), and select the optimal model combination based on the cross-validation method; S333. Calculate the residual between the actual observed value and the model predicted value, and the residual reflects the abnormal information not captured by the model; and based on the residual monitoring mechanism, track the change trend of the residual in real time; S334. Further analyze the residual data through a deep learning model (such as convolutional neural network CNN, recurrent neural network RNN or their variants), and automatically extract deeper abnormal features; S335. Use the trained deep learning model to perform pattern matching on the extracted abnormal features, identify common abnormal patterns (such as periodic fluctuations, sudden jumps, etc.), and perform preliminary classification; S336. Based on the significance of the abnormal features and the statistical characteristics of historical abnormal data; through an abnormal scoring system, quantify and score each detected abnormal event; S337. According to the abnormal score, rank the detected abnormal events in priority, and combine the working environment (such as temperature, humidity, voltage fluctuation, etc.) and operating status (such as load size, working time, etc.) information of the circuit board to perform context-aware analysis on the anomalies; S338. Refine the classification of the initially identified anomalies based on context information and provide an explanation of the anomaly causes.

[0024] The working principle of the above technical solution is as follows: Preprocess the real-time collected circuit board status data, such as denoising, normalization, etc., to improve the data quality; Divide the continuous data stream into multiple time windows according to the typical time scale of the circuit board status change. The data within each time window is used as a sample for a model input for subsequent model prediction; Use multiple machine learning algorithms (such as random forest, support vector machine, neural network, etc.) to build a prediction model; Evaluate the performance of each model through the cross-validation method and select the model combination with the best performance for subsequent analysis. This can ensure that the model has high prediction accuracy when facing unknown data; Calculate the residual between the actual observed value and the model predicted value. The residual reflects the anomaly information not captured by the model; Based on the residual monitoring mechanism, track the change trend of the residual in real time. When the residual exceeds the preset threshold, it may indicate an anomaly; Use a deep learning model (such as convolutional neural network CNN, recurrent neural network RNN or its variants) to further analyze the residual data and automatically extract deeper anomaly features; Use the trained deep learning model to perform pattern matching on the extracted anomaly features to identify common anomaly patterns (such as periodic fluctuations, sudden jumps, etc.); Conduct a preliminary classification of the identified anomaly patterns to provide a basis for subsequent anomaly scoring and priority ranking; Based on the significance of the anomaly features and the statistical characteristics of historical anomaly data, quantitatively score each detected anomaly event through an anomaly scoring system; According to the anomaly score, rank the detected anomaly events by priority. Prioritize the high-risk anomalies that may cause serious failures or affect system security; Combine the working environment (such as temperature, humidity, voltage fluctuations, etc.) and operating status (such as load size, working hours, etc.) information of the circuit board to conduct context-aware analysis of the anomaly; Refine the classification of the initially identified anomalies based on context information and provide an explanation of the anomaly causes. For example, "overheating anomaly caused by too high environmental temperature" or "component aging anomaly caused by long-term high-load operation".

[0025] The effects of the above technical solution are as follows: By collecting and preprocessing the circuit board status data in real time, this solution can quickly respond to changes in the circuit board status; dividing the continuous data stream into multiple time windows, and using the data within each window as samples for model input, ensuring the timeliness and continuity of data processing; using multiple machine learning algorithms such as random forest, support vector machine, and neural network to build a prediction model, and selecting the optimal model combination with the best performance through cross-validation; this multi-algorithm fusion method improves the prediction accuracy of the model for changes in the circuit board status, thus enhancing the reliability of anomaly detection; by calculating the residuals between the actual observed values and the model predicted values, this solution can capture anomaly information that has not been captured by the model; using deep learning models such as convolutional neural network (CNN) and recurrent neural network (RNN) to further analyze the residual data, automatically extracting deeper anomaly features; using the trained deep learning model to perform pattern matching on the extracted anomaly features, it can identify common anomaly patterns (such as periodic fluctuations, sudden jumps, etc.) and perform preliminary classification; this helps maintenance personnel quickly locate the root cause of the problem and improve the maintenance efficiency; based on the significance of anomaly features and the statistical characteristics of historical anomaly data, quantifying and scoring each detected anomaly event through an anomaly scoring system; sorting the detected anomaly events according to the anomaly scores to ensure that high-risk anomalies that may cause serious failures or affect system safety are processed first; combining information on the working environment (such as temperature, humidity, voltage fluctuations, etc.) and operating status (such as load size, working hours, etc.) of the circuit board to perform context-aware analysis of anomalies; refining the classification of the initially identified anomalies based on the context information and providing explanations for the anomaly causes, such as "overheating anomaly due to too high environmental temperature" or "component aging anomaly caused by long-term high-load operation"; this helps maintenance personnel more accurately understand the cause of the anomaly and formulate more effective maintenance strategies; through real-time anomaly detection and identification, this solution can timely discover and handle potential failures, thus preventing the occurrence of serious failures; it helps reduce the equipment failure rate, extend the service life of the circuit board, and reduce maintenance costs and time; the anomaly detection and identification results provided by this solution can be used as an important basis for optimizing the production process and formulating maintenance plans; through data analysis, enterprises can better understand the trend of changes in the circuit board status and provide decision-making support for future equipment procurement, maintenance, and management.

[0026] In one embodiment of the present invention, the S337 includes: On the basis of the original anomaly scoring system, introducing more dimensions of consideration factors to form a more comprehensive anomaly scoring system; Dynamically adjust the weights of each scoring dimension according to the current working environment and operating status of the circuit board; for example, in a high-temperature environment, the weight of temperature-related anomalies (such as overheating) should be appropriately increased; in a high-load operating state, the weight of load-related anomalies should also be increased accordingly; based on the anomaly scores, prioritize the anomaly events through sorting algorithms (such as TOPSIS, grey relational analysis, etc.); Based on the emergency response strategy library, provide preset response measures and emergency plans for anomaly events with different priorities; Integrate the working environment data of the circuit board (such as temperature, humidity, voltage fluctuations, etc.), operating status data (such as load size, working hours, etc.), and historical fault records to form a comprehensive context information library; Use machine learning algorithms (such as association rule mining, clustering analysis, etc.) to conduct intelligent correlation analysis on the context information to reveal the potential relationships between anomaly events, environmental factors, and operating status; Based on deep learning models (such as deep belief networks, generative adversarial networks, etc.), perform deeper feature extraction and pattern recognition on the anomaly features, and apply causal inference models (such as Bayesian networks, structural equation models, etc.) to infer the possible causes of the anomaly events in combination with the context information and anomaly features; Based on the anomaly classification, cause inference, and priority ranking results, provide personalized anomaly handling suggestions for circuit board maintenance personnel, including emergency handling measures, long-term maintenance strategies, etc.; Combine historical anomaly data and the current operating status to formulate a preventive maintenance plan to predict and prevent the occurrence of potential faults.

