Highway electromechanical autonomous decision-making method and system based on big data and knowledge graph

Abstract

The application provides a highway electromechanical autonomous decision method and system based on big data and a knowledge graph, relates to the technical field of intelligent highway traffic, and first acquires a running data set of a highway electromechanical device in a continuous running cycle, which contains device state information, environmental perception information and operation instruction records, then carries out feature extraction on the running data set to obtain a key feature set reflecting the running state of the device, then calls a pre-constructed knowledge graph to carry out correlation analysis on the key feature set to generate decision correlation information, determines the current running state evaluation result and the potential risk prediction result of the highway electromechanical system based on the decision correlation information, and finally generates an autonomous decision instruction for triggering device adjustment or resource scheduling operation, so that intelligent and automated decision of the highway electromechanical system is realized, and the system running efficiency and safety are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent highway transportation technology, and more specifically, to a highway electromechanical autonomous decision-making method and system based on big data and knowledge graphs. Background Technology

[0002] In the highway transportation sector, highway electromechanical systems play a crucial role in ensuring road safety and efficient operation. These systems encompass numerous electromechanical devices, such as lighting equipment, monitoring equipment, and communication equipment. With the increasing volume of highway traffic and the growing complexity of electromechanical equipment, higher demands are being placed on the operation and management of highway electromechanical systems.

[0003] Currently, the operation and management of highway electromechanical systems mainly rely on manual experience or simple automated monitoring systems. Manual experience-based decision-making suffers from strong subjectivity and slow response times, making it difficult to adapt to the complex and ever-changing highway operating environment. While simple automated monitoring systems can collect some equipment data, they lack the ability to deeply analyze and comprehensively utilize this data, failing to accurately grasp the changing trends of equipment operating status and hindering the early detection of potential risks. Furthermore, existing technologies do not fully integrate knowledge from the highway electromechanical field, failing to effectively link equipment operating data with professional expertise, resulting in insufficient scientific rigor and accuracy in decision-making, thus impacting the operational efficiency and safety of highway electromechanical systems. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a highway electromechanical autonomous decision-making method based on big data and knowledge graphs, the method comprising:

[0005] Acquire a set of operational data generated by highway electromechanical equipment during a continuous operating cycle, the set of operational data including equipment status information, environmental perception information and operation instruction records;

[0006] The operational data set is subjected to feature extraction processing to obtain a key feature set reflecting the equipment's operating status. The key feature set includes equipment performance correlation features, environmental impact coupling features, and operational response lag features.

[0007] The pre-built knowledge graph is invoked to perform correlation analysis on the key feature set, generating decision-related information containing the device state evolution path and the relationship between influencing factors;

[0008] Based on the aforementioned decision-related information, the current operating status assessment results and potential risk prediction results of the highway electromechanical system are determined.

[0009] Based on the operational status assessment results and potential risk prediction results, autonomous decision-making instructions for the highway electromechanical system are generated. These autonomous decision-making instructions are used to trigger equipment adjustment operations or resource scheduling operations.

[0010] In another aspect, embodiments of the present invention also provide a highway electromechanical autonomous decision-making system based on big data and knowledge graphs, including a processor and a machine-readable storage medium connected to the processor. The machine-readable storage medium is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the machine-readable storage medium to implement the above-described method.

[0011] Based on the above, this invention, by comprehensively acquiring multi-dimensional operational data of highway electromechanical equipment during continuous operation cycles and performing deep feature extraction, can accurately capture key features of equipment operating status. It utilizes a pre-constructed knowledge graph to perform correlation analysis on these key features, effectively integrating professional knowledge in the field of highway electromechanical systems. This deeply integrates equipment operating data with professional knowledge, generating decision-related information that includes equipment state evolution paths and the relationships between influencing factors. Based on this decision-related information, it determines the current operating status assessment results and potential risk prediction results, and generates autonomous decision-making instructions. This achieves intelligent and automated decision-making for highway electromechanical systems, enabling timely and accurate triggering of equipment adjustment operations or resource scheduling operations. This significantly improves the operating efficiency and safety of highway electromechanical systems while reducing maintenance costs and risks. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the execution flow of the highway electromechanical autonomous decision-making method based on big data and knowledge graph provided in the embodiments of the present invention.

[0013] Figure 2 This is a schematic diagram of exemplary hardware and software components of the highway electromechanical autonomous decision-making system based on big data and knowledge graphs provided in an embodiment of the present invention. Detailed Implementation

[0014] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a highway electromechanical autonomous decision-making method based on big data and knowledge graphs, provided in one embodiment of the present invention. The following is a detailed description of this highway electromechanical autonomous decision-making method based on big data and knowledge graphs.

[0015] Step S110: Obtain the set of operating data generated by the highway electromechanical equipment during a continuous operating cycle. The set of operating data includes equipment status information, environmental perception information, and operation instruction records.

[0016] In this embodiment, highway electromechanical equipment generates various operational data during continuous operation, reflecting its operational status, surrounding environmental conditions, and operator behavior. Highway electromechanical equipment encompasses multiple types, such as lighting equipment, surveillance cameras, and traffic lights. Equipment status information is a series of data describing the equipment's own operating status. For example, for lighting equipment, this might include lamp brightness, power consumption, and operating time; for surveillance cameras, it might involve image clarity, frame rate, and signal strength; and for traffic lights, it might include the frequency of color changes and flashing patterns. Environmental perception information is data related to the environment in which the equipment is located, mainly including ambient temperature, humidity, light intensity, and air quality. Different environmental conditions may affect equipment operation to varying degrees; for example, high temperatures may cause overheating, affecting performance; high humidity may cause short circuits and other malfunctions. Operation command records record various commands issued by the operator to the equipment and the execution status of these commands, including the type of command (e.g., turn on, turn off, adjust parameters), the time the command was issued, and the corresponding equipment response time.

[0017] Step S120: Perform feature extraction processing on the running data set to obtain a key feature set reflecting the equipment operating status. The key feature set includes equipment performance correlation features, environmental influence coupling features, and operation response lag features.

[0018] Step S121: Perform feature filtering processing on the running data set, filter out redundant data items that are not directly related to the equipment operating status, and retain the functional module working status parameters in the equipment status information, the temperature and humidity fluctuation records in the environmental perception information, and the instruction type sequence in the operation instruction records.

[0019] In this step, the runtime dataset may contain a large amount of redundant data that is not directly related to the device's operating status. Without filtering, this could increase the complexity and computational load of subsequent processing and potentially interfere with the feature extraction results. Therefore, feature filtering is necessary for the runtime dataset.

[0020] During the screening process, the relevance of data to the equipment's operating status can be used to determine whether to retain a data item. For equipment status information, the focus is on retaining functional module operating status parameters, as these parameters directly reflect the equipment's performance. For example, for lighting fixtures, parameters such as brightness and power can be retained because these parameters are closely related to the normal operation of the fixtures; while information unrelated to the equipment's operating status, such as the equipment's production batch and installation location, will be filtered out. For environmental sensing information, temperature and humidity fluctuation records are primarily retained because temperature and humidity are environmental factors that significantly affect equipment operation. Other environmental information, such as the content of certain components in air quality, will also be filtered out if it is not strongly correlated with the equipment's operating status. For operation command records, the command type sequence is retained, which helps in analyzing operators' operating habits and the equipment's response to different commands. Data such as notes in the commands that are not directly related to the equipment's operating status will be removed.

[0021] Specific filtering methods can include rule matching, where pre-defined filtering rules are used to evaluate data items. Alternatively, machine learning algorithms can be used to train a model that identifies data items related to the device's operating status. For example, a decision tree algorithm can be used, taking data items as input and the device's operating status as output, to train a model to determine whether each data item should be retained.

[0022] Step S122: Perform time-series correlation analysis on the filtered equipment status information, and extract the changing trend characteristics of the functional module working status parameters within the continuous operation cycle as equipment performance correlation characteristics. The equipment performance correlation characteristics include the stability distribution pattern of the status parameters and the sudden change triggering conditions.

[0023] This step primarily involves in-depth analysis of the filtered equipment status information to extract key features reflecting equipment performance. The functional module operating status parameters in the equipment status information change over time. By performing time-series correlation analysis on these parameters, we can discover the trends and patterns of change, thereby obtaining the correlation characteristics of equipment performance.

[0024] Step S1221: Arrange the functional module working status parameters in the device status information according to the timestamp order to generate a time series data chain.

[0025] Before performing time-series correlation analysis, the functional module operating status parameters in the filtered device status information need to be arranged in chronological order according to their timestamps. A timestamp is an identifier that records the time the data was generated; by sorting by timestamps, the functional module operating status parameters can be organized into an ordered time-series data chain.

[0026] For each functional module of a device, its operating status parameters are generated at different points in time. For example, the brightness and power of a lighting fixture change continuously over time, with each measurement corresponding to a specific time. Therefore, arranging these brightness and power values ​​in ascending order of timestamps creates a time-series data chain regarding the brightness and power of the lighting fixture. This time-series data chain clearly demonstrates how the device's status parameters change over time.

[0027] The sorting process can be implemented using sorting algorithms such as quicksort and mergesort. Each data item in the device status information is compared and sorted according to its timestamp, ultimately generating an ordered time-series data chain.

[0028] Step S1222: Perform sliding window segmentation on the time series data chain to obtain multiple sub-data windows containing a preset time span.

