Internet-based automatic detection equipment remote diagnosis system

Abstract

The invention relates to the technical field of equipment fault diagnosis, and discloses an automatic detection equipment remote diagnosis system based on the Internet. A data acquisition module of the system acquires the running state and environmental parameter data flow of distributed detection equipment in real time through the Internet; the abnormal feature recognition module is used for performing multi-dimensional analysis on the state data flow based on the fault judgment rule base to generate an equipment abnormal feature map; the diagnosis strategy generation module combines a diagnosis knowledge graph and strategy effectiveness evaluation, fuses abnormal features and environment parameters, and outputs an initial diagnosis strategy set; the state prediction module is used for carrying out time sequence evolution analysis on the abnormal characteristics and the environmental parameters to obtain equipment state prediction characteristics and future environmental parameter prediction values; the strategy optimization module is used for dynamically adjusting and correcting the initial diagnosis strategy set according to the prediction result and generating an optimized diagnosis strategy set; and the remote execution module is used for converting the optimization strategy into a diagnosis control instruction through the Internet and sending the diagnosis control instruction to corresponding equipment.

Description

Technical Field

[0001] This invention relates to the field of equipment fault diagnosis technology, specifically to an Internet-based automated remote diagnostic system for testing equipment. Background Technology

[0002] Current remote diagnostics of automated testing equipment primarily employs a combination of periodic data uploads and manual analysis. Existing technologies have limited analytical dimensions for equipment operating status data, failing to achieve systematic extraction of multi-dimensional anomaly features. Fault determination rule bases are outdated and unable to adapt to new fault modes. The diagnostic strategy generation process is rigid, failing to fully consider the impact of environmental parameters. Status prediction functionality is lacking, resulting in a lack of forward-looking judgment on equipment fault development trends. Strategy optimization mechanisms are simplistic and cannot dynamically adjust based on prediction results. Existing systems need to address key technical challenges such as multi-source data fusion, intelligent feature extraction, accurate predictive analysis, and adaptive strategy optimization.

[0003] Traditional remote diagnostic systems suffer from significant shortcomings in accuracy and foresight. Inconsistent data acquisition frequencies lead to severe loss of key state features. Abnormal feature extraction methods are limited, resulting in low recognition rates for complex fault modes. Fault rule bases are finite in size, with incomplete coverage of special operating conditions. Insufficient quantification of the impact of environmental parameters leads to poorly targeted diagnostic strategies. Time-series prediction models lack accuracy, resulting in frequent false alarms and missed alarms. Inflexible strategy adjustment mechanisms fail to adapt to dynamically changing needs. Summary of the Invention

[0004] The purpose of this invention is to provide an Internet-based automated testing equipment remote diagnostic system to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an Internet-based automated remote diagnostic system for testing equipment, the system comprising: The data acquisition module is used to collect real-time data streams of equipment operating status and equipment environmental parameters from multiple distributed automated testing devices via the Internet. An anomaly feature recognition module is used to extract multi-dimensional anomaly features from the equipment operating status data stream based on a preset equipment fault judgment rule library, and generate an equipment anomaly feature map. The diagnostic strategy generation module, based on the built-in diagnostic knowledge graph and strategy effectiveness evaluation criteria, integrates the device anomaly feature graph and the device environmental parameter data stream to output an initial diagnostic strategy set; The state prediction module is used to perform time-series evolution analysis on the equipment anomaly feature map and the equipment environmental parameter data stream to obtain equipment state prediction features and future environmental parameter prediction values. The strategy optimization module is used to dynamically adjust and correct the initial diagnostic strategy set by combining the device status prediction features and the future environmental parameter prediction values, and generate an optimized diagnostic strategy set. The remote execution module is used to send diagnostic control commands to the corresponding automated testing equipment via the Internet according to the optimized diagnostic strategy set.

[0006] Preferably, the anomaly feature recognition module is used to extract multi-dimensional anomaly features from the equipment operating status data stream based on a preset equipment fault judgment rule base, generating an equipment anomaly feature map, including: Obtain the structural configuration information and historical operation logs of the target automated testing equipment; A virtual simulation model of the device is constructed based on the structural configuration information and historical operation logs; The real-time collected equipment operating status data stream is mapped to the equipment virtual simulation model to generate a three-dimensional cloud map of the equipment operating status; Based on the equipment fault determination rule base, pattern recognition is performed on the three-dimensional cloud map of the equipment's operating status to generate the equipment's abnormal feature map.

[0007] Preferably, the construction process of the anomaly feature recognition module includes: Select a group of devices with the same model as the target automated testing equipment as the reference equipment set; Load the database of group anomaly characteristics recorded by the reference device set during the historical period; Load the database of individual abnormal features recorded by the target automated detection equipment within a historical period; An initial anomaly identification model is trained based on the aforementioned group anomaly feature database; The initial anomaly identification model is optimized for feature recognition accuracy using the individual anomaly feature database to form the final deployed anomaly feature recognition module.