[0027] The working principle of the above technical solution is to introduce more dimensional consideration factors on the basis of the original anomaly scoring system, such as anomaly duration, scope of influence (such as the number of circuit boards or key functions affected), difficulty and results of handling historical similar anomalies, etc.; dynamically adjust the weights of each scoring dimension according to the current working environment and operating status of the circuit board to ensure the accuracy and pertinence of anomaly scoring; based on the anomaly scoring, perform priority sorting on anomaly events through sorting algorithms (such as TOPSIS, grey relational analysis, etc.); according to anomaly events with different priorities, combine with the emergency response strategy library to provide preset response measures and emergency plans to ensure that anomaly events can be handled in a timely and effective manner; integrate the working environment data of the circuit board (such as temperature, humidity, voltage fluctuation, etc.), operating status data (such as load size, working time, etc.) and historical fault records to form a comprehensive context information library; utilize the context information library to provide data support for subsequent intelligent correlation analysis, feature extraction and pattern recognition, and causal reasoning; use machine learning algorithms (such as association rule mining, clustering analysis, etc.) to perform intelligent correlation analysis on the context information to reveal the potential connections between anomaly events and environmental factors and operating status; perform deeper feature extraction and pattern recognition on anomaly features through deep learning models (such as deep belief network, generative adversarial network, etc.) to improve the accuracy and reliability of anomaly recognition; introduce causal reasoning models (such as Bayesian network, structural equation model, etc.), combine with context information and anomaly features to infer the possible causes of anomaly events; based on anomaly classification, cause inference and priority sorting results, provide personalized anomaly handling suggestions for circuit board maintenance personnel, including emergency handling measures, long-term maintenance strategies, etc.; combine historical anomaly data and current operating status to formulate preventive maintenance plans to predict and prevent the occurrence of potential faults; through regular inspections and maintenance, ensure that the circuit board is in good working condition, extend its service life, and reduce maintenance costs.

[0028] The effects of the above technical solution are as follows: By introducing more dimensional consideration factors such as abnormal duration, influence range, processing difficulty and results of historical similar abnormalities, the abnormal scoring system becomes more comprehensive and can more accurately reflect the severity and urgency of abnormalities; Dynamically adjusting the weights of each scoring dimension according to the current working environment and operating status of the circuit board improves the pertinence and accuracy of abnormal scoring. For example, in a high-temperature environment, the weight of overheating abnormalities increases, enabling more sensitive detection of such abnormalities and timely measures; Based on the abnormal score, priority ranking of abnormal events is performed through a sorting algorithm, ensuring reasonable allocation of processing resources and enabling timely handling of high-priority and high-risk abnormalities; For abnormal events with different priorities, preset response measures and emergency plans are provided, improving the efficiency and pertinence of abnormal handling, reducing handling time, and reducing the impact of faults; Integrating the working environment data, operating status data, and historical fault records of the circuit board to form a comprehensive context information library provides rich data support for subsequent intelligent correlation analysis and cause inference; Using machine learning algorithms to perform intelligent correlation analysis on the context information reveals the potential connections between abnormal events, environmental factors, and operating status, helping to discover the root causes of abnormalities; Referring to the causal reasoning model, combined with the context information and abnormal characteristics, to infer the possible causes of abnormal events, improving the accuracy and reliability of abnormal diagnosis; Based on the abnormal classification, cause inference, and priority ranking results, personalized abnormal handling suggestions are provided for circuit board maintenance personnel, including emergency handling measures and long-term maintenance strategies, improving the pertinence and effectiveness of maintenance work; Combining historical abnormal data and current operating status to formulate a preventive maintenance plan, predicting and preventing the occurrence of potential faults, reducing the failure rate and maintenance cost, and extending the service life of the circuit board; Through the above technical solution, the management and maintenance of circuit board abnormalities become more efficient and accurate, helping to improve the stability and reliability of the entire system; Through timely and accurate abnormal handling and preventive maintenance, the downtime and maintenance cost caused by faults are reduced, and the operation and maintenance efficiency are improved; For systems or devices that rely on the stable operation of the circuit board, this technical solution helps to improve the user experience and enhance user satisfaction.

[0029] In one embodiment of the present invention, S35 includes: S351. Mapping the spatial distribution information of abnormal features onto the physical layout diagram of the circuit board to form a feature heat map or distribution map; and performing trend analysis on the time series data of abnormal features to identify the variation law of abnormalities over time, such as periodic fluctuations, progressive deterioration, or sudden events; S352. Applying pattern recognition algorithms, such as Hidden Markov Model (HMM), Long Short-Term Memory Network (LSTM), etc., to deeply mine the time series data to identify common abnormal patterns, such as overheating patterns, short-circuit patterns, etc.; S353. Combine the spatial distribution characteristics and time series trends, and adopt multi-source information fusion technologies such as Bayesian networks and deep learning models to comprehensively evaluate the anomalies; based on the fused information, use high-precision positioning algorithms such as received signal strength indication (RSSI) and time difference of arrival (TDOA) to accurately determine the specific location where the anomalies occur; at the same time, combine the design drawings and component information of the circuit board to provide technicians with a detailed description of the anomaly location and a list of possible faulty components; S354. Evaluate the impact range of the anomalies on the overall performance of the circuit board according to the spatial distribution and time series trends of the anomaly characteristics, including the number of affected components, circuit paths, and potential performance degradation; based on historical data and anomaly characteristics, use machine learning algorithms to predict the risk levels that the anomalies may trigger, such as minor faults, serious faults, or system crashes; S355. Automatically generate anomaly handling suggestions according to the anomaly location, impact range evaluation, and risk prediction results, including possible replacement of faulty components, adjustment of circuit paths, or system restart.