[0029] To analyze the changing trends of equipment status parameters in more detail, it is necessary to perform sliding window segmentation on the generated time series data chain. The sliding window is a commonly used time series analysis method. By setting a window of fixed length, it slides across the time series data chain, and each slide captures a segment of data to form a sub-data window.

[0030] The preset time span is determined based on specific analysis needs and the operating characteristics of the equipment. For example, a shorter time span can be set for equipment parameters that change rapidly, while a longer time span can be set for parameters that change slowly. Each sub-data window contains equipment status parameter data within the set time range. By analyzing these sub-data windows, the changing patterns of equipment status parameters over different time periods can be discovered.

[0031] The sliding step size of the sliding window also needs to be set according to the actual situation. If the sliding step size is small, more sub-data windows can be obtained, but the amount of computation will also increase accordingly; if the sliding step size is large, the number of sub-data windows will be reduced, but some important change information may be missed. For example, for a time series data link of lamp brightness of lighting equipment, setting an appropriate time span and sliding step size can divide it into multiple sub-data windows, each containing lamp brightness data over a period of time.

[0032] Step S1223: Perform trend fitting processing on the working status parameters in each sub-data window to generate trend type labels that reflect whether the parameters are rising, falling, or stable.

[0033] After obtaining multiple sub-data windows, trend fitting processing needs to be performed on the working status parameters within each sub-data window. The purpose of trend fitting is to find the changing trend of the parameters within each sub-data window and represent it using a trend type identifier.

[0034] Various methods can be used for trend fitting, such as linear fitting and polynomial fitting. Linear fitting is a simple and commonly used method that assumes the parameter changes linearly, approximating the parameter change by fitting a straight line. Polynomial fitting, on the other hand, can handle more complex trends by fitting a polynomial function to describe the parameter change.

[0035] After trend fitting, the parameter's changing trend is determined based on the fitting results. If the slope of the fitted curve is positive, the parameter is trending upward; if the slope is negative, the parameter is trending downward; if the slope is close to zero, the parameter is in a stable state. Based on these results, a corresponding trend type label, such as "rising," "falling," or "stable," is generated for each sub-data window. For example, for the luminaire brightness sub-data window of lighting equipment, the slope of its fitted curve is obtained through linear fitting. The changing trend of luminaire brightness within this sub-data window is determined based on the slope, and a corresponding trend type label is generated.

[0036] Step S1224: Statistically analyze the frequency of occurrence of different trend type identifiers in continuous sub-data windows, and extract the persistence characteristics of trend types as the stability distribution pattern of state parameters.

[0037] After generating trend type identifiers for each sub-data window, it is necessary to count the frequency of occurrence of different trend type identifiers in consecutive sub-data windows. By counting the frequency of occurrence, the stability of the device status parameters can be understood.

[0038] The specific method involves iterating through consecutive sub-data windows and recording the frequency of each trend type identifier. Then, the frequency of occurrence of each trend type identifier is calculated, which is the number of times that trend type identifier appears divided by the total number of sub-data windows. These frequencies constitute the persistence characteristics of the trend type.

[0039] The stability distribution pattern of state parameters can be reflected through these persistence characteristics. If a certain trend type indicator appears frequently, it indicates that the trend type is relatively persistent, and the equipment state parameters are relatively stable under that trend. If the frequency of various trend type indicators is relatively even, it indicates that the equipment state parameters change frequently and have poor stability. For example, for the brightness of lighting equipment, the frequency of occurrence of the three trend type indicators "increasing," "decreasing," and "stable" in a continuous sub-data window can be statistically analyzed. Based on these frequencies, the stability distribution pattern of the lighting brightness can be determined.

[0040] Step S1225: Detect the change range of working status parameters between adjacent sub-data windows, record the trigger time point when the change range exceeds the preset threshold and the corresponding device operation scenario, and generate a description of the trigger condition for the trend change.

[0041] When analyzing the changing trends of equipment status parameters, in addition to focusing on the stability distribution pattern, it is also necessary to detect the magnitude of changes in operating status parameters between adjacent sub-data windows. When the magnitude of change exceeds a preset threshold, it indicates that a sudden change has occurred in the equipment status parameters, and relevant information needs to be recorded.

[0042] The preset threshold is determined based on the characteristics of the equipment and actual operating conditions, and it represents the maximum normal range of change in the equipment's status parameters. When the change in the operating status parameters between adjacent sub-data windows exceeds this preset threshold, the trigger time point, i.e., the time when the sudden change occurs, is recorded. Simultaneously, the corresponding equipment operating scenario is recorded, including ambient temperature, humidity, and whether the operator has issued instructions.

[0043] The information recorded above constitutes a description of the triggering conditions for trend changes. By analyzing these descriptions, the causes and patterns of sudden changes in equipment status parameters can be identified, providing a basis for equipment fault warnings and maintenance. For example, regarding the brightness of lighting equipment, when the brightness change between adjacent sub-data windows exceeds a preset threshold, the time, ambient temperature, humidity, and whether the operator has issued a brightness adjustment command are recorded to form a description of the triggering conditions for trend changes.

[0044] Step S1226: Combine the stability distribution pattern and triggering condition description to form the device performance correlation feature.

[0045] Finally, the stability distribution pattern of state parameters and the triggering conditions for trend changes are combined to form the equipment performance correlation characteristics. These characteristics can comprehensively reflect the performance status of the equipment.

[0046] The stability distribution pattern illustrates the stability of the device's state parameters under normal conditions. The description of the triggering conditions for trend changes provides information related to sudden changes in the device's state parameters.

[0047] Combining these two pieces of information forms a complete equipment performance correlation characteristic. This characteristic can serve as an important basis for subsequent analysis and decision-making, such as assessing the health status of equipment and predicting the probability of equipment failure. For lighting fixtures, combining the description of their brightness stability distribution pattern and the triggering conditions for trend changes yields an equipment performance correlation characteristic that can be used to determine whether the fixtures are working properly and whether maintenance is required.

[0048] Step S123: Perform multi-source data coupling analysis on the screened environmental perception information, and extract the synchronous change features of temperature and humidity fluctuation records and equipment status information as environmental impact coupling features. The environmental impact coupling features include a description of the correlation strength between temperature and humidity fluctuation amplitude and equipment status parameters.

[0049] This step primarily analyzes the filtered environmental sensing information, focusing on the synchronous change characteristics between temperature and humidity fluctuation records and equipment status information. Environmental factors, especially temperature and humidity, can significantly impact equipment operation; multi-source data coupling analysis can provide a deeper understanding of the extent and manner of these impacts.

[0050] Step S1231: Align the temperature and humidity fluctuation records in the environmental sensing information with the functional module working status parameters in the equipment status information according to the timestamp to generate time-synchronized multi-source data pairs.

[0051] Before conducting multi-source data coupling analysis, it is necessary to align the temperature and humidity fluctuation records from the environmental sensing information with the functional module operating status parameters from the equipment status information. Since these two types of data may be collected at different times, it is necessary to ensure that they are synchronized in time in order to accurately analyze the relationship between them.

[0052] The specific approach involves matching temperature and humidity fluctuation records with functional module operating status parameters based on the timestamp information in the data. For each time point, the corresponding temperature and humidity values ​​and equipment status parameter values ​​are found, forming a multi-source data pair. If only temperature and humidity records or only equipment status parameter records are available at a certain time point, interpolation processing or discarding that data point may be necessary to ensure the integrity of the data pair.

[0053] Data alignment processes generate time-synchronized multi-source data pairs. These data pairs can serve as the basis for subsequent correlation calculations and analyses. For example, for lighting equipment, the temperature and humidity fluctuation records of its environment are aligned with the operating status parameters of functional modules such as the brightness and power of the lamps by timestamps, generating a series of time-synchronized multi-source data pairs.

[0054] Step S1232: Perform correlation calculation on each pair of multi-source data, and statistically analyze the consistency ratio between the temperature and humidity fluctuation amplitude and the change direction of the functional module's working status parameters to obtain the correlation calculation results.

[0055] After obtaining time-synchronized multi-source data pairs, correlation calculations need to be performed on each pair. The purpose of the correlation calculation is to assess the degree of correlation between the temperature and humidity fluctuation amplitude and the operating status parameters of the functional modules.

[0056] For example, various methods can be used to calculate correlation, such as the Pearson correlation coefficient method and the Spearman correlation coefficient method. These methods represent the degree of association between two variables by calculating the correlation coefficient between them. In this step, the focus is on statistically analyzing the proportion of consistency between the temperature and humidity fluctuation amplitude and the direction of change of the functional module's operating status parameters.

[0057] For example, for each pair of multi-source data, the direction of change in temperature and humidity fluctuation amplitude is compared with the direction of change in the functional module's operating status parameters. If the two directions of change are the same, they are considered consistent; if the directions of change are opposite, they are considered inconsistent. The number of pairs of multi-source data with consistent directions of change is counted and divided by the total number of pairs to obtain the consistency ratio of change direction. This consistency ratio of change direction is the correlation calculation result, which reflects the degree of correlation between temperature and humidity fluctuation amplitude and the functional module's operating status parameters. For example, for lighting equipment, by calculating the consistency ratio between the temperature and humidity fluctuation amplitude of its environment and the direction of change in the brightness and power of the lamp, the obtained correlation calculation result can be used to determine the degree of influence of temperature and humidity on the lamp's performance.

[0058] Step S1233: Extract the probability of abnormal occurrence of the working status parameters of the functional module when the temperature and humidity fluctuation range reaches the set range, and generate a mapping relationship table between the fluctuation range and the abnormal probability.