[0008] Preferably, the step of optimizing the feature recognition accuracy of the initial anomaly recognition model using the individual anomaly feature database to form the finally deployed anomaly feature recognition module includes: The initial anomaly recognition model was validated and tested using the individual anomaly feature database to obtain the model recognition accuracy index. Determine whether the model recognition accuracy index is lower than a preset accuracy threshold; When the model recognition accuracy index is lower than the preset accuracy threshold, the parameters of the initial anomaly recognition model are fine-tuned using the individual anomaly feature database until the model recognition accuracy index reaches or exceeds the accuracy threshold, thereby forming the anomaly feature recognition module.

[0009] Preferably, the diagnostic strategy generation module, based on a built-in diagnostic knowledge graph and strategy effectiveness evaluation criteria, integrates the device anomaly feature graph and the device environmental parameter data stream to output an initial diagnostic strategy set, including: Based on the device anomaly feature map and the device environmental parameter data stream, the diagnostic knowledge graph is subjected to context association matching to obtain an adapted diagnostic knowledge subgraph; In the adaptive diagnostic knowledge subgraph, identify the policy triggering condition interval and determine the effective policy generation space; Based on the aforementioned effective strategy generation space, candidate diagnostic strategies are generated; Calculate the strategy effectiveness score of the candidate diagnostic strategy, and determine whether the strategy effectiveness score meets the strategy effectiveness evaluation criteria; When the strategy effectiveness score meets the strategy effectiveness evaluation criteria, the candidate diagnostic strategy is included in the initial diagnostic strategy set.

[0010] Preferably, the step of performing context-based matching on the diagnostic knowledge graph based on the device anomaly feature graph and the device environmental parameter data stream to obtain an adapted diagnostic knowledge subgraph includes: Traverse the diagnostic knowledge graph and read the historical diagnostic case records stored therein. The historical diagnostic case records include historical equipment abnormality feature information, historical environmental parameter data, and historical diagnostic strategies that have been executed. Calculate the feature similarity between the equipment anomaly feature map and the historical equipment anomaly feature information; Calculate the environmental parameter similarity between the device environmental parameter data stream and the historical environmental parameter data; The feature similarity and the environmental parameter similarity are weighted and fused according to predefined fusion rules to generate case correlation. Determine whether the case correlation degree reaches or exceeds the case correlation threshold; When the case correlation reaches or exceeds the case correlation threshold, the knowledge nodes associated with the corresponding historical diagnostic case records are extracted and used to form the adaptive diagnostic knowledge subgraph.

[0011] Preferably, the strategy optimization module is used to dynamically adjust and correct the initial diagnostic strategy set by combining the device state prediction features and the predicted values ​​of future environmental parameters, to generate an optimized diagnostic strategy set, including: Based on the predicted features of the equipment status and the predicted values ​​of the future environmental parameters, a prospective association matching is performed on the diagnostic knowledge graph to obtain a predictive diagnostic knowledge subgraph. In the predictive diagnostic knowledge subgraph, identify the interval of predictive policy triggering conditions and determine the predictive policy generation space; Based on the strategy effectiveness evaluation criteria and the predictive strategy generation space, an iterative strategy search is performed to generate a predictive diagnostic strategy set; The predictive diagnostic strategy set is fused and conflict-resolved with the initial diagnostic strategy set to generate the optimized diagnostic strategy set.

[0012] Preferably, the state prediction module is used to perform time-series evolution analysis on the equipment anomaly feature map and the equipment environmental parameter data stream to obtain equipment state prediction features and future environmental parameter prediction values, including: The abnormal feature map of the equipment is decomposed into time series to extract the feature evolution trend components; Periodically analyze the data stream of the equipment's environmental parameters to extract the patterns of change in these parameters; Using a trained time-series prediction model, joint reasoning is performed on the feature evolution trend components and the environmental parameter change patterns to predict the equipment state prediction features and the future environmental parameter prediction values.

[0013] Preferably, the data acquisition module is used to acquire real-time data streams of equipment operating status and equipment environmental parameters from multiple distributed automated testing devices via the Internet, including: Establish secure communication connections with various automated testing devices through the device gateway protocol; The device's sensor readings, operation logs, and alarm information are read according to the preset sampling frequency to form raw operation data; The system collects temperature, humidity, and vibration intensity data at the location of the device through an environmental sensor node network to form raw environmental data. The original operating data and the original environmental data are subjected to data verification, filtering, and format unification processing to generate standardized data streams of the device operating status and the device environmental parameters.