[0030] The working principle of the above technical solution is as follows: Map the spatial distribution information of abnormal features onto the physical layout diagram of the circuit board, and visually display the distribution of abnormalities on the circuit board through visualization means (such as feature heat maps or distribution maps); this helps technicians quickly identify areas with concentrated abnormalities or components with frequent abnormalities, providing clues for subsequent positioning and handling; conduct trend analysis on the time series data of abnormal features to identify the variation patterns of abnormalities over time; including periodic fluctuations (such as seasonal faults), progressive deterioration (such as performance degradation caused by component aging), or sudden events (such as accidental short circuits or overloads); through trend analysis, the future development trend of abnormalities can be predicted, providing a basis for preventive maintenance; apply pattern recognition algorithms (such as Hidden Markov Model HMM, Long Short-Term Memory Network LSTM, etc.) to deeply mine the time series data; these algorithms can identify common abnormal patterns, such as overheating patterns (faults caused by excessive component temperature), short circuit patterns (excessive current caused by abnormal circuit paths), etc.; through pattern recognition, complex abnormal data can be simplified into patterns that are easy to understand and process, improving the efficiency and accuracy of abnormal handling; combine spatial distribution features and time series trends, and use multi-source information fusion technologies (such as Bayesian networks, deep learning models, etc.) to comprehensively evaluate abnormalities; this helps integrate information from different sources, improving the accuracy and reliability of abnormal positioning; based on the fused information, use high-precision positioning algorithms (such as Received Signal Strength Indication - RSSI, Time Difference of Arrival - TDOA, etc.) to accurately determine the specific location where the abnormality occurs; combine the design drawings and component information of the circuit board to provide technicians with a detailed description of the abnormal location and a list of possible faulty components; this helps technicians quickly locate the fault point and take targeted handling measures; evaluate the impact range of the abnormality on the overall performance of the circuit board according to the spatial distribution and time series trend of the abnormal features; this includes the number of affected components, circuit paths, and the potential degree of performance degradation; through the impact range evaluation, the severity of the abnormality and its impact on the system can be understood, providing a basis for formulating handling strategies; based on historical data and abnormal features, use machine learning algorithms to predict the risk level that the abnormality may trigger; this includes minor faults (such as minor performance fluctuations caused by component performance degradation), severe faults (such as system function loss caused by damage to key components), or system crashes (such as the entire circuit board being unable to work properly), etc.; through risk prediction, measures can be taken in advance to reduce the probability and impact degree of faults; according to the results of abnormal positioning, impact range evaluation, and risk prediction, automatically generate abnormal handling suggestions; these suggestions may include possible replacement of faulty components (such as replacing damaged components to restore system function), circuit path adjustment (such as modifying the circuit path to bypass the fault area), or system restart (such as clearing temporary faults or restoring the system to its initial state), etc.; through automatically generating handling suggestions, the fault handling time can be shortened, and the reliability and stability of the system can be improved.

[0031] The effects of the above technical solutions are as follows: mapping the spatial distribution information of abnormal features onto the physical layout diagram of the circuit board to form a feature heat map or distribution map, enabling technicians to visually see the distribution of abnormalities on the circuit board; this helps to quickly identify abnormal areas and facilitates subsequent positioning and processing; performing trend analysis on the time series data of abnormal features can identify the changing patterns of abnormalities over time, such as periodic fluctuations, progressive deterioration, or sudden events; it helps to predict the development trend of abnormalities, take preventive measures in advance to intervene, and prevent the occurrence or expansion of faults; using high-precision positioning algorithms such as RSSI and TDOA can accurately determine the specific location where the abnormality occurs; combining with the design drawings and component information of the circuit board provides technicians with a detailed description of the abnormal location and a list of possible faulty components, improving the accuracy and reliability of positioning; applying pattern recognition algorithms such as Hidden Markov Model (HMM) and Long Short-Term Memory Network (LSTM) to deeply mine the time series data can identify common abnormal patterns, such as overheating patterns and short-circuit patterns; it helps to simplify complex abnormal data into easy-to-understand and process patterns, improving the intelligent level of abnormal handling; combining spatial distribution features and time series trends, using multi-source information fusion technologies such as Bayesian networks and deep learning models to comprehensively evaluate abnormalities; it can integrate information from different sources, improve the accuracy and reliability of abnormal identification, and provide a more comprehensive basis for subsequent processing; according to the spatial distribution and time series trends of abnormal features, evaluating the impact range of abnormalities on the overall performance of the circuit board, including the number of affected components, circuit paths, and potential performance degradation; it helps to understand the severity of abnormalities and their impact on the system, providing a basis for formulating treatment strategies; based on historical data and abnormal features, using machine learning algorithms to predict the risk level that abnormalities may trigger, such as minor faults, serious faults, or system crashes; it helps to take preventive measures in advance to reduce the probability and impact of faults, improving the reliability and stability of the system; according to the results of abnormal positioning, impact range evaluation, and risk prediction, automatically generating abnormal handling suggestions, including possible replacement of faulty components, adjustment of circuit paths, or system restart, etc.; reducing the time for technicians to manually analyze and handle abnormalities and improving work efficiency; the generated abnormal handling suggestions can be customized according to specific abnormal situations and system characteristics, improving the pertinence and effectiveness of handling.