[0059] To gain a deeper understanding of the impact of temperature and humidity fluctuations on equipment status, it is necessary to extract the probability of abnormal occurrence of functional module operating status parameters when the temperature and humidity fluctuation range reaches a set range.

[0060] First, the temperature and humidity fluctuations are divided into different intervals. For each interval, the number of times the functional module's operating status parameters become abnormal within that interval is counted. The definition of abnormality can be determined based on the normal operating range of the equipment. For example, if the brightness of the lighting fixtures exceeds the normal range or the power consumption of the lighting equipment increases abnormally, it can be considered an abnormal situation.

[0061] Then, the probability of an anomaly occurring within each interval is calculated, which is the number of times an anomaly occurs within that interval divided by the total amount of data within that interval. Using this method, the correspondence between different intervals of temperature and humidity fluctuation and the probability of anomalies occurring in the functional module's operating status parameters is obtained.

[0062] These correspondences are organized into a mapping table, which clearly shows the relationship between the amplitude of temperature and humidity fluctuations and the probability of anomalies. For example, for lighting equipment, what is the probability of abnormal lamp brightness when the ambient temperature fluctuates within a certain range? What is the probability of abnormal lamp power when the humidity fluctuates within another range? This mapping table can be used to predict the likelihood of equipment malfunctions under different temperature and humidity conditions.

[0063] Step S1234: Analyze the degree of matching between the periodic characteristics of temperature and humidity fluctuations and the periodic changes of equipment status parameters, and extract the synchronization index between the fluctuation period and the status period.

[0064] In addition to focusing on the correlation between temperature and humidity fluctuations and equipment status parameters, it is also necessary to analyze the degree of matching between the periodic characteristics of temperature and humidity fluctuations and the periodic changes of equipment status parameters.

[0065] For temperature and humidity fluctuation records and equipment status parameter data, periodic analysis should be performed separately. Methods such as spectral analysis can be used to determine their periods. Spectral analysis determines the period by converting time-series data to the frequency domain and identifying the main frequency components.

[0066] After determining the period of temperature and humidity fluctuations and the period of equipment status parameters, compare their period lengths and phase relationships. If their period lengths are similar and their phase relationships are stable, it indicates that they have high synchronization; if the period lengths differ significantly or the phase relationships are unstable, it indicates low synchronization.

[0067] Based on the comparison results, a synchronicity index is extracted to quantify the degree of matching between the temperature and humidity fluctuation cycle and the equipment status parameter cycle. This synchronicity index can be used to assess the impact of environmental factors on equipment operation. For example, for lighting equipment, by analyzing the degree of matching between the temperature and humidity fluctuation cycle of its environment and the periodic changes in the brightness and power of the lamps, the extracted synchronicity index can be used to determine whether the periodic changes in temperature and humidity will interfere with the normal operation of the lamps.

[0068] Step S1235: Combine the correlation calculation results, mapping relationship table and synchronization index to form environmental impact coupling characteristics.

[0069] Finally, the correlation calculation results between temperature and humidity fluctuation amplitude and functional module working status parameters, the mapping relationship table between temperature and humidity fluctuation amplitude and abnormal probability, and the synchronization index between temperature and humidity fluctuation cycle and equipment status parameter cycle are combined to form the environmental impact coupling characteristics.

[0070] This environmental impact coupling characteristic comprehensively describes the synchronous changes between temperature and humidity fluctuations and equipment status information, as well as the strength of their correlation. The correlation calculation results reflect the direct correlation between the amplitude of temperature and humidity fluctuations and equipment status parameters; the mapping relationship table provides the probability information of equipment anomalies under different temperature and humidity fluctuation ranges; and the synchronicity index illustrates the matching between the periodicity of temperature and humidity fluctuations and the periodic changes of equipment status parameters.

[0071] Combining these three pieces of information forms a complete environmental impact coupling characteristic. This characteristic can be used to assess the degree of impact of environmental factors on equipment operation and predict performance changes of equipment under different environmental conditions. For lighting equipment, the environmental impact coupling characteristic can be used to determine the impact of ambient temperature and humidity on the normal operation and lifespan of the luminaires, and to formulate corresponding maintenance and management strategies accordingly.

[0072] Step S124: Perform response delay analysis on the filtered operation instruction records, and extract the time difference feature between the instruction type sequence and the change of device status parameters as the operation response lag feature. The operation response lag feature includes the average response delay duration and delay fluctuation range corresponding to different instruction types.

[0073] This step primarily involves analyzing the filtered operation command records to extract operation response lag characteristics. Operation response lag characteristics reflect the speed and delay of the equipment's response to commands issued by the operator.

[0074] Step S1241: Traverse the sequence of instruction types in the operation instruction record and mark the sending timestamp of each operation instruction.

[0075] When performing response delay analysis, the first step is to traverse the sequence of instruction types in the operation instruction log. The operation instruction log contains various instructions issued by the operator to the equipment, and each instruction corresponds to a specific time.

[0076] By traversing the sequence of instruction types, each instruction is marked with its sent timestamp. A timestamp is a precise identifier that records the exact moment an instruction was issued; it can be a specific date and time. For example, for lighting equipment, if an operator issues an instruction to turn on the lights at a certain moment, that moment is the timestamp of that instruction. The purpose of marking timestamps is to accurately calculate the response delay of the instruction later.

[0077] Step S1242: Locate the start timestamp of the status parameter change corresponding to each operation command in the device status information, and calculate the difference between the sending timestamp and the change start timestamp as the response delay duration of a single operation command.

[0078] After marking the sending timestamp of each operation command, it is necessary to find the start timestamp of the status parameter change corresponding to that command in the device status information. The device status information records the operating status parameters of the device at different times; when the operator issues an operation command, the device's status parameters will change accordingly.

[0079] For example, when an operator issues a command to adjust the brightness of a lighting fixture, the brightness of the fixture will begin to change from its current value. The time point at which the brightness began to change is located in the equipment status information; this time point is the start timestamp of the status parameter change corresponding to that operation command.

[0080] The response delay for a single operation command is calculated by comparing the timestamp of the command's transmission with the timestamp of the corresponding change in the status parameter. This difference reflects the time interval between receiving the command and initiating a response. The response delay may vary depending on the operation command's performance and current operating status.

[0081] Step S1243: Classify and statistically analyze the response delay duration according to the instruction type, and calculate the average response delay duration and the standard deviation of the delay duration for the same type of operation instruction as the delay fluctuation range.

[0082] After obtaining the response delay duration for each operation command, these delay durations need to be categorized and statistically analyzed according to the command type. The device's response may differ depending on the type of operation command.

[0083] For each instruction type, the response latency of all operation instructions under that type is summarized. Then, the average of these response latency times is calculated to obtain the average response latency time for that instruction type. The average response latency time reflects the device's average response speed for that instruction type.

[0084] Simultaneously, calculate the standard deviation of the response latency for this command type. Standard deviation is a statistic that measures the dispersion of data, representing the fluctuation of the response latency relative to the mean. The standard deviation of the latency, as the range of latency fluctuation, reflects the stability of the device's response to this command type. A smaller standard deviation indicates that the device's response to this command type is relatively stable; a larger standard deviation indicates that the device's response exhibits greater fluctuations.

[0085] For example, for lighting equipment, the average response delay and delay fluctuation range of turn-on commands, turn-off commands, and brightness adjustment commands can be calculated separately. By comparing these values, the response performance of the equipment to different commands can be understood.

[0086] Step S1244: Analyze the relationship between the time interval and response delay between consecutively sent similar operation commands, and extract the characteristics of the influence of command sending frequency on response delay.

[0087] In addition to focusing on the average response speed and stability of the device to different command types, it is also necessary to analyze the impact of command sending frequency on response latency.

[0088] Observe the time interval between consecutively sent commands of the same type, that is, the difference in the timestamps of two adjacent commands of the same type. At the same time, record the response delay of each command.

[0089] Analyze the relationship between time intervals and response latency. For example, when sending commands to turn on lighting equipment rapidly and continuously, observe whether the response latency increases as the time interval decreases. If so, it indicates that the command sending frequency affects the response latency, and the device's response speed may slow down under high-frequency commands.

[0090] Based on the analysis results, the impact characteristics of command transmission frequency on response latency were extracted. These characteristics can be used to predict the device's response under different command transmission frequencies. For example, in practical operation, the command transmission time interval can be reasonably arranged according to the device's response characteristics to command transmission frequency to improve the device's response efficiency.

[0091] Step S1245: Combine the average response delay duration, delay fluctuation range, and influencing characteristics to form the operational response hysteresis characteristics.

[0092] Finally, the average response delay duration, delay fluctuation range, and the impact of command sending frequency on response delay of similar operation commands are combined to form the operation response lag characteristics.

[0093] Average response latency reflects the average response speed of the device to different command types; latency fluctuation range indicates the stability of the device's response to the same command type; and the influence of command sending frequency on response latency provides information on the device's response under different command sending frequencies.

[0094] Combining these three pieces of information forms a complete operational response hysteresis characteristic. This characteristic can serve as an important basis for evaluating equipment operational performance, such as determining whether maintenance or operational strategy optimization is needed. For lighting equipment, the operational response hysteresis characteristic helps operators understand how the equipment responds to different commands, thereby enabling more rational control of equipment operation.