[0014] Preferably, the remote execution module is used to send diagnostic control commands to the corresponding automated testing equipment via the Internet according to the optimized diagnostic strategy set, including: The optimized diagnostic strategy set is parsed into a sequence of operation instructions that the device can recognize; The sequence of operation instructions is sent to the target automated detection equipment via a secure communication link; Receive instruction execution status feedback information from the target automated inspection equipment; The instruction execution status feedback information is sent back to the data acquisition module to update the device operating status data stream.

[0015] Compared with the prior art, the beneficial effects of the present invention are: Based on a pre-defined equipment fault diagnosis rule base, multi-dimensional anomaly features are extracted from the equipment operating status data stream to generate an equipment anomaly feature map. The fault diagnosis rule base includes historical fault cases and expert experience, providing a basis for judgment. Anomaly feature extraction employs a multi-scale analysis method to capture fault features at different granularities. The feature map is represented by a graph structure, with nodes representing anomaly patterns and edges representing anomaly relationships. The map is dynamically updated to reflect the equipment health status in real time. Through multi-dimensional feature extraction, accurate fault identification is achieved.

[0016] An initial diagnostic strategy set is output by fusing equipment anomaly feature graphs and equipment environmental parameter data streams based on a built-in diagnostic knowledge graph and strategy effectiveness evaluation criteria. The diagnostic knowledge graph stores the correlation between fault causes and solutions. The strategy effectiveness evaluation criteria are dynamically optimized based on historical execution results. The fusion process considers the influence weight of environmental parameters on faults. The initial strategy set contains multiple processing schemes, supporting multi-objective optimization. Through intelligent fusion, the relevance of diagnostic strategies is improved.

[0017] Time-series evolution analysis is performed on equipment anomaly characteristic maps and equipment environmental parameter data streams to obtain equipment state prediction characteristics and future environmental parameter prediction values. The time-series analysis employs time series prediction algorithms to infer state change trends. Predicted characteristics include key indicators such as fault development speed and impact range. Environmental parameter predictions consider seasonal and periodic variations. Confidence intervals are included in the prediction results to improve reliability. Through time-series prediction, forward-looking judgments on fault development can be achieved.

[0018] By combining equipment status prediction features and future environmental parameter predictions, the initial diagnostic strategy set is dynamically adjusted and corrected to generate an optimized diagnostic strategy set. The adjustment algorithm considers the difference between the prediction results and the current state. The correction process balances processing efficiency and resource consumption. The optimized strategy set supports priority ranking to improve execution efficiency. Through dynamic optimization, the adaptability of the diagnostic strategies is enhanced. Attached Figure Description

[0019] Figure 1 This is a schematic diagram illustrating the working principle of the Internet-based automated testing equipment remote diagnostic system described in this invention. Figure 2 A flowchart illustrating the core operation of the anomaly feature recognition module; Figure 3 A flowchart generated for the initial diagnostic strategy set; Figure 4 Optimization analysis diagram for equipment status prediction and diagnosis strategies; Figure 5 This is a diagram illustrating data acquisition, monitoring, and analysis of automated testing equipment. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 The present invention provides an Internet-based remote diagnostic system for automated testing equipment. The system includes a data acquisition module, an anomaly feature recognition module, a diagnostic strategy generation module, a status prediction module, a strategy optimization module, and a remote execution module.

[0022] The data acquisition module collects real-time data streams of equipment operating status and environmental parameters from distributed automated testing equipment via the internet. These data streams are then transmitted to the anomaly feature identification module. Based on a pre-defined equipment fault judgment rule base, the anomaly feature identification module extracts multi-dimensional anomaly features from the equipment operating status data streams, generating an anomaly feature map. The diagnostic strategy generation module, based on a built-in diagnostic knowledge graph and strategy effectiveness evaluation criteria, integrates the anomaly feature map and the equipment environmental parameter data streams to output an initial diagnostic strategy set. The status prediction module performs time-series evolution analysis on the anomaly feature map and the equipment environmental parameter data streams to obtain predicted equipment status features and predicted future environmental parameters. The strategy optimization module dynamically adjusts and corrects the initial diagnostic strategy set based on the predicted equipment status features and predicted future environmental parameters, generating an optimized diagnostic strategy set. The remote execution module sends diagnostic control commands to the corresponding automated testing equipment via the internet according to the optimized diagnostic strategy set, completing the remote diagnostic process.

[0023] Example 1: See Figure 2In practical implementation, the anomaly feature identification module constructs a virtual simulation model of the target automated inspection equipment by acquiring its structural configuration information and historical operation logs. The structural configuration information includes the equipment's mechanical layout, sensor types, and connection topology, while the historical operation logs contain performance parameters and event sequences recorded during past operating cycles. The process of constructing the virtual simulation model based on the structural configuration information and historical operation logs involves using 3D modeling software to convert the structural configuration information into a geometric model and importing data points from the historical operation logs as the initial state of the model, thereby generating a digital twin capable of simulating the equipment's real-time operating behavior. When the real-time acquired equipment operating status data stream is mapped to the virtual simulation model, sensor readings in the data stream are mapped to corresponding nodes in the model. A 3D cloud map of the equipment's operating status is generated by the rendering engine, visually displaying the dynamic operating status of each component of the equipment. Based on the equipment fault determination rule base, pattern recognition is performed on the 3D cloud map of the equipment operating status. The equipment fault determination rule base stores feature templates for various fault modes. The matching degree between the 3D cloud map and the feature template is calculated by the image recognition algorithm. When the matching degree exceeds the threshold, the abnormal area is extracted and an abnormal feature map of the equipment is generated. This map records the type, location and intensity of the abnormal features in the form of graph structure data.