[0032] In one embodiment of the present invention, the S351 includes: Obtaining abnormal feature data from the anomaly detection module, where the feature data includes voltage fluctuations, current anomalies, temperature rises, etc.; and using Geographic Information System (GIS) to accurately map the spatial distribution information of the abnormal features onto the physical layout diagram of the circuit board; Generate a feature heat map or distribution map based on the mapping result; among them, in the heat map, different colors represent different abnormal intensities or frequencies, enabling technicians to visually identify the abnormal concentration areas. The distribution map may be displayed in the form of a scatter plot, contour map, etc., to help understand the spatial distribution law of the anomalies. Conduct in-depth analysis on the time series data of the abnormal features, and use time series analysis algorithms (such as ARIMA, STL decomposition, etc.) to identify the variation law of the anomalies over time; including periodic fluctuations (such as overheating phenomena occurring at specific times every day), progressive deterioration (such as the gradual decline of component performance over time), or sudden events (such as sudden short circuits); Combine the spatial distribution information and time series trends for spatio-temporal correlation analysis; based on the spatio-temporal correlation analysis result, construct an abnormal propagation model; this model can simulate the diffusion process of the anomalies on the circuit board and predict the areas and time ranges that the anomalies may affect; Use pattern recognition algorithms (such as K-means clustering, DBSCAN density clustering, etc.) to perform preliminary pattern matching and clustering on the spatio-temporal correlation analysis result; extract and describe the features of the identified abnormal patterns to form a pattern feature library.

[0033] The working principle of the above technical solution is as follows: Obtain abnormal feature data such as voltage fluctuations, current anomalies, and temperature rises from the anomaly detection module; use Geographic Information System (GIS) technology to accurately map the spatial distribution information of these abnormal features onto the physical layout diagram of the circuit board. This step ensures the accurate correspondence between the abnormal features and the actual layout of the circuit board; based on the mapping result, generate a feature heat map or distribution map; the heat map represents different abnormal intensities or frequencies through different colors, enabling technicians to visually identify the abnormal concentration areas. The distribution map may be displayed in the form of a scatter plot, contour map, etc., to help understand the spatial distribution law of the anomalies; conduct in-depth analysis on the time series data of the abnormal features; use time series analysis algorithms (such as ARIMA, STL decomposition, etc.) to identify the variation law of the anomalies over time. This includes periodic fluctuations (such as overheating phenomena occurring at specific times every day), progressive deterioration (such as the gradual decline of component performance over time), or sudden events (such as sudden short circuits); combine the spatial distribution information and time series trends for spatio-temporal correlation analysis; based on the spatio-temporal correlation analysis result, construct an abnormal propagation model. This model can simulate the diffusion process of the anomalies on the circuit board and predict the areas and time ranges that the anomalies may affect. This is of great significance for understanding the propagation mechanism of the anomalies and formulating preventive measures; use pattern recognition algorithms (such as K-means clustering, DBSCAN density clustering, etc.) to perform preliminary pattern matching and clustering on the spatio-temporal correlation analysis result; extract and describe the features of the identified abnormal patterns to form a pattern feature library. The pattern feature library provides a basis for subsequent pattern recognition and anomaly warning.

[0034] The effects of the above technical solution are as follows: By using the Geographic Information System (GIS), the spatial distribution information of abnormal features is accurately mapped onto the physical layout diagram of the circuit board, enabling technicians to clearly see the specific locations of abnormalities on the circuit board; This intuitive display method helps technicians quickly locate the abnormal areas and facilitates subsequent analysis and processing; By generating a feature heat map or distribution map, different colors represent different abnormal intensities or frequencies, enabling technicians to visually identify the concentrated areas of abnormalities; The heat map and distribution map provide an intuitive visualization of abnormal features, helping technicians better understand the spatial distribution law of abnormalities; By using time series analysis algorithms (such as ARIMA, STL decomposition, etc.) to deeply analyze the time series data of abnormal features, the changing law of abnormalities over time can be identified; It helps technicians understand the development trend of abnormalities, predict future abnormal situations, and thus take measures in advance for intervention; Through time series analysis, abnormal change laws such as periodic fluctuations, progressive deterioration, and sudden events can be identified; The identification of these laws helps technicians better understand the causes and impacts of abnormalities and provides a scientific basis for subsequent processing; By combining the spatial distribution information and time series trends, spatio-temporal correlation analysis is carried out to reveal the mutual relationship of abnormalities in space and time; It helps technicians better understand the propagation mechanism and diffusion process of abnormalities; Based on the results of spatio-temporal correlation analysis, an abnormal propagation model is constructed to simulate the diffusion process of abnormalities on the circuit board and predict the areas and time ranges that may be affected by the abnormalities; The construction of the abnormal propagation model provides a tool for prediction and early warning for technicians, helping to take measures in advance to prevent the diffusion and deterioration of abnormalities; By using pattern recognition algorithms (such as K-means clustering, DBSCAN density clustering, etc.) to perform preliminary pattern matching and clustering on the results of spatio-temporal correlation analysis; It helps technicians simplify complex abnormal data into patterns that are easy to understand and process, improving the efficiency and accuracy of abnormal processing; Feature extraction and description are carried out on the identified abnormal patterns to form a pattern feature library; The pattern feature library provides basic data support for subsequent pattern recognition and abnormal early warning, helping technicians better understand and respond to abnormal situations on the circuit board.

[0035] In one embodiment of the present invention, step S4 includes: S41. Based on time series analysis algorithms such as ARIMA, LSTM, etc., model the historical data; According to the model, predict the future performance change trend of the circuit board (for example, one week), including the temperature change trend, electrical parameter change trend, etc.; S42. Evaluate the prediction results, analyze the prediction errors and sources of uncertainty; And quantify the uncertainty of the prediction results using confidence intervals; S43. Combine an expert system with a knowledge graph to analyze the reasons for the performance degradation of the circuit board. The reasons include material aging, design defects, and manufacturing process problems. Based on the reasons for performance degradation, formulate targeted optimization suggestions, including design improvement, manufacturing process optimization, maintenance strategy adjustment, etc. S44. And feedback the optimization suggestions to the relevant personnel, where the relevant personnel include the circuit board design team, manufacturing team, or maintenance team.