[0095] Step S125: Standardize and splice the equipment performance correlation characteristics, environmental impact coupling characteristics, and operational response lag characteristics to obtain the key feature set.

[0096] After obtaining the equipment performance correlation characteristics, environmental impact coupling characteristics, and operational response lag characteristics, they need to be standardized and spliced ​​to obtain the final set of key features.

[0097] Since these three features may have different dimensions and ranges of values, standardization is necessary to enable effective comparison and combination in subsequent analyses. Standardization can unify the ranges of values ​​for different features into a relatively consistent interval; common standardization methods include normalization.

[0098] Normalization maps the value of each feature to a specific interval, such as [0, 1]. By normalizing, the influence of different dimensions between features is eliminated, making them numerically comparable.

[0099] After standardization, the equipment performance correlation characteristics, environmental impact coupling characteristics, and operational response lag characteristics are spliced ​​together. Splicing involves combining these three characteristics in a predetermined order to form a new feature set.

[0100] The assembled key feature set contains information on various aspects such as equipment performance, environmental impact, and operational response, providing a more comprehensive reflection of the operational status of highway electromechanical equipment. For highway lighting equipment, the key feature set can be used to assess the health status of the lamps, predict the probability of failure, and develop reasonable maintenance plans.

[0101] Step S130: Call the pre-built knowledge graph to perform association analysis on the key feature set to generate decision association information containing the device state evolution path and the relationship between influencing factors.

[0102] In this step, a pre-built knowledge graph is used to perform association analysis on the obtained key feature set to generate decision-related information. A knowledge graph is a structured knowledge representation method that displays the relationships and knowledge between entities in the form of a graph.

[0103] Step S131: Extract the stability distribution pattern from the device performance correlation features from the key feature set, match the corresponding device state node in the knowledge graph, and establish the correlation between the stability distribution pattern and the device state node.

[0104] Step S1311: Analyze the stability distribution pattern in the device performance correlation characteristics, and extract the proportion of stable state and the proportion of abrupt change in the trend type persistence characteristics.

[0105] After extracting the stability distribution pattern of equipment performance correlation features from the key feature set, it needs to be analyzed. The stability distribution pattern contains the trend type persistence features of equipment state parameters, and through analysis, the proportion of stable states and the proportion of abrupt changes can be extracted.

[0106] The trend type persistence characteristic describes the persistence of equipment state parameters under different trend types. The stable state percentage refers to the proportion of the total observation time during which the equipment state parameters are in a stable trend (such as a steady trend); the abrupt state percentage refers to the proportion of the total observation time during which the equipment state parameters undergo an abrupt change (such as suddenly changing from a steady trend to an upward or downward trend).

[0107] In the specific analysis process, based on the previous analysis of the changing trends of equipment state parameters, the duration of stable and abrupt trends is statistically analyzed. Then, the proportion of stable states and the proportion of abrupt states are calculated separately. For example, for the brightness of lighting equipment, the duration of its stable state and the duration of abrupt changes over a period of time are statistically analyzed, and then the proportion of stable states and the proportion of abrupt states are calculated. These proportions can reflect the stability and abrupt changes of the equipment state, providing an important basis for subsequently finding matching equipment state nodes in the knowledge graph.

[0108] Step S1312: In the device state node library of the knowledge graph, search for historical device state nodes that contain the same percentage of stable states and percentage of abrupt changes.

[0109] The knowledge graph's device state node library stores a large amount of historical device state information. After extracting the proportion of stable states and the proportion of abrupt changes in the device performance correlation features, the library is used to search for historical device state nodes that match these characteristics.

[0110] Each device state node in the knowledge graph corresponds to a specific device state and contains various related information for that state. By traversing the device state node library, the percentage of stable states and percentage of abrupt changes recorded by each node are compared with the currently extracted values.

[0111] Since the number of device state nodes in a knowledge graph can be large, indexing techniques can be used to improve search efficiency. An index is pre-built for the device state node library, dividing the nodes into different index intervals based on the range of stable and mutated state proportions. During a search, the index interval that may contain matching nodes is quickly located based on the currently extracted stable and mutated state proportions, and then a precise search is performed within that interval.

[0112] During the search process, it is necessary to ensure that the calculation methods and accuracy of the proportion of stable states and the proportion of abrupt changes are consistent between the current device state and the historical device state nodes in the knowledge graph, in order to guarantee the accuracy of the search results. For example, for the brightness parameters of highway lighting equipment, the time range and statistical methods used in calculating the proportion of stable states and the proportion of abrupt changes in the current device state need to be the same as the calculation methods used for historical device state nodes in the knowledge graph.

[0113] Step S1312: In the device state node library of the knowledge graph, search for historical device state nodes that contain the same percentage of stable states and percentage of abrupt changes.

[0114] The search process can be implemented using search algorithms. Breadth-first search or depth-first search algorithms can be used, starting from the root node of the knowledge graph and traversing the nodes layer by layer, checking whether the percentage of stable states and percentage of mutated states recorded in each node are the same as the currently extracted values.

[0115] During the traversal, if a matching node is found, it is recorded. If multiple matching nodes exist, further filtering based on other relevant information is required. For example, factors such as device type and environment can be considered, prioritizing historical nodes more similar to the current device.

[0116] To improve search accuracy, a certain error range can be set for the proportions of stable and abrupt states. Since actual data may contain measurement errors or fluctuations, allowing a certain range of differences can increase the likelihood of finding matching nodes. For example, if the current device's stable state proportion is a certain value, during the search, nodes in the knowledge graph whose stable state proportions are within a certain error range of that value can be considered a match.

[0117] Step S1313: Extract the device operation scenario description and subsequent state change records corresponding to the historical device status nodes, and generate a matching score between the current stability distribution pattern and the historical device status nodes.

[0118] After identifying historical device state nodes containing the same percentages of stable and abrupt states, the corresponding device operating scenario description and subsequent state change records are extracted. The device operating scenario description includes the environmental conditions of the device (e.g., temperature, humidity, light intensity), operating instructions (e.g., instruction type, instruction sending frequency), and other relevant factors. The subsequent state change records show the evolution of the historical device state over a subsequent period, such as whether the device malfunctioned or its performance degraded.

[0119] Based on the extracted description of the device's operating scenario and subsequent state change records, a matching score is generated between the current stability distribution pattern and historical device state nodes. The matching score comprehensively considers multiple factors to measure the degree of similarity between the current device state and historical nodes.

[0120] To describe the equipment's operating scenario, the current environmental conditions and operational commands can be compared with historical data. Differences in environmental conditions can be assessed by calculating the differences in various environmental parameters, while differences in operational commands can be considered based on command types and transmission frequencies. For each comparative factor, different weights can be assigned according to its impact on the equipment's state. For example, ambient temperature has a significant impact on equipment performance and can be assigned a higher weight; while some less important environmental factors have relatively lower weights.

[0121] For subsequent state change records, we can analyze the consistency between the development trend of historical devices under similar states and the potential development trend of the current device. If a historical device experienced a certain fault or performance change after a similar state, and the current device also has the potential to experience a similar situation, a higher score will be given in the matching score; conversely, if the development trends differ significantly, the score will be reduced accordingly.

[0122] The comparison results of each factor are weighted according to their respective weights to obtain a matching score between the current stability distribution pattern and the historical device state nodes. For example, assuming the weight of the environmental condition comparison result is A, the weight of the operation command comparison result is B, and the weight of the subsequent state change record comparison result is C, the comparison scores of these three factors are multiplied by their respective weights and then summed to obtain the matching score.

[0123] Step S1314: Select the historical device status node with the highest matching score as the mapping node for the current device status.

[0124] After generating the matching scores between the current stability distribution pattern and each historical device state node, all matching scores are compared. By iterating through all the matching scores, the historical device state node corresponding to the highest score is identified.

[0125] The historical device state node with the highest matching score is chosen as the mapping node for the current device state because this node is most similar to the current device state in terms of the proportion of stable states, the proportion of abrupt changes, the device's operating scenario, and subsequent state change trends. Using this mapping node as a reference allows for a more accurate prediction of the current device's future state and potential problems.

[0126] For example, for traffic signal equipment on highways, the mapping node selected by matching degree scoring may be a historical state node of a previous signal light under similar stable state and environmental conditions. Based on the subsequent state change records of the mapping node, it is possible to predict whether the current signal light may malfunction or experience performance degradation.

[0127] Step S1315: Establish direct association edges between the current stability distribution pattern and the mapping node, wherein the attribute values ​​of the direct association edges are matching degree score and historical state change records.

[0128] After determining the mapping node for the current device state, a direct association edge is established between the current stability distribution pattern and the mapping node. Association edges are a way to represent relationships between entities in a knowledge graph; by establishing association edges, the current device state can be connected to historical knowledge in the knowledge graph.

[0129] Set attribute values ​​for this directly related edge. The attribute values ​​include a matching score and historical state change records. The matching score indicates the similarity between the current stability distribution pattern and the mapped node. The higher the matching score, the more valuable the information from the reference mapped node is for analyzing the current device state.

[0130] Historical state change records detail the evolution of the mapped node's device state over time. This information is transmitted to the current device state through the attribute values ​​of the associated edges, allowing for the prediction of the device's future state based on historical experience. For example, for highway surveillance cameras, by establishing an association between the current stability distribution pattern and the mapped node, and recording the matching score and historical state change records, when analyzing the current camera's operational status, past faults and performance trends of the mapped node can be referenced, enabling proactive maintenance measures.