[0024] In some embodiments, the construction process of the anomaly feature recognition module includes selecting a group of devices of the same model as the target automated inspection device as a reference device set. The reference device set contains complete records of the operation of multiple devices of the same model within a historical period. A group anomaly feature database recorded by the reference device set within the historical period is loaded. This database summarizes common anomaly patterns and their frequencies within the device group. An individual anomaly feature database recorded by the target automated inspection device within the historical period is loaded. This database focuses on specific operational deviations and custom alarm events for that particular device. An initial anomaly recognition model is trained based on the group anomaly feature database. The training process uses machine learning algorithms such as convolutional neural networks, employing group anomaly features as training samples to learn general anomaly patterns. The feature recognition accuracy of the initial anomaly recognition model is optimized using the individual anomaly feature database. By fine-tuning the model parameters to adapt to the uniqueness of the target device, the final deployed anomaly feature recognition module is formed.

[0025] It is understandable that the specific steps for optimizing the feature recognition accuracy of the initial anomaly recognition model using an individual anomaly feature database include: validating the initial anomaly recognition model using the individual anomaly feature database; obtaining model recognition accuracy metrics such as accuracy or F1 score by inputting individual anomaly feature data and comparing the model output with the true labels; determining whether the model recognition accuracy metric is lower than a preset accuracy threshold, which is set according to the security requirements of the application scenario; and fine-tuning the parameters of the initial anomaly recognition model using the individual anomaly feature database, which uses the gradient descent algorithm to adjust the network weights. This iterative process continues until the model recognition accuracy metric reaches or exceeds the accuracy threshold, thus forming the anomaly feature recognition module. Optionally, the accuracy metric can be calculated using the following formula: Where: A represents accuracy, TP represents the number of true positive samples, TN represents the number of true negative samples, FP represents the number of false positive samples, and FN represents the number of false negative samples. This formula is used to quantify the model's performance on the validation set and guide the direction of parameter fine-tuning.

[0026] In some embodiments, the construction of the virtual simulation model of the device can integrate a physics engine to simulate the dynamic behavior of the device, enhancing the realism of the 3D cloud map. Optionally, multi-scale feature extraction techniques can be employed in the pattern recognition stage to capture anomalies at different granularities. It is understood that the update of the individual anomaly feature database can continue as the device operates, ensuring model adaptability.

[0027] Example 2: See Figure 3In practical implementation, the diagnostic strategy generation module performs contextual matching on the diagnostic knowledge graph based on the equipment anomaly feature graph and the equipment environmental parameter data stream. The equipment anomaly feature graph represents anomaly types and their relationships in the form of graph nodes and edges. The equipment environmental parameter data stream contains time-series data of temperature, humidity, and vibration intensity. The process of contextual matching on the diagnostic knowledge graph involves traversing the historical diagnostic case records stored in the diagnostic knowledge graph. These historical diagnostic case records contain historical equipment anomaly feature information, historical environmental parameter data, and historically executed diagnostic strategies. The feature similarity between the equipment anomaly feature graph and historical equipment anomaly feature information is calculated, involving a comparison of the topological similarity of the graph structure and the similarity of node attributes. The environmental parameter similarity between the equipment environmental parameter data stream and historical environmental parameter data is calculated, assessing the morphological differences between the current environmental data sequence and historical sequences. The feature similarity and environmental parameter similarity are weighted and fused according to predefined fusion rules to generate a case correlation score. The case correlation score numerically represents the degree of matching between the current scenario and historical cases. Finally, it is determined whether the case correlation score reaches or exceeds a case correlation threshold, which is dynamically set according to the diagnostic accuracy requirements. When the case correlation reaches or exceeds the case correlation threshold, the knowledge nodes associated with the corresponding historical diagnostic case records are extracted. The knowledge nodes include the cause of the fault, the handling method, and the expected effect, which together constitute the adaptive diagnostic knowledge subgraph.