[0036] The working principle of the above technical solution is as follows: Collect historical performance data of the circuit board, such as temperature, electrical parameters, etc.; Preprocess the data, including steps such as cleaning, denoising, and normalization, to ensure data quality; Use time series analysis algorithms (such as ARIMA, LSTM, etc.) to model the historical data; These algorithms can capture the time dependence and trends in the data, thereby constructing a prediction model; Predict the future performance change trends of the circuit board based on the model, such as the temperature change trend and electrical parameter change trend within a week, etc.; The prediction results can provide important references for subsequent analysis and optimization; Evaluate the prediction results and calculate the prediction error; Analyze the error sources, such as data quality, model selection, parameter settings, etc.; Quantify the uncertainty of the prediction results using confidence intervals; The confidence interval represents the range of the prediction results under a certain probability and helps to understand the reliability of the prediction results; Combine an expert system with a knowledge graph to analyze the reasons for the performance degradation of the circuit board; The reasons may include material aging, design defects, manufacturing process problems, etc.; The expert system uses the knowledge and experience of domain experts for reasoning, and the knowledge graph provides a structured knowledge representation and query ability; Based on the reasons for performance degradation, formulate targeted optimization suggestions; The optimization suggestions may include design improvement, manufacturing process optimization, maintenance strategy adjustment, etc.; These suggestions aim to improve the performance, reliability, and service life of the circuit board; Feedback the optimization suggestions to the relevant personnel, such as the circuit board design team, manufacturing team, or maintenance team; The feedback method can be reports, meetings, emails, etc.; Guide the relevant personnel to carry out improvement and optimization work according to the optimization suggestions; Provide necessary technical support and training to ensure the effective implementation of the improvement and optimization measures.

[0037] The effects of the above technical solutions are as follows: By using time series analysis algorithms such as ARIMA and LSTM to model historical data, the time dependence and trends in the circuit board performance data can be accurately captured; this makes the prediction results more accurate, helps to discover potential performance problems in advance, and provides a scientific basis for subsequent prevention and maintenance work; predicting the future performance change trends of the circuit board, such as temperature change trends, electrical parameter change trends, etc., can help technicians timely understand the operating conditions of the circuit board; through the prediction results, technicians can take measures in advance, such as adjusting the working environment, optimizing the circuit layout, etc., to avoid performance degradation and failures; evaluating the prediction results and analyzing the sources of prediction errors and uncertainties helps technicians understand the reliability and limitations of the prediction results; by using confidence intervals to quantify the uncertainty of the prediction results, the risk level of the prediction results can be more accurately evaluated; quantifying the uncertainty of the prediction results helps to formulate more reasonable risk management strategies; technicians can formulate targeted preventive measures and emergency plans based on the width of the confidence interval and the reliability of the prediction results to reduce potential risks; combining an expert system and a knowledge graph to analyze the reasons for the performance degradation of the circuit board can make full use of the knowledge and experience of domain experts, as well as the structured knowledge representation and query capabilities; this makes the analysis of the reasons for performance degradation more in-depth and comprehensive, and helps to discover potential design defects, manufacturing process problems, etc.; formulating targeted optimization suggestions according to the reasons for performance degradation, such as design improvement, manufacturing process optimization, maintenance strategy adjustment, etc., can directly improve the problem at the root; these optimization suggestions help to improve the performance, reliability, and service life of the circuit board, reduce maintenance costs and failure rates; feeding back the optimization suggestions to relevant personnel, such as the circuit board design team, manufacturing team, or maintenance team, helps them understand the existing problems and improvement directions of the current circuit board; the feedback optimization suggestions can be used as a reference for subsequent design and manufacturing work, promoting the continuous improvement and optimization of the circuit board performance; through the feedback mechanism, a virtuous cycle of continuous improvement can be established; technicians can continuously optimize and improve the design, manufacturing, and maintenance strategies of the circuit board based on the feedback optimization suggestions and practical experience to improve the overall performance level.

[0038] In one embodiment of the present invention, S41 includes: Deeply clean the historical data, including removing outliers, filling in missing values, data smoothing, etc., extract time series features, such as trend features, periodic features, seasonal features, etc., according to the key indicators of circuit board performance monitoring, such as temperature, voltage, current, etc., and construct a feature set; Correlation analysis is used to screen out the features that have a significant impact on the prediction of circuit board performance from the feature set, and according to the characteristics of the circuit board performance data, a time series analysis model is selected, such as the autoregressive integrated moving average model (ARIMA), long short-term memory network (LSTM); The selected model is trained using historical data, and the model is tested multiple times using K-fold cross-validation. According to the cross-validation results, the model parameters are adjusted; Based on the validation results, the optimal model is selected as the benchmark model for predicting the future performance of the circuit board; using the trained model, the performance change trends of the circuit board in the future for a period of time (such as one week) are predicted, including the temperature change trend, electrical parameter change trend, etc.

[0039] The working principle of the above technical solution is as follows: Identify and eliminate outliers in the data, which may be caused by sensor failures, data transmission errors, etc.; For missing values in the data, use appropriate methods for filling, such as interpolation method, mean filling method, etc., to ensure the integrity of the data; Smooth the data to reduce noise and fluctuations, and improve the accuracy and reliability of the data; Extract time series features according to the key indicators for monitoring the performance of the circuit board (such as temperature, voltage, current, etc.); The features include trend features (such as linear trend, non-linear trend, etc.), periodic features (such as periodic fluctuations, seasonal variations, etc.) and other relevant features; Construct a feature set to provide a basis for subsequent correlation analysis and model selection; Use correlation analysis (such as Pearson correlation coefficient, Spearman rank correlation coefficient, etc.) to screen out the features that have a significant impact on the prediction of the circuit board performance from the feature set; These features will be used as the main input variables for subsequent model training and prediction; Select a suitable time series analysis model according to the characteristics of the circuit board performance data (such as data stationarity, periodicity, non-linearity, etc.); Example models include autoregressive integrated moving average model (ARIMA) and long short-term memory network (LSTM). ARIMA is suitable for linear or stationary time series data, while LSTM is suitable for non-linear or complex time series data; Use historical data to train the selected model; During the training process, the model will learn the mapping relationship between the features and the target variable to achieve the prediction of the circuit board performance; Use K-fold cross-validation to test the model multiple times; Divide the data set into K subsets, and each time select one subset as the test set, and the remaining subsets as the training set for model training and testing; Through multiple tests, evaluate the stability and generalization ability of the model; Adjust the model parameters according to the cross-validation results; Optimize the parameters to improve the prediction accuracy and generalization ability of the model; Based on the cross-validation results, select the optimal model as the benchmark model for predicting the future performance of the circuit board; The optimal model should have the highest prediction accuracy and stability; Use the trained optimal model to predict the performance change trend of the circuit board in the future for a period of time (such as one week); The prediction results include key performance indicators such as temperature change trend and electrical parameter change trend.