[0131] Step S132: Input the mapping relationship table between fluctuation amplitude and anomaly probability in the environmental impact coupling characteristics into the relation reasoning module of the knowledge graph, and analyze the influence weight between temperature and humidity fluctuation nodes and equipment status nodes.

[0132] The relational reasoning module of the knowledge graph is an important component for analyzing the relationships between entities and reasoning new knowledge. The mapping table of fluctuation amplitude and anomaly probability in the environmental impact coupling characteristics is input into this module to analyze the influence weight between temperature and humidity fluctuation nodes and equipment status nodes.

[0133] The relational reasoning module first parses the input mapping table. This table details the probability of abnormal device status parameters occurring across different temperature and humidity fluctuation ranges. Based on this information, the module analyzes the potential relationships between temperature and humidity fluctuation nodes and device status nodes.

[0134] For each temperature and humidity fluctuation range, the degree of impact of the temperature and humidity fluctuations on the equipment status is determined based on the corresponding anomaly probability in the mapping table. The higher the anomaly probability, the greater the impact of the temperature and humidity fluctuation range on the equipment status.

[0135] Then, based on the analysis results, influence weights are assigned to the edges between temperature and humidity fluctuation nodes and equipment status nodes. These influence weights reflect the strength of the effect of temperature and humidity fluctuations on equipment status. For example, if the probability of abnormal equipment status parameters is high within a certain temperature and humidity fluctuation range, a higher influence weight is assigned to the edge between the corresponding temperature and humidity fluctuation node and the equipment status node; conversely, if the probability of abnormality is low, a lower influence weight is assigned.

[0136] Using the methods described above, the relational reasoning module can quantify the impact of temperature and humidity fluctuations on equipment status, providing a basis for subsequent equipment status analysis and decision-making. For example, for communication equipment in highway electromechanical systems, by determining the weight of the impact of temperature and humidity fluctuations on its status, the likelihood of equipment failure under different temperature and humidity conditions can be predicted, thereby allowing for proactive preventative measures.

[0137] Step S133: Using the average response delay duration in the operation response hysteresis feature, find the delay association edge between the instruction type node and the device status node in the knowledge graph, and update the attribute value of the delay association edge.

[0138] The average response delay in the operational response hysteresis feature reflects the average response speed of the device to different command types. In the knowledge graph, command type nodes and device status nodes are connected by delay-related edges, and the attribute values ​​of these edges represent the device's response delay to the corresponding command.

[0139] By utilizing the average response latency, latency-related edges between instruction type nodes and device status nodes are found in the knowledge graph. The search process can be based on matching the instruction type and the identification information of the device status node. For example, based on the instruction type in the operation instruction record, the corresponding instruction type node is found in the knowledge graph, and then the latency-related edge between that node and the current device status node is searched.

[0140] After finding a delay-related edge, its attribute value is updated to the average response delay duration in the current operation response hysteresis feature. Updating the attribute value can more accurately reflect the device's current response to commands. For example, if the attribute value of a delay-related edge for a certain command type previously recorded in the knowledge graph is a single value, and the currently calculated average response delay duration has changed, then the attribute value is updated to the new average response delay duration.

[0141] By updating the attribute values ​​of the delay-related edges, the response performance of the equipment to operation commands can be tracked in real time, providing a reference for subsequent operation optimization and fault diagnosis. For example, for highway traffic signal equipment, if a significant increase in the response delay time of a certain command type is found, it may indicate a potential fault in the equipment, requiring timely inspection and maintenance.

[0142] Step S134: Based on the association, influence weight and delay association edge attribute value, construct the adjacency relationship network of the device status node. The adjacency relationship network includes the connection paths between the device status node and the temperature and humidity fluctuation node and the instruction type node.

[0143] After establishing the correlation between equipment performance characteristics and equipment status nodes in the knowledge graph, determining the influence weight between temperature and humidity fluctuation nodes and equipment status nodes, and updating the attribute values ​​of the delay association edges between instruction type nodes and equipment status nodes, an adjacency relationship network of equipment status nodes is constructed.

[0144] Step S1341: Using the mapping node of the current device state as the center node, collect all temperature and humidity fluctuation nodes and instruction type nodes connected by associated edges as adjacent nodes.

[0145] When constructing the adjacency network of device state nodes, the central node is first determined. The mapping node for the current device state is obtained by matching it with historical device state nodes in the knowledge graph through previous steps, and it represents the corresponding representation of the current device state in the knowledge graph.

[0146] Using this mapping node as the center, find all nodes connected to it via association edges. Association edges represent relationships between entities in a knowledge graph; they allow you to find other nodes related to the central node. In this embodiment, we focus on collecting nodes related to temperature and humidity fluctuations and instruction type nodes as adjacent nodes.

[0147] Temperature and humidity fluctuation nodes represent the temperature and humidity changes in the environment in which the equipment is located, while instruction type nodes represent different types of instructions issued by operators to the equipment. Collecting these adjacent nodes allows us to construct a network relating equipment status to environmental factors and operational instructions. For example, for electronic information signage on highways, we can use the mapping node of its current status as the center, collecting temperature and humidity fluctuation nodes (such as nodes corresponding to different temperature and humidity ranges) and instruction type nodes (such as instruction nodes for turning on, turning off, and updating information) connected to that node through associated edges. This lays the foundation for subsequent analysis of the impact of the environment and operations on the equipment status.

[0148] Step S1342: Set an influence weight as the weight value of each associated edge. The influence weight is derived from the correlation calculation result of the environmental influence coupling characteristics.

[0149] After collecting the adjacent nodes, an influence weight is assigned to each edge connecting the central node (the mapping node of the current device state) and its adjacent nodes. The influence weight reflects the degree of influence of the adjacent node on the central node, and it is derived from the correlation calculation results of the coupling characteristics of environmental influence.

[0150] The correlation calculation results in the environmental impact coupling feature demonstrate the degree of correlation between temperature and humidity fluctuation amplitudes and equipment status parameters. Based on these correlation calculation results, each associated edge is assigned a corresponding influence weight. For example, if the correlation calculation results show a high degree of consistency between a certain temperature and humidity fluctuation range and the direction of change of equipment status parameters, it indicates that the temperature and humidity fluctuation has a significant impact on the equipment status, and thus the associated edge connecting the corresponding temperature and humidity fluctuation node and the equipment status node is assigned a higher influence weight; conversely, if the consistency ratio is low, a lower influence weight is assigned.

[0151] By setting influence weights, the impact of adjacent nodes on equipment status can be quantified, enabling a more accurate assessment of the effects of various factors when analyzing adjacency networks. For example, for meteorological monitoring equipment on highways, influence weights can be set for the edges connecting equipment status nodes and temperature and humidity fluctuation nodes. Based on the weight magnitude, the degree of impact of different temperature and humidity conditions on the performance of the monitoring equipment can be determined, allowing for timely implementation of appropriate maintenance measures.

[0152] Step S1343: Set a delay-related edge attribute value for each delay-related edge as the edge's delay attribute. The delay attribute is derived from the average response delay duration of the operation response hysteresis characteristic.

[0153] In addition to setting influence weights for associated edges, delay attributes also need to be set for delayed associated edges. Delayed associated edges connect instruction type nodes and device status nodes, reflecting the device's response delay to operation instructions.

[0154] The average response delay duration from the operation response hysteresis feature is used as the delay attribute of delay-related edges. For each delay-related edge, the delay attribute value is set according to the average response delay duration of the corresponding instruction type. For example, if the average response delay duration of an instruction type is long, it indicates that the device responds slowly to the instruction, so a larger delay attribute value is assigned to the delay-related edge connecting the node of that instruction type and the device status node; conversely, if the average response delay duration is short, a smaller delay attribute value is assigned.

[0155] By setting delay attributes, the operational response performance of devices can be reflected in the adjacency network. This is crucial for analyzing the real-time operating status of devices and optimizing operational strategies. For example, in a traffic flow control system on a highway, by examining the delay attributes of delay-related edges, one can understand the system's response speed to different control commands, adjust command transmission strategies in a timely manner, and improve system operating efficiency.

[0156] Step S1344: Perform hierarchical partitioning on the adjacent nodes, designating directly connected temperature and humidity fluctuation nodes and instruction type nodes as first-order adjacent nodes, and designating other environmental parameter nodes or instruction classification nodes connected through first-order adjacent nodes as second-order adjacent nodes.

[0157] To more clearly illustrate the structure of the adjacency network, the adjacency nodes are hierarchically divided.

[0158] Temperature and humidity fluctuation nodes and instruction type nodes that are directly connected to the central node (the mapping node of the current device state) through associated edges are designated as first-order adjacency nodes. First-order adjacency nodes are directly related to the device state, and their impact on the device state is relatively direct and obvious.

[0159] Other environmental parameter nodes or command classification nodes connected through first-order adjacency nodes are designated as second-order adjacency nodes. The relationship between second-order adjacency nodes and equipment status is relatively indirect; they influence equipment status through first-order adjacency nodes. For example, for bridge monitoring equipment on highways, first-order adjacency nodes might be temperature and humidity fluctuation nodes and operation command nodes that directly affect equipment status, while second-order adjacency nodes might be other environmental parameter nodes related to temperature and humidity (such as wind speed, air pressure, etc.) or more granular command classification nodes.