[0028] In some embodiments, policy triggering condition intervals are identified in the adaptation diagnostic knowledge subgraph, and these intervals are defined by rule prerequisites in the knowledge nodes. An effective policy generation space is determined; this space is the set of policy solutions that satisfy all constraints. Candidate diagnostic policies are generated based on this space, formed by combining operation instructions from the adaptation diagnostic knowledge subgraph. The policy effectiveness score of each candidate diagnostic policy is calculated, based on a comprehensive evaluation of its historical success rate and implementation cost. The policy effectiveness score is then assessed to determine if it meets the policy effectiveness evaluation criteria, which set a minimum allowable score. When the policy effectiveness score meets the criteria, the candidate diagnostic policy is included in the initial diagnostic policy set, which serves as the basis for subsequent optimization.

[0029] It is understandable that the calculation of case relevance can employ a multi-attribute decision-making method. Optionally, the weighted fusion formula can be expressed as: Where: R represents the case relevance. Weight coefficients representing feature similarity This represents the calculated value of feature similarity. Weighting coefficients representing the similarity of environmental parameters. This represents the calculated similarity value of environmental parameters. Weighting coefficients. and weighting coefficients Assigned according to feature importance, and satisfying + =1 constraint condition.

[0030] In some embodiments, feature similarity calculation can employ a graph neural network embedding method, mapping the device anomaly feature map and historical device anomaly feature information into vectors before calculating cosine similarity. Optionally, environmental parameter similarity calculation can employ a dynamic time warping algorithm to eliminate the influence of time series scaling and distortion. It is understood that the identification of policy triggering condition intervals can be combined with fuzzy logic reasoning to handle condition judgments with unclear boundaries. The construction process of the adaptive diagnostic knowledge subgraph can introduce a caching mechanism to store high-frequency matching results to improve response speed.

[0031] Example 3: In specific implementation, the strategy optimization module performs prospective association matching on the diagnostic knowledge graph based on equipment status prediction features and future environmental parameter prediction values. Equipment status prediction features include predicted performance degradation values ​​of key equipment components at future time points, and future environmental parameter prediction values ​​include forecast data for temperature and humidity. The prospective association matching process involves scanning knowledge nodes with time-constrained relationships within the diagnostic knowledge graph. These knowledge nodes contain fault evolution paths and environmental impact factors. Predictive strategy triggering condition intervals are identified in the prospective association matching process. These intervals are jointly defined by equipment status thresholds and environmental parameter limits at future time points. A predictive strategy generation space is determined, which is the set of all preventative strategies that satisfy potential future fault conditions. Iterative strategy search is performed based on strategy effectiveness evaluation criteria and the predictive strategy generation space. The iterative strategy search uses a genetic algorithm to find the optimal solution within the strategy space. The implementation process of the genetic algorithm first randomly initializes a set of candidate strategies from the predictive strategy generation space as an initial population, with each candidate strategy representing a possible diagnostic scheme. Subsequently, the fitness of each individual in the population is evaluated using a strategy effectiveness assessment criterion. This criterion is dynamically adjusted based on the execution effects of historical diagnostic cases, and the evaluation factors include the historical success rate of the strategy, implementation cost, and expected results. After the fitness assessment, a selection operator is used to preferentially retain strategy individuals with high fitness, and a crossover operator is used to exchange and combine the parameters of the selected strategy pairs to generate new offspring strategies to explore the solution space. A predictive diagnostic strategy set is generated, which contains a sequence of maintenance instructions for predictive faults. The predictive diagnostic strategy set is then fused with the initial diagnostic strategy set, and conflict resolution is performed. The strategy fusion and conflict resolution employs a multi-objective optimization method to balance the needs of immediate processing and preventive maintenance, generating an optimized diagnostic strategy set.

[0032] The status prediction module performs time-series evolution analysis on the equipment anomaly feature map and the equipment environmental parameter data stream. The equipment anomaly feature map provides time stamps for historical anomaly patterns, while the equipment environmental parameter data stream provides continuous environmental monitoring values. The equipment anomaly feature map is decomposed into time series components using a seasonal decomposition method to extract feature evolution trend components, which reflect the long-term direction of change in anomaly features. The equipment environmental parameter data stream is analyzed periodically, using Fourier transform to extract the patterns of environmental parameter changes, revealing diurnal fluctuations in temperature or seasonal fluctuations in humidity. A trained time-series prediction model is used to jointly infer the feature evolution trend components and environmental parameter change patterns. The time-series prediction model is a Long Short-Term Memory (LSTM) network model, and the joint inference process uses the feature evolution trend components and environmental parameter change patterns as parallel inputs. The module predicts equipment status features and future environmental parameter values. The equipment status prediction features include a bearing wear prediction curve, and the future environmental parameter prediction values ​​include a 24-hour working chamber temperature prediction curve.

[0033] In some embodiments, forward-looking association matching may introduce a time decay factor, assigning higher weights to recent knowledge nodes. Optionally, the policy fusion and conflict resolution weights can be calculated using the following formula: in: Indicates the policy fusion weights, This indicates the priority coefficient of the immediate processing strategy. This indicates the confidence level of the initial diagnostic strategy set. This represents the confidence level of the strategies in the predictive diagnostic strategy set. The strategy fusion weight W is used to determine the priority of strategy selection during conflict resolution.