[0040] The effects of the above technical solution are as follows: By steps such as removing outliers, filling in missing values, and data smoothing, the data quality is significantly improved, and the influence of noise and errors on subsequent analysis is reduced; this provides a more accurate and reliable data basis for subsequent time series feature extraction and model training; according to the key indicators of circuit board performance monitoring, time series features such as trend features, periodic features, and seasonal features are extracted; these features can comprehensively reflect the change laws and potential trends of circuit board performance, providing key information for subsequent correlation analysis and model selection; correlation analysis is used to screen out the features that have a significant impact on circuit board performance prediction from the feature set; this helps to reduce the number of input variables of the model, reduce the complexity of the model, and at the same time improve the prediction accuracy and stability of the model; according to the characteristics of circuit board performance data, suitable time series analysis models such as ARIMA and LSTM are selected; these models can capture the time dependence and non-linear relationships in circuit board performance data, providing strong support for the accurate prediction of circuit board performance; K-fold cross-validation is used to test the model multiple times to evaluate the stability and generalization ability of the model; the model parameters are adjusted according to the cross-validation results to optimize the prediction performance of the model; based on the verification results, the optimal model is selected as the benchmark model for predicting the future performance of the circuit board; this ensures the accuracy and reliability of the prediction results, providing a scientific basis for subsequent analysis of the reasons for performance decline and formulation of optimization suggestions; using the trained optimal model, the performance change trend of the circuit board in the future for a period of time (such as one week) is predicted; this helps technicians to timely understand the operating status and potential problems of the circuit board, providing strong support for preventive maintenance and fault troubleshooting; the entire technical solution realizes the automated processing of historical data and model training, significantly improving work efficiency; at the same time, intelligent prediction and decision support reduce manual intervention and subjective judgment, improving the accuracy and objectivity of prediction; through accurate prediction and timely maintenance, downtime and maintenance costs caused by circuit board failures can be avoided; in addition, the formulation and implementation of optimization suggestions can improve the design, manufacturing, and maintenance strategies of circuit boards, further reducing production costs and operating expenses.

[0041] In one embodiment of the present invention, S5 includes: S51. Display relevant information through an intuitive information visualization interface, where the relevant information includes detection results, abnormal information, performance prediction, and optimization suggestions; S52. Apply visualization elements to the interface to visually present data features and trends, where the visualization elements include charts (line charts, bar charts, pie charts, etc.), curve charts, heat maps, and 3D models; and automatically generate a detection report.

[0042] The working principle of the above technical solution is as follows: Integrate key information such as detection results, abnormal information, performance prediction, and optimization suggestions to ensure the accuracy and integrity of the information; Design an intuitive and user-friendly visual interface to enable users to quickly locate the required information; The interface layout is clear and the color matching is reasonable to ensure the readability and comprehensibility of the information; Display the integrated information through the visual interface, including the specific values of the detection results, the detailed descriptions of the abnormal information, the trend charts of the performance prediction, and the specific contents of the optimization suggestions; Select appropriate visual elements for display according to the type and characteristics of the information; Charts (such as line charts, bar charts, pie charts, etc.) are suitable for showing the comparison and trend changes of values; Curve charts are suitable for showing the change trends of continuous data; Heat maps are suitable for showing the distribution and correlation of data; 3D models are suitable for showing complex structures or spatial relationships; Apply the selected visual elements to the visual interface to ensure the reasonable and beautiful layout of the elements; Set appropriate interaction methods according to the hierarchical and logical relationships of the information, such as clicking, sliding, zooming, etc., to enable users to understand the data more deeply; Automatically generate a detection report according to the displayed information and visual elements; The report content includes but is not limited to the overview of the detection results, the analysis of the abnormal information, the conclusion of the performance prediction, and the detailed description of the optimization suggestions; The report format is standardized and the content is detailed, making it easy for users to understand and use; Through the intuitive visual interface, users can quickly obtain key information and reduce the decision-making time; The visual elements transform complex data into easy-to-understand graphics and images, reducing the user's understanding threshold; Users can reasonably allocate resources and optimize the work process according to the information on the visual interface; The intuitive and user-friendly interface design and rich interaction methods enhance the user experience.

[0043] The effects of the above technical solutions are as follows: Through an intuitive information visualization interface, key information such as detection results, abnormal information, performance prediction, and optimization suggestions is presented in a graphical manner, enabling users to quickly capture the key points they need; this intuitiveness helps users understand complex data and information faster, reducing the time and difficulty of information interpretation; the intuitive information display method enables users to make decisions quickly, especially in emergency situations, to quickly identify problems and take corresponding measures; this is of great significance for improving production efficiency, reducing failure rates, and optimizing resource allocation; using visualization elements such as charts (line charts, bar charts, pie charts, etc.), line graphs, heat maps, 3D models, etc., can clearly present the characteristics and trends of data; these elements can intuitively display the fluctuations, distributions, correlations, and spatial relationships of data, helping users better understand the internal laws and potential problems of data; through visualization elements, users can more easily identify abnormal fluctuations and potential trends in data, so as to take preventive measures in advance to avoid the occurrence of potential problems; at the same time, the visual display of performance prediction helps users understand future development trends and provides strong support for formulating long-term plans and optimization strategies; the automatic detection report generation function mentioned in step S52 can greatly improve the efficiency and accuracy of report generation; users do not need to manually organize data and write reports, and can automatically generate detection reports containing key information and visualization elements through the visualization interface; the automated report generation function not only saves users' time and energy, but also reduces the possibility of human errors and improves overall work efficiency; this is especially valuable for users who need to generate reports frequently; the intuitive information visualization interface and rich visualization elements provide users with a more convenient and efficient information acquisition method; users can choose appropriate visualization elements and interface layouts according to their own needs and preferences to obtain a better user experience.