[0160] By dividing the network into hierarchical parts, the adjacency network can be organized into a hierarchical structure, which facilitates subsequent analysis and reasoning. For example, when analyzing the cause of equipment failure, one can start with the first-order adjacency nodes to check the directly related environmental factors and operating instructions; if the problem is still not resolved, the influence of the second-order adjacency nodes can be further considered.

[0161] Step S1345: Construct an adjacency network containing node hierarchy and edge attribute information based on the hierarchy of adjacent nodes, edge weights, and delay attributes.

[0162] After completing the hierarchical partitioning of adjacent nodes and setting weight values ​​and delay attributes for edges, an adjacency relationship network containing node hierarchical structure and edge attribute information is constructed.

[0163] The adjacency network is represented graphically, with the central node located at the center of the graph. First-order adjacency nodes are directly connected around the central node, while second-order adjacency nodes are indirectly connected to the central node through first-order adjacency nodes.

[0164] In the graph, each node is labeled with its meaning, such as device status node, temperature and humidity fluctuation node, and command type node. Simultaneously, each edge is labeled with its weight value and delay attribute to visually demonstrate the degree of influence of various factors on the device status and the device's operational response.

[0165] The constructed adjacency network can serve as an important analytical tool for studying the evolution of equipment states and the interactions between various factors. For example, for intelligent parking systems on highways, analyzing the adjacency network can reveal the impact paths and intensity of environmental factors (such as temperature and humidity) and operating commands on the system state, predict potential system failures, and formulate corresponding maintenance and management strategies.

[0166] Step S135: Perform path traversal processing on the adjacency relationship network, extract all possible evolution paths from the initial device state node to the abnormal device state node, and generate a set of device state evolution paths.

[0167] After the adjacency network is constructed, it undergoes path traversal to extract all possible evolution paths from the initial device state node to the abnormal device state node. Path traversal helps to understand how the device state evolves from a normal state to an abnormal state, and the role of various factors in this process.

[0168] Graph traversal algorithms, such as depth-first search or breadth-first search, can be used to traverse the system starting from the initial device state node. The initial device state node represents the normal operating state of the device at a certain moment, while the abnormal device state node indicates that the device has malfunctioned or is experiencing performance abnormalities.

[0169] Taking the depth-first search algorithm as an example, starting from the initial device state node, the algorithm explores along the edges in the adjacency network. At each node reached, it is checked whether the node is an abnormal device state node. If not, the algorithm continues to explore its unvisited neighboring nodes along its adjacent edges. During the exploration process, the nodes and edges traversed are recorded, forming a path. When an abnormal device state node is reached, this path is recorded.

[0170] During traversal, to avoid repeatedly visiting already explored paths, it is necessary to mark visited nodes and edges. For example, a marker array can be used to record the visit status of each node; once a node is visited, it is marked as visited.

[0171] Since there may be multiple paths from the initial device state node to the abnormal device state node in the adjacency network, the above search process needs to be repeated until all possible paths have been traversed.

[0172] During the traversal, the edge weights and latency attributes also need to be considered. The edge weights represent the degree of influence of adjacent nodes on the device state, while the latency attributes represent the device's response delay to operation commands.

[0173] After traversal, all paths from the initial device state node to the abnormal device state node are summarized to generate a set of device state evolution paths. This set of device state evolution paths contains all possible development paths of the device from a normal state to an abnormal state. For example, for tunnel ventilation equipment on a highway, the set of device state evolution paths can show all possible paths from normal operation to failure under different environmental factors and operating instructions, helping managers to take preventive measures in advance.

[0174] Step S136: Analyze the influence weights and delay-related edge attribute values ​​of each node in each evolution path to determine the description of the relationship between influencing factors.

[0175] After obtaining the set of equipment state evolution paths, each path needs to be analyzed in depth to determine the description of the relationships between influencing factors. This description illustrates the strength and interrelationships of different factors during the equipment state evolution process.

[0176] For each evolution path, the influence weight of each node is first analyzed. The influence weight reflects the degree of influence of adjacent nodes on the device state. By analyzing the influence weight, we can understand the importance of each factor in the evolution of the device state. For example, in a certain evolution path, a node with a large influence weight for temperature and humidity fluctuations indicates that the temperature and humidity fluctuations have a significant impact on the device state; while a node with a specific instruction type has a small influence weight, indicating that the instruction has a relatively small impact on the device state.

[0177] Simultaneously, the attribute values ​​of delayed-related edges are analyzed. These attribute values ​​represent the delay in the device's response to operational commands, reflecting the time factor in the device's state evolution process. A large attribute value for a delayed-related edge indicates a slow response to the operational command, which may lead to or exacerbate device malfunctions.

[0178] Based on the influence weights of each node and the attribute values ​​of delayed associated edges, the interrelationships between different factors are analyzed. For example, some factors may mutually reinforce each other's influence on the device state, while others may cancel each other out. By analyzing these interrelationships, a more comprehensive understanding of the mechanism of device state evolution can be achieved.

[0179] During the analysis, methods such as weighted summation can be used to comprehensively consider the influence weights of each node and the attribute values ​​of delayed associated edges. For each evolution path, the influence weights of each node are multiplied by the corresponding coefficients, and the influence of the attribute values ​​of delayed associated edges is added to obtain the comprehensive effect value of each factor on that path. By comparing the comprehensive effect values ​​of different paths, it is possible to determine which factors play a major role in the evolution of equipment state.

[0180] Based on the analysis results, a description of the relationships between influencing factors is generated. This description can be presented in textual form, explaining the strength of the influence of different factors during the equipment's state evolution, their interrelationships, and their impact on equipment failure. For example, for intelligent traffic signal systems on highways, the description could indicate that, under certain circumstances, inappropriate combinations of temperature and humidity fluctuations and operational commands may lead to signal system failures, with temperature and humidity fluctuations having a significant impact, and response delays in operational commands also exacerbating the occurrence of failures.

[0181] Step S140: Determine the current operating status assessment results and potential risk prediction results of the highway electromechanical system based on the decision association information.

[0182] Step S141: Analyze the set of device state evolution paths in the decision association information, count the frequency of the current device state node in the evolution path, and generate the stability assessment score of the current state.

[0183] In the set of device state evolution paths related to decision-making information, the frequency of occurrence of the current device state node is an important basis for evaluating the stability of the current device state.

[0184] Traverse the set of device state evolution paths and count the number of times the current device state node appears in each path. Sum the counts of the current device state node in all paths to obtain the total number of times the node appears.

[0185] Simultaneously, the total number of paths in the device state evolution path set is counted. The frequency of the current device state node in the evolution path is obtained by dividing the total number of occurrences of the current device state node by the total number of paths.

[0186] A stability assessment score for the current state is generated based on the frequency of occurrence. A mapping relationship can be preset to correspond the frequency of occurrence with the stability assessment score. For example, when the frequency of occurrence is low, it indicates that the current device state is relatively stable and is assigned a higher stability assessment score; when the frequency of occurrence is high, it indicates that the current device state is prone to developing into an abnormal state and is assigned a lower stability assessment score.

[0187] When generating a stability assessment score, the influence of other factors can also be considered. For example, the stability assessment score can be adjusted based on the degree of influence of each factor on the current state as described in the description of the relationships between influencing factors. If certain important factors have a significant negative impact on the stability of the current state, the stability assessment score can be appropriately reduced.

[0188] For power monitoring equipment on highways, a stability assessment score is generated by statistically analyzing the frequency of the occurrence of its current state node in the set of equipment state evolution paths. If the frequency is high, it indicates that the current state of the equipment is unstable and may require enhanced monitoring and maintenance.

[0189] Step S142: Extract the types of subsequent state nodes starting from the current state node in the evolution path, and count the occurrence ratio of abnormal state nodes as the probability of potential risks.

[0190] From the set of device state evolution paths, extract the types of subsequent state nodes starting from the current state node. The types of subsequent state nodes reflect the possible development direction of the device in the current state.

[0191] The extracted subsequent status nodes are categorized into normal status nodes and abnormal status nodes. Normal status nodes indicate that the device continues to operate normally, while abnormal status nodes indicate that the device has malfunctioned or is experiencing performance abnormalities.

[0192] The number of abnormal state nodes is counted and divided by the total number of subsequent state nodes originating from the current state node to obtain the occurrence ratio of abnormal state nodes. This occurrence ratio serves as the potential risk probability, representing the likelihood that the device will experience an anomaly in the future given its current state.

[0193] During the statistical process, it is essential to ensure the accuracy of subsequent status node classifications. Clear criteria for judging abnormal states can be established based on historical equipment failure data and expert experience. For example, for communication base station equipment on highways, when the signal strength of the equipment falls below a certain threshold or communication is interrupted, its corresponding status node is defined as an abnormal status node.

[0194] By statistically analyzing the occurrence rate of abnormal nodes, the probability of potential risks can be obtained. This probability helps managers anticipate the likelihood of equipment failure and take appropriate preventative measures. For example, a high probability of potential risk indicates a greater likelihood of equipment failure in the near future, requiring timely scheduling of maintenance personnel for inspection and repair.

[0195] Step S143: Based on the strength of the main factors in the description of the relationship between influencing factors, assess the contribution of each influencing factor to the stability of the current state, and generate a ranking of the importance of the influencing factors.

[0196] The description of the relationships between influencing factors includes the strength of each factor's influence and their interrelationships during the equipment's state evolution. Combining this information, the contribution of each influencing factor to the current state stability can be assessed, and a ranking of the influencing factors' importance can be generated.