[0034] It is understood that time series decomposition can employ the STL algorithm to decompose the equipment anomaly feature map into trend, seasonal, and residual terms. In some embodiments, the time series prediction model is trained using aligned sequences of historical equipment operating data and corresponding environmental data. Optionally, periodicity analysis can be combined with wavelet transform to capture the multi-scale periodic features of environmental parameters.

[0035] See Figure 4 This paper demonstrates the optimization process of equipment condition prediction and diagnostic strategies based on time-series evolution analysis. The charts include performance degradation prediction curves for key equipment components, reflecting the changing trend of bearing wear over time. Combined with future predictions of ambient temperature, this provides data support for preventative maintenance strategies. Through the fusion and analysis of multi-dimensional data, the system can generate optimized diagnostic strategies to ensure reliable equipment operation under complex conditions.

[0036] Example 4: In specific implementation, the data acquisition module establishes a secure communication connection with each automated testing device through a device gateway protocol, including industry standard protocols such as ModbusTCP, OPCUA, or MQTT. The process of establishing a secure communication connection involves authentication and communication encryption. Authentication uses digital certificates to verify device identity, and communication encryption uses the TLS protocol to encrypt transmitted data. The module reads sensor readings, operation logs, and alarm information from the devices according to a preset sampling frequency. The preset sampling frequency is configured based on the device type and monitoring requirements. Sensor readings include analog or digital signals such as voltage, current, temperature, and pressure. The operation log records device start / stop events, operation mode switching, and error codes. Alarm information includes real-time alarms triggered by the device's own diagnostic system. This forms raw operating data, which is a collection of unprocessed, timestamped raw data points.

[0037] Temperature, humidity, and vibration intensity data at the device's location are collected via an environmental sensor node network. This network consists of wireless sensor nodes deployed around the device, which aggregate the collected data to a gateway via Zigbee or LoRaWAN protocols. This forms raw environmental data, which consists of asynchronous measurements from sensor nodes at different physical locations. The raw operational and environmental data undergo data verification, filtering, and format standardization. Data verification checks the completeness and rationality of the data, filtering out outliers that significantly exceed the physical measurement range. Filtering uses digital filters to smooth high-frequency noise in the data. Format standardization converts data from different sources into a standard format with the same time base and data structure. Standardized device operating status data streams and device environmental parameter data streams are generated, and these standardized data streams serve as input for subsequent modules.

[0038] In some embodiments, the selection of the device gateway protocol can be dynamically adapted based on the device interface type and network conditions. Optionally, a secure communication connection can be established using virtual private network technology to construct an encrypted tunnel. It is understood that the data verification process can apply a cyclic redundancy check algorithm to verify the integrity of the data frame, and its checksum calculation formula is: Where: CRC represents Cyclic Redundancy Check, Data represents the original data bit sequence to be checked, Generator represents the preset generator polynomial, n represents the order of the generator polynomial, and Remainder represents the remainder operation. This check code is used at the receiving end to compare and detect data transmission errors. For the device gateway protocols supported by the data acquisition module and their typical application scenarios, please refer to Table 1. Table 1: Device Gateway Protocol Types and Applications In some embodiments, the preset sampling frequency can be dynamically adjusted according to the criticality level of the equipment, with a higher sampling frequency used for core equipment. Optionally, redundant nodes can be deployed in the environmental sensor node network to improve the reliability of data acquisition. It is understood that format unification processing includes timestamp synchronization and data unit standardization to ensure the consistency of multi-source data fusion. A Kalman filter can be used for filtering to optimize signal quality in the presence of noise.

[0039] See Figure 5 This demonstrates the effectiveness of the data acquisition module, showcasing the real-time collection of equipment operating parameters and environmental monitoring data via the device gateway protocol. The figure displays real-time changes in key operating parameters such as voltage and current, as well as monitored values ​​for ambient temperature and humidity, demonstrating the system's ability to acquire and process multi-source heterogeneous data. The data quality score curve reflects the reliability and completeness of the acquired data, providing a high-quality data foundation for subsequent anomaly detection and fault diagnosis.

[0040] Example 5: In specific implementation, the remote execution module parses the optimized diagnostic strategy set into a sequence of operation instructions recognizable by the device. The optimized diagnostic strategy set includes diagnostic steps and parameters in text format. The parsing process uses an instruction compiler to convert high-level strategy descriptions into low-level device control commands. The sequence of operation instructions is an ordered list of binary or structured text commands conforming to the device communication protocol. The sequence of operation instructions is sent to the target automated detection device through a secure communication link. The secure communication link uses an encrypted channel based on the TLS 1.3 protocol, and the sending process follows a request-confirmation mechanism to ensure instruction delivery. The module receives instruction execution status feedback information from the target automated detection device. This feedback information includes instruction reception confirmation, execution progress percentage, execution success flag, or error code. The instruction execution status feedback information is then sent back to the data acquisition module. This feedback information is encapsulated into a data packet of a specific format and used to update the real-time control status field in the device operating status data stream.