[0044] An embodiment of the present invention, as Figure 2 shown, is a circuit board detection system, which includes: Data acquisition module: Collects the original data of the circuit experiment board through a high-precision sensing network. The original data includes static data and dynamic data; the static data includes the output data of each sensor, including static information such as surface topography, material composition, and electrical characteristics; the dynamic data includes the dynamic data of the sensors during the operation of the circuit board, such as temperature changes and electrical parameter fluctuations, and preprocesses the collected data. Feature extraction module: Extracts time-related features from the dynamic data, performs Fourier transform on the dynamic data to extract frequency-related features, and extracts spatial distribution features from the static data, such as surface roughness and material distribution. Model construction module: Based on machine learning algorithms, construct a circuit board status recognition model, train and validate the model using circuit board data with known statuses, and apply the trained model to perform anomaly detection on real-time collected data to identify potential defects or abnormal statuses; Suggestion generation module: Based on time series analysis algorithms, model historical data to predict the future performance change trends of the circuit board; According to the prediction results, combined with an expert system and a knowledge graph, automatically generate targeted optimization suggestions to guide the design and manufacturing process of the circuit board; Information display module: Display relevant information in a visual manner, where the relevant information includes detection results, anomaly information, performance predictions, and optimization suggestions; And automatically generate a detection report, where the monitoring report includes the detection time, detection conditions, detection results, anomaly information, and optimization suggestions.

[0045] The working principle of the above technical solution is as follows: Through a high-precision sensing network, perform comprehensive data collection on the circuit experiment board, including static data and dynamic data. Static data mainly reflects the physical and chemical characteristics of the circuit board, such as surface topography, material composition, and electrical characteristics, etc.; Dynamic data focuses on the real-time changes during the operation of the circuit board, such as temperature fluctuations, electrical parameter changes, etc. Extract key features from the preprocessed data for subsequent model training and anomaly detection. For dynamic data, extract time-related features (such as mean, variance, peak value, etc.) to capture the time change trends of the data; At the same time, use Fourier transform to extract frequency-related features to analyze the periodic components of the data. For static data, extract spatially distributed features, such as surface roughness, material distribution, etc., to reflect the physical structure characteristics of the circuit board; Based on machine learning algorithms, construct a circuit board status recognition model. Use circuit board data with known statuses to train the model so that it can accurately identify different statuses of the circuit board. After training, apply this model to perform anomaly detection on real-time collected data. By comparing the differences between real-time data and normal status data, identify potential defects or abnormal statuses; Based on time series analysis algorithms, model historical data to predict the future performance change trends of the circuit board. By analyzing the time series features in historical data, the laws of circuit board performance changes over time can be revealed. According to the prediction results, combined with an expert system and a knowledge graph, automatically generate targeted optimization suggestions, which are aimed at guiding the design and manufacturing process of the circuit board to improve the reliability and performance of the circuit board; Display relevant information such as detection results, anomaly information, performance predictions, and optimization suggestions in a visual way to facilitate users to intuitively understand the status and potential problems of the circuit board. At the same time, automatically generate a detection report, which details the detection time, detection conditions, detection results, anomaly information, and optimization suggestions, etc., providing an important reference for the subsequent processing and maintenance of the circuit board.

[0046] The effects of the above technical solutions are as follows: By collecting the original data of the circuit experiment board through a high-precision sensing network, more accurate and comprehensive data can be obtained, including static data and dynamic data. These data provide a solid foundation for subsequent anomaly detection and performance prediction; Preprocessing the collected data, such as denoising and standardization, can further improve the data quality, reduce the impact of noise and outliers on subsequent analysis, and thus enhance the accuracy and reliability of detection; Extracting time-related features and frequency-related features from dynamic data, as well as extracting spatial distribution features from static data, these features can comprehensively reflect the state and performance of the circuit board. The efficiency of feature extraction helps to accelerate the subsequent model training and anomaly detection; Constructing a circuit board state recognition model based on machine learning algorithms and using the circuit board data with known states for training and verification. This intelligent method can automatically identify potential defects or abnormal states, improving the detection efficiency and accuracy; Modeling the historical data based on time series analysis algorithms can predict the future performance change trend of the circuit board. This forward-looking prediction helps to timely discover potential problems and take corresponding measures to avoid or reduce the occurrence of faults; Combining an expert system with a knowledge graph to automatically generate targeted optimization suggestions. These suggestions can guide the design and manufacturing process of the circuit board, improve the reliability and performance of the circuit board, and reduce the maintenance cost; Displaying the relevant information in a visual way makes the detection results, anomaly information, performance prediction, optimization suggestions and other contents more intuitive and easy to understand. This helps users quickly understand the state and potential problems of the circuit board; Automatically generating a detection report, which details the detection time, detection conditions, detection results, anomaly information, optimization suggestions and other contents. This standardized report helps subsequent analysis and traceability, improving the work efficiency and accuracy.

[0047] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these changes and modifications.

Claims

1. A circuit board detection method, characterized in that: The method comprises: S1. Collecting raw data of the circuit experiment board through a high-precision sensor network, wherein the raw data includes static data and dynamic data; and preprocessing the collected data; S2, extract time-related features from dynamic data, perform Fourier transform on dynamic data, extract frequency-related features, and extract spatial distribution features from static data; S3. Based on the machine learning algorithm, a circuit board status recognition model is constructed. The model is trained and verified using circuit board data with known status. The trained model is then used to perform anomaly detection on the real-time collected data to identify potential defects or abnormal status. S4. Based on the time series analysis algorithm, historical data is modeled to predict the future performance trend of the circuit board; based on the prediction results, the expert system and knowledge graph are combined to automatically generate targeted optimization suggestions; S5. Display relevant information in a visual way and automatically generate a test report.

2. A circuit board detection method according to claim 1, characterized in that: Said S1 comprises: S11, selecting sensors according to the structural characteristics, working environment and detection requirements of the circuit board, wherein the sensors include infrared temperature sensors, high-precision resistance measurement sensors, capacitance measurement sensors and inductance measurement sensors; S12, connecting the selected sensors to the data acquisition system via wired or wireless means to form a sensor network; and calibrating the sensor network; S13, when the circuit board is in a non-working state or a stable state, start the sensor network to collect static data of each sensor, and monitor and record the dynamic data of the sensor in real time during the operation of the circuit board; S14. Clean the collected raw data to remove invalid data, abnormal data or duplicate data, and use filtering algorithms or denoising techniques to eliminate the impact of environmental noise or instrument errors on the data; S15. Normalize the data, convert the data into a unified dimension and range, and standardize the data to eliminate measurement deviations between different sensors.