[0197] First, extract the strength of each factor's influence from the description of their relationships. Influence strength can be reflected by indicators such as influence weight and the attribute value of delayed association edges. For example, the strength of influence for temperature and humidity fluctuation factors can be determined based on the weight value of the edge between the temperature and humidity fluctuation node and the device status node in the adjacency network; the strength of influence for operation command factors can be determined based on the delay attribute value of delayed association edges.

[0198] The contribution of each factor to the stability of the current state is assessed based on the strength of its effect. A weighted summation method can be used, multiplying the strength of each factor by its corresponding weight coefficient to obtain the contribution value of each factor to the stability of the current state. The weight coefficient can be determined based on the importance of the factor and its degree of influence on the equipment state. For example, temperature and humidity factors that have a significant impact on the equipment state can be assigned a higher weight coefficient, while less important factors can be assigned a lower weight coefficient.

[0199] By comparing the contribution values ​​of each factor and ranking them from largest to smallest, an importance ranking of the influencing factors is generated. This importance ranking helps managers understand which factors have the greatest impact on the stability of the current equipment state, allowing them to take targeted measures to improve equipment stability.

[0200] For example, for tunnel lighting systems on highways, the importance of each influencing factor is ranked by assessing its contribution to the stability of the current state. If the contribution of temperature and humidity fluctuations is significant, it indicates a need to strengthen the monitoring and control of temperature and humidity within the tunnel; if the contribution of operational instructions is significant, it indicates a need to optimize operational strategies and reduce the negative impact of operational instructions on the equipment's state.

[0201] Step S144: Normalize the stability assessment score and potential risk probability to generate a current operating status assessment result containing stability level identifier and risk level identifier.

[0202] Since the stability assessment score and the potential risk probability may have different ranges, they need to be normalized for easier comprehensive analysis and comparison.

[0203] Normalization can unify the range of stability assessment scores and potential risk probabilities into a relatively consistent interval, such as [0, 1]. Linear normalization can be used to process the stability assessment scores and potential risk probabilities separately.

[0204] For the stability assessment score, find its minimum and maximum values. Subtract the minimum value from the stability assessment score, and then divide by the difference between the maximum and minimum values ​​to obtain the normalized stability assessment score.

[0205] For the potential risk probability, we find its minimum and maximum values ​​and perform similar linear normalization to obtain the normalized potential risk probability.

[0206] Based on the normalized stability assessment score and potential risk probability, a current operational status assessment result is generated, including stability level and risk level indicators. Multiple level classification standards can be preset to map the normalized values ​​to the levels.

[0207] For stability level designations, a high normalized stability assessment score is defined as "stable"; a moderate score is defined as "relatively stable"; and a low score is defined as "unstable".

[0208] For risk level identification, when the normalized probability of potential risk is low, it is defined as "low risk"; when the probability is at a medium level, it is defined as "medium risk"; and when the probability is high, it is defined as "high risk".

[0209] Combining stability level and risk level labels forms the current operational status assessment result. For example, for traffic flow monitoring equipment on highways, through normalization and level classification, the current operational status assessment result is "relatively stable, medium risk," which can help managers quickly understand the current status and potential risks of the equipment.

[0210] Step S145: Based on the trend of subsequent state node changes in the device state evolution path, predict the types of abnormal states that may occur within a preset time period and their occurrence time range, and generate potential risk prediction results containing abnormal type identifiers and time range descriptions.

[0211] For example, step S1451: Extract all subsequent state node sequences starting from the current state node in the device state evolution path, and count the frequency of each abnormal state type in the sequence.

[0212] Within the set of device state evolution paths, we focus on paths originating from the current state node. For each of these paths, we extract the sequence of subsequent state nodes. This sequence records the possible evolution of the device starting from the current state.

[0213] All extracted subsequent state node sequences are summarized. Abnormal state nodes are identified within these sequences. The identification of abnormal state nodes can be based on pre-defined criteria, such as device performance indicators and operating parameters.

[0214] Different types of abnormal state nodes are classified and statistically analyzed. The frequency of occurrence of each abnormal state type is obtained by counting the number of times it appears in all subsequent state node sequences.

[0215] For example, for automated toll collection equipment on highways, abnormal state types may include toll collection system malfunctions, communication interruptions, and data errors. By statistically analyzing the frequency of these abnormal state types in the sequence of subsequent state nodes starting from the current state node, we can understand the likelihood of each abnormal state occurring in the future.

[0216] Step S1452: Select the most frequently occurring abnormal state type as the main potential abnormal type.

[0217] After counting the frequency of each abnormal state type in the subsequent state node sequence, these frequencies are compared.

[0218] Identify the most frequently occurring abnormal state type. The higher the frequency of this abnormal state type in the device's state evolution path, the more likely the device is to experience this type of abnormality in the future.

[0219] The most frequently occurring abnormal state types were identified as the primary potential anomaly types. These primary potential anomaly types are the focus of attention for predicting future equipment failures.

[0220] For example, for meteorological monitoring equipment on highways, if temperature sensor failure occurs most frequently in subsequent state node sequences, then temperature sensor failure is considered the primary potential anomaly type. This means that the meteorological monitoring equipment is most likely to experience temperature sensor failure within a predetermined time period.

[0221] Step S1453: Analyze the time interval characteristics of each node in the evolution path corresponding to the main potential anomaly types, and calculate the average time interval from the current state node to the abnormal state node.

[0222] For the main potential anomaly types, their corresponding evolutionary paths are identified. Within these evolutionary paths, the time interval characteristics between each node are analyzed.

[0223] The time interval characteristic reflects the time required for the device state to transition from one node to another. For each major potential anomaly type, the time interval between each node from the current state node to the anomaly state node is recorded.

[0224] The time intervals from the current state node to the anomalous state node in the evolution paths corresponding to all major potential anomaly types are summarized. The sum of these time intervals is calculated and then divided by the total number of paths to obtain the average time interval from the current state node to the anomalous state node.

[0225] Average time intervals can serve as an important reference for predicting the timing of abnormal states. For example, in a road lighting control system on a highway, if the main potential anomaly type is lamp failure, by analyzing the time interval characteristics of each node in its corresponding evolution path, the average time interval from the current state node to the lamp failure abnormal state node can be calculated, thereby predicting the approximate time when the lamp failure may occur.

[0226] Step S1454: Combine the running time of the current cycle with the current time point as the starting point and add the average time interval to generate the predicted occurrence time of the abnormal state.

[0227] After obtaining the average time interval from the current state node to the abnormal state node, a prediction is made by combining it with the running time of the current running cycle.

[0228] The elapsed duration of the current operating cycle represents the time elapsed from the start of the current operating cycle to the current moment. It is calculated by combining the average time interval with the elapsed duration of the current operating cycle, starting from the current time point.

[0229] The average time interval is converted to a duration consistent with the current operating cycle's time unit. For example, if the current operating cycle's time unit is hours, then the average time interval also needs to be converted to hours. Then, by adding the converted average time interval to the current time point, the predicted time point of occurrence of the abnormal state can be obtained.

[0230] This requires ensuring the accuracy of time calculations, taking into account both the continuity and periodicity of time. For example, in highway electromechanical systems, for traffic signal control equipment, if its operating cycle is measured in days, the specific characteristics of different time periods within a day and the changing of dates must be considered when calculating and predicting the occurrence time. If the average time interval calculation results span across days, the date and time need to be accurately adjusted.

[0231] Using the above calculation method, the approximate time when the main potential anomalies of the equipment will occur can be predicted relatively accurately. For example, for vehicle weighing equipment on highways, based on the calculated average time interval and the duration of the current operating cycle, the specific time point when the main potential anomaly of inaccurate weighing data may occur can be predicted. In this way, relevant personnel can make preparations in advance, such as arranging maintenance personnel and preparing replacement parts.

[0232] Step S1455: Combine the fluctuation range of time intervals in the evolution path to expand the generation time range description with the predicted occurrence time point as the center.

[0233] After obtaining the predicted time of occurrence of the abnormal state, it is also necessary to consider that the time interval may fluctuate in the actual situation. The fluctuation range of the time interval in the evolution path reflects the uncertainty of the equipment state transition time.

[0234] First, analyze the time interval data from the current state node to the abnormal state node in the evolution path, and statistically analyze the fluctuation of these time intervals. The fluctuation range can be measured by calculating the difference between the maximum and minimum values ​​of the time intervals, or by calculating the standard deviation of the time intervals.

[0235] Centered on the predicted occurrence time, the range of fluctuation is expanded. The fluctuation range can be evenly distributed before and after the predicted occurrence time to form a time interval. For example, if the fluctuation range is a specific duration, half of that duration is added before the predicted occurrence time, and the other half is added after the predicted occurrence time to obtain a time interval.

[0236] When expanding the time range, it is necessary to ensure its rationality and logic. The time range should not exceed the reasonable operating cycle of the equipment, and potential limiting factors in actual situations must be taken into account. For example, for tunnel monitoring equipment on highways, there are certain maintenance cycles and usage specifications; the time range should not exceed these reasonable time limits.

[0237] The generated time range description should be clear and concise, accurately conveying the approximate time period during which abnormal conditions may occur. For various equipment in highway electromechanical systems, such as communication base stations and lighting facilities, generating a time range description by combining the fluctuation range of time intervals allows managers to have a more comprehensive understanding of the possible time intervals in which equipment anomalies may occur, thereby enabling better scheduling of maintenance plans and resource allocation. For example, for communication base station equipment, the time range description can clearly indicate that the time interval during which the main potential anomaly type of equipment, signal interruption, may occur is from a specific time period on a specific date to another time period. This allows managers to strengthen monitoring within that time interval and make advance emergency preparations.