[0041] In some embodiments, the parsing rules of the instruction compiler are predefined based on the device model. Optionally, the operation instruction sequence may include a checksum field to ensure instruction integrity. It is understood that establishing a secure communication link requires two-way certificate authentication to prevent unauthorized access. The process of receiving instruction execution status feedback information is equipped with a timeout retransmission mechanism to handle network latency or packet loss. After the instruction execution status feedback information is sent back to the data acquisition module, it triggers an immediate update of the device operating status data stream.

[0042] The mapping relationship between the optimized diagnostic strategy set and the sequence of operation instructions can be represented by the following formulaic rules: in: This represents the i-th operation instruction. This represents the j-th strategy entry in the set of optimization diagnostic strategies. Let f represent the hardware specification description of the target automated detection device k, and let f represent the compilation process of mapping the policy to specific instructions according to the device specification. This formula describes the transformation relationship from abstract policy to specific instructions.

[0043] In some embodiments, the operation command sequence can support a batch issuance mode to improve the efficiency of operating multiple devices. Optionally, the command execution status feedback information can include a timestamp and executor number for accurate tracking of the command execution process. It is understood that the secure communication link can adaptively adjust the transmission rate and encryption strength under different network environments. The command compiler can implement version management to ensure compatibility with devices of different firmware versions. The command execution status feedback information returned to the data acquisition module will be correlated and matched with the original command to ensure the accuracy of status updates.

[0044] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0045] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A remote diagnostic system for automated testing equipment based on the Internet, characterized in that, The system includes: The data acquisition module is used to collect real-time data streams of equipment operating status and equipment environmental parameters from multiple distributed automated testing devices via the Internet. An anomaly feature recognition module is used to extract multi-dimensional anomaly features from the equipment operating status data stream based on a preset equipment fault judgment rule library, and generate an equipment anomaly feature map. The diagnostic strategy generation module, based on the built-in diagnostic knowledge graph and strategy effectiveness evaluation criteria, integrates the device anomaly feature graph and the device environmental parameter data stream to output an initial diagnostic strategy set; The state prediction module is used to perform time-series evolution analysis on the equipment anomaly feature map and the equipment environmental parameter data stream to obtain equipment state prediction features and future environmental parameter prediction values. The strategy optimization module is used to dynamically adjust and correct the initial diagnostic strategy set by combining the device status prediction features and the future environmental parameter prediction values, and generate an optimized diagnostic strategy set. The remote execution module is used to send diagnostic control commands to the corresponding automated testing equipment via the Internet according to the optimized diagnostic strategy set.

2. The Internet-based remote diagnostic system for automated testing equipment as described in claim 1, characterized in that, The anomaly feature recognition module is used to extract multi-dimensional anomaly features from the equipment operating status data stream based on a preset equipment fault judgment rule base, and generate an equipment anomaly feature map, including: Obtain the structural configuration information and historical operation logs of the target automated testing equipment; A virtual simulation model of the device is constructed based on the structural configuration information and historical operation logs; The real-time collected equipment operating status data stream is mapped to the equipment virtual simulation model to generate a three-dimensional cloud map of the equipment operating status; Based on the equipment fault determination rule base, pattern recognition is performed on the three-dimensional cloud map of the equipment's operating status to generate the equipment's abnormal feature map.

3. The Internet-based automated testing equipment remote diagnostic system as described in claim 1, characterized in that, The construction process of the anomaly feature recognition module includes: Select a group of devices with the same model as the target automated testing equipment as the reference equipment set; Load the database of group anomaly characteristics recorded by the reference device set during the historical period; Load the database of individual abnormal features recorded by the target automated detection equipment within a historical period; An initial anomaly identification model is trained based on the aforementioned group anomaly feature database; The initial anomaly identification model is optimized for feature recognition accuracy using the individual anomaly feature database to form the final deployed anomaly feature recognition module.

4. The Internet-based remote diagnostic system for automated testing equipment as described in claim 3, characterized in that, The step of optimizing the feature recognition accuracy of the initial anomaly recognition model using the individual anomaly feature database to form the final deployed anomaly feature recognition module includes: The initial anomaly recognition model was validated and tested using the individual anomaly feature database to obtain the model recognition accuracy index. Determine whether the model recognition accuracy index is lower than a preset accuracy threshold; When the model recognition accuracy index is lower than the preset accuracy threshold, the parameters of the initial anomaly recognition model are fine-tuned using the individual anomaly feature database until the model recognition accuracy index reaches or exceeds the accuracy threshold, thereby forming the anomaly feature recognition module.