3. A circuit board detection method according to claim 1, characterized in that: The S2 comprises: S21, extracting time-related features from dynamic data, and analyzing the changing rules of the time-related features over time to obtain dynamic behavior information of the circuit board during operation; S22, performing Fourier transform or wavelet transform on the dynamic data, extracting frequency-related features, and analyzing the distribution law of the frequency-related features in the frequency domain to obtain the vibration and noise characteristics of the circuit board; S23, extracting spatial distribution features from static data, and analyzing the spatial distribution patterns of the spatial distribution features to obtain the microstructure and material properties of the circuit board; S24, performing image processing on the surface morphology data to extract corresponding features, which include edge features, texture features, and shape features; and analyzing the performance of the corresponding features on the image to obtain the surface morphology and structural characteristics of the circuit board.

4. A circuit board detection method according to claim 1, characterized in that: The S3 includes: S31, selecting features useful for circuit board status identification from the extracted features, and removing redundant or irrelevant features; and reducing the feature dimension by principal component analysis; S32, select a machine learning algorithm to train the model using circuit board data with known states, and adjust the model parameters. S33. Use the trained machine learning model to perform anomaly detection on the real-time collected data to identify potential defects or abnormal conditions; S34, setting an abnormality threshold, when the detected abnormality exceeds the threshold, triggering an alarm mechanism, outputting abnormality information; classifying and identifying the detected abnormality, and determining the type of abnormality; S35. Locate the specific location where the anomaly occurs based on the spatial distribution and time series information of the anomaly characteristics.

5. A circuit board detection method according to claim 4, characterized in that: The S33 comprises: Preprocess the circuit board status data collected in real time from sensors or other data sources; divide the continuous data stream into multiple time windows according to the typical time scale of circuit board status changes, and use the data in each window as a sample for model input; Build prediction models through a variety of machine learning algorithms and select the model combination with the best performance based on the cross-validation method; Calculate the residual between the actual observation value and the model prediction value, and track the residual change trend in real time based on the residual monitoring mechanism; Further analyze the residual data through deep learning models to automatically extract deep abnormal features; Use the trained deep learning model to perform pattern matching on the extracted abnormal features, identify common abnormal patterns, and make preliminary classifications; Based on the significance of abnormal features and the statistical characteristics of historical abnormal data; through the abnormal scoring system, each detected abnormal event is quantitatively scored; Prioritize detected abnormal events based on anomaly scores, and perform context-aware analysis of anomalies based on the circuit board’s working environment and operating status information; Based on the context information, the initially identified anomalies are refined and classified, and explanations of the causes of the anomalies are provided.

6. A circuit board detection method according to claim 4, characterized in that: The S35 comprises: Map the spatial distribution information of abnormal features onto the physical layout of the circuit board to form a feature heat map or distribution map; and perform trend analysis on the time series data of abnormal features to identify the change pattern of abnormalities over time; Apply pattern recognition algorithms to conduct in-depth mining of time series data and identify common abnormal patterns; Combining spatial distribution characteristics and time series trends, multi-source information fusion technology is used to conduct a comprehensive assessment of anomalies; based on the fused information, a high-precision positioning algorithm is used to accurately determine the specific location of the anomaly; According to the spatial distribution and time series trend of abnormal characteristics, the impact of abnormalities on the overall performance of the circuit board is evaluated, and based on historical data and abnormal characteristics, machine learning algorithms are used to predict the risk level that may be caused by abnormalities; Automatically generate exception handling suggestions based on exception location, impact scope assessment and risk prediction results.

7. A circuit board detection method according to claim 1, characterized in that: The S4 comprises: S41. Model the historical data based on the time series analysis algorithm; predict the future performance trend of the circuit board based on the model; S42. Evaluate the forecast results, analyze the sources of forecast errors and uncertainties, and quantify the uncertainty of the forecast results using confidence intervals; S43. Combine the expert system and knowledge graph to analyze the causes of circuit board performance degradation, and formulate targeted optimization suggestions based on the causes of performance degradation; S44, and feeding back the optimization suggestions to relevant personnel, wherein the relevant personnel include a circuit board design team, a manufacturing team or a maintenance team.

8. A circuit board detection method according to claim 7, characterized in that: The S41 includes: Deeply clean the historical data, extract time series features based on the key indicators of circuit board performance monitoring, and build feature sets; Correlation analysis is used to select features that have a significant impact on circuit board performance prediction from the feature set, and time series analysis model selection is performed based on the characteristics of circuit board performance data; Use historical data to train the selected model, and use K-fold cross-validation to test the model multiple times. According to the cross-validation results, adjust the model parameters; Based on the verification results, the optimal model is selected as the benchmark model for predicting the future performance of the circuit board; the trained model is used to predict the performance change trend of the circuit board in the future.

9. A circuit board detection method according to claim 1, characterized in that: The S5 comprises: S51. Displaying relevant information through an intuitive information visualization interface, wherein the relevant information includes test results, abnormal information, performance prediction, and optimization suggestions; S52. Apply visualization elements to the interface to intuitively present data features and trends, and automatically generate test reports.

10. A circuit board detection system, characterized in that: The system comprises: Data acquisition module: collects the original data of the circuit experiment board through a high-precision sensor network, the original data includes static data and dynamic data; and pre-processes the collected data; Feature extraction module: extract time-related features from dynamic data, perform Fourier transform on dynamic data to extract frequency-related features, and extract spatial distribution features from static data; Model building module: Based on machine learning algorithms, a circuit board status recognition model is built. The model is trained and verified using circuit board data with known status. The trained model is then used to perform anomaly detection on real-time collected data to identify potential defects or abnormal status. Suggestion generation module: Based on the time series analysis algorithm, historical data is modeled to predict the future performance trend of the circuit board; based on the prediction results, combined with the expert system and knowledge graph, targeted optimization suggestions are automatically generated; Information display module: displays relevant information in a visual way and automatically generates a test report.

Images