5. The Internet-based remote diagnostic system for automated testing equipment as described in claim 1, characterized in that, The diagnostic strategy generation module, based on a built-in diagnostic knowledge graph and strategy effectiveness evaluation criteria, integrates the device anomaly feature graph and the device environmental parameter data stream to output an initial diagnostic strategy set, including: Based on the device anomaly feature map and the device environmental parameter data stream, the diagnostic knowledge graph is subjected to context association matching to obtain an adapted diagnostic knowledge subgraph; In the adaptive diagnostic knowledge subgraph, identify the policy triggering condition interval and determine the effective policy generation space; Based on the aforementioned effective strategy generation space, candidate diagnostic strategies are generated; Calculate the strategy effectiveness score of the candidate diagnostic strategy, and determine whether the strategy effectiveness score meets the strategy effectiveness evaluation criteria; When the strategy effectiveness score meets the strategy effectiveness evaluation criteria, the candidate diagnostic strategy is included in the initial diagnostic strategy set.

6. The Internet-based remote diagnostic system for automated testing equipment as described in claim 5, characterized in that, The step of performing context-based matching on the diagnostic knowledge graph based on the device anomaly feature graph and the device environmental parameter data stream to obtain an adapted diagnostic knowledge subgraph includes: Traverse the diagnostic knowledge graph and read the historical diagnostic case records stored therein. The historical diagnostic case records include historical equipment abnormality feature information, historical environmental parameter data, and historical diagnostic strategies that have been executed. Calculate the feature similarity between the equipment anomaly feature map and the historical equipment anomaly feature information; Calculate the environmental parameter similarity between the device environmental parameter data stream and the historical environmental parameter data; The feature similarity and the environmental parameter similarity are weighted and fused according to predefined fusion rules to generate case correlation. Determine whether the case correlation degree reaches or exceeds the case correlation threshold; When the case correlation reaches or exceeds the case correlation threshold, the knowledge nodes associated with the corresponding historical diagnostic case records are extracted and used to form the adaptive diagnostic knowledge subgraph.

7. The Internet-based remote diagnostic system for automated testing equipment as described in claim 1, characterized in that, The strategy optimization module is used to dynamically adjust and correct the initial diagnostic strategy set by combining the device status prediction features and the future environmental parameter prediction values, to generate an optimized diagnostic strategy set, including: Based on the predicted features of the equipment status and the predicted values ​​of the future environmental parameters, a prospective association matching is performed on the diagnostic knowledge graph to obtain a predictive diagnostic knowledge subgraph. In the predictive diagnostic knowledge subgraph, identify the interval of predictive policy triggering conditions and determine the predictive policy generation space; Based on the strategy effectiveness evaluation criteria and the predictive strategy generation space, an iterative strategy search is performed to generate a predictive diagnostic strategy set; The predictive diagnostic strategy set is fused and conflict-resolved with the initial diagnostic strategy set to generate the optimized diagnostic strategy set.

8. The Internet-based remote diagnostic system for automated testing equipment as described in claim 1, characterized in that, The state prediction module is used to perform time-series evolution analysis on the equipment anomaly feature map and the equipment environmental parameter data stream to obtain equipment state prediction features and future environmental parameter prediction values, including: The abnormal feature map of the equipment is decomposed into time series to extract the feature evolution trend components; Periodically analyze the data stream of the equipment's environmental parameters to extract the patterns of change in these parameters; Using a trained time-series prediction model, joint reasoning is performed on the feature evolution trend components and the environmental parameter change patterns to predict the equipment state prediction features and the future environmental parameter prediction values.

9. The Internet-based remote diagnostic system for automated testing equipment as described in claim 1, characterized in that, The data acquisition module is used to acquire real-time data streams of equipment operating status and equipment environmental parameters from multiple distributed automated testing devices via the Internet, including: Establish secure communication connections with various automated testing devices through the device gateway protocol; The device's sensor readings, operation logs, and alarm information are read according to the preset sampling frequency to form raw operation data; The system collects temperature, humidity, and vibration intensity data at the location of the device through an environmental sensor node network to form raw environmental data. The original operating data and the original environmental data are subjected to data verification, filtering, and format unification processing to generate standardized data streams of the device operating status and the device environmental parameters.

10. The Internet-based remote diagnostic system for automated testing equipment as described in claim 1, characterized in that, The remote execution module is used to send diagnostic control commands to the corresponding automated testing equipment via the Internet according to the optimized diagnostic strategy set, including: The optimized diagnostic strategy set is parsed into a sequence of operation instructions that the device can recognize; The sequence of operation instructions is sent to the target automated detection equipment via a secure communication link; Receive instruction execution status feedback information from the target automated inspection equipment; The instruction execution status feedback information is sent back to the data acquisition module to update the device operating status data stream.

Images