[0238] Step S150: Generate autonomous decision-making instructions for the highway electromechanical system based on the operational status assessment results and potential risk prediction results. The autonomous decision-making instructions are used to trigger equipment adjustment operations or resource scheduling operations.

[0239] By combining the operational status assessment results and potential risk prediction results of the integrated highway electromechanical system, corresponding autonomous decision-making instructions can be generated. The purpose of these autonomous decision-making instructions is to ensure the stable operation of the highway electromechanical system and reduce potential risks by triggering equipment adjustment operations or resource scheduling operations.

[0240] Based on the stability and risk levels identified in the operational status assessment, the overall current status of the system is determined. If the stability level is "stable" and the risk level is "low risk," it indicates that the system is currently operating well, but some monitoring is still required. The generated autonomous decision-making instructions at this point may focus on maintaining the existing equipment's operational status, performing routine inspections and maintenance. For example, for highway lighting equipment, it might instruct it to continue operating at the current brightness settings and schedule regular lamp checks.

[0241] If the stability level is "relatively stable" but the risk level is "medium risk," then certain adjustments need to be made to the system to reduce the risk. Autonomous decision-making commands may trigger equipment adjustment operations, fine-tuning the parameters of relevant equipment. For example, for traffic signal control equipment on highways, if the risk of inaccurate signal timing is predicted, the decision-making command may adjust the signal cycle length or phase time to optimize traffic flow and reduce the likelihood of accidents.

[0242] When the stability level is "unstable" and the risk level is "high risk," more proactive measures are required. Autonomous decision-making instructions may trigger resource scheduling operations, allocating more resources to ensure the system's normal operation. For example, for communication base station equipment on highways, if a high risk of communication outage is predicted, the decision-making instructions may arrange for backup base stations to be put into use, or increase maintenance personnel to conduct real-time monitoring and maintenance on-site.

[0243] Based on the descriptions of the main potential anomaly types and time ranges in the potential risk prediction results, the autonomous decision-making instructions are further refined. If a specific piece of equipment is predicted to malfunction within a specific time range, preventative adjustments can be made to the equipment before that time range begins. For example, for vehicle weighing equipment on highways, if inaccurate weighing data is predicted to occur in the future, the equipment can be calibrated and adjusted before the start of that time range.

[0244] When generating autonomous decision-making instructions, it is necessary to ensure their feasibility and effectiveness. Instructions must comply with equipment operating procedures and system operational requirements, while also considering resource availability and the rationality of allocation. For example, when performing resource scheduling operations, it is necessary to ensure that backup equipment can be properly connected to the system and that maintenance personnel possess the appropriate skills and tools.

[0245] The generated autonomous decision-making instructions must be clearly and explicitly communicated to the relevant implementing departments or equipment. Instructions can be sent to the equipment's control system via the system's communication interface or to notify relevant maintenance personnel. The instructions must include specific operational details, timing, and target information to ensure accurate understanding and execution by the personnel. For example, for power supply equipment on a highway, an autonomous decision-making instruction might explicitly specify the parameter adjustment of a particular transformer at a specific time to address potential power fluctuation risks.

[0246] Based on the above steps, through comprehensive analysis of the operation data of highway electromechanical equipment, and by using knowledge graphs for correlation analysis and decision reasoning, the resulting autonomous decision-making instructions can effectively ensure the stable operation of highway electromechanical systems, reduce potential risks, and improve the safety and efficiency of highway traffic.

[0247] Figure 2 The illustration shows exemplary hardware and software components of a highway electromechanical autonomous decision-making system 100 based on big data and knowledge graphs, which can implement the ideas of this application, according to some embodiments of this application. For example, processor 120 can be used in the highway electromechanical autonomous decision-making system 100 based on big data and knowledge graphs and to perform the functions in this application.

[0248] The highway electromechanical autonomous decision-making system 100 based on big data and knowledge graphs can be a general-purpose server or a special-purpose server; both can be used to implement the highway electromechanical autonomous decision-making method based on big data and knowledge graphs of this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the load.

[0249] For example, the highway electromechanical autonomous decision-making system 100 based on big data and knowledge graphs may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the highway electromechanical autonomous decision-making system 100 based on big data and knowledge graphs may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The methods of this application can be implemented according to these program instructions. The highway electromechanical autonomous decision-making system 100 based on big data and knowledge graphs also includes an I / O interface 150 between the computer and other input / output devices.

[0250] Furthermore, this embodiment of the invention also provides a readable storage medium, wherein computer-executable instructions are preset in the readable storage medium, and when the processor executes the computer-executable instructions, the above-mentioned highway electromechanical autonomous decision-making method based on big data and knowledge graph is realized.

[0251] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A highway electromechanical autonomous decision-making method based on big data and knowledge graphs, characterized in that, The method includes: Acquire a set of operational data generated by highway electromechanical equipment during a continuous operating cycle, the set of operational data including equipment status information, environmental perception information and operation instruction records; The operational data set is subjected to feature extraction processing to obtain a key feature set reflecting the equipment's operating status. The key feature set includes equipment performance correlation features, environmental impact coupling features, and operational response lag features. The pre-built knowledge graph is invoked to perform correlation analysis on the key feature set, generating decision-related information containing the device state evolution path and the relationship between influencing factors; Based on the aforementioned decision-related information, the current operating status assessment results and potential risk prediction results of the highway electromechanical system are determined. Based on the operational status assessment results and potential risk prediction results, autonomous decision-making instructions for the highway electromechanical system are generated. These autonomous decision-making instructions are used to trigger equipment adjustment operations or resource scheduling operations. The process of calling a pre-built knowledge graph to perform association analysis on the key feature set generates decision association information containing the device state evolution path and the relationships between influencing factors, including: The stability distribution pattern is extracted from the device performance correlation features of the key feature set, and the corresponding device state node is matched in the knowledge graph to establish the correlation between the stability distribution pattern and the device state node. The table mapping relationship between fluctuation amplitude and anomaly probability in the environmental impact coupling characteristics is input into the relation reasoning module of the knowledge graph to analyze the influence weight between temperature and humidity fluctuation nodes and equipment status nodes. By utilizing the average response delay duration in the operation response hysteresis characteristics, the delay association edge between the instruction type node and the device status node is found in the knowledge graph, and the attribute value of the delay association edge is updated. Based on the aforementioned association relationships, influence weights, and delayed association edge attribute values, an adjacency relationship network for device status nodes is constructed. This adjacency relationship network includes connection paths between device status nodes and temperature and humidity fluctuation nodes, as well as instruction type nodes. The adjacency network is traversed to extract all possible evolution paths from the initial device state node to the abnormal device state node, generating a set of device state evolution paths. Analyze the influence weights and delay-related edge attribute values ​​of each node in each evolution path to determine the description of the relationships between influencing factors; The step of extracting stability distribution patterns from the key feature set of equipment performance correlation features, matching the corresponding equipment state nodes in the knowledge graph, and establishing the association between stability distribution patterns and equipment state nodes includes: Analyze the stability distribution pattern in the correlation characteristics of equipment performance, and extract the proportion of stable state and the proportion of abrupt state in the persistence characteristics of trend type; In the device state node library of the knowledge graph, search for historical device state nodes that contain the same percentage of stable states and percentage of abrupt states. Extract the device operation scenario description and subsequent state change records corresponding to the historical device status nodes, and generate a matching score between the current stability distribution pattern and the historical device status nodes; Select the historical device status node with the highest matching score as the mapping node for the current device status; Establish direct association edges between the current stability distribution pattern and the mapping nodes. The attribute values ​​of the direct association edges are the matching degree score and historical state change records.

2. The highway electromechanical autonomous decision-making method based on big data and knowledge graphs according to claim 1, characterized in that, The feature extraction process performed on the operational data set to obtain a set of key features reflecting the operating status of the equipment includes: The operational data set is subjected to feature filtering processing to filter out redundant data items that are not directly related to the equipment's operating status, while retaining the functional module working status parameters in the equipment status information, the temperature and humidity fluctuation records in the environmental perception information, and the instruction type sequence in the operation instruction records. The filtered equipment status information is subjected to time-series correlation analysis, and the changing trend characteristics of the working status parameters of the functional modules within the continuous operating cycle are extracted as equipment performance correlation characteristics. The equipment performance correlation characteristics include the stability distribution pattern of the status parameters and the sudden change triggering conditions. The filtered environmental perception information is subjected to multi-source data coupling analysis and processing. The synchronous change characteristics of temperature and humidity fluctuation records and equipment status information are extracted as environmental impact coupling characteristics. The environmental impact coupling characteristics include a description of the correlation strength between temperature and humidity fluctuation amplitude and equipment status parameters. The filtered operation instruction records are subjected to response delay analysis. The time difference between the instruction type sequence and the change of equipment status parameters is extracted as the operation response lag feature. The operation response lag feature includes the average response delay duration and delay fluctuation range corresponding to different instruction types. The key feature set is obtained by standardizing and splicing the equipment performance correlation characteristics, environmental impact coupling characteristics, and operational response lag characteristics.

3. The highway electromechanical autonomous decision-making method based on big data and knowledge graphs according to claim 2, characterized in that, The step of performing time-series correlation analysis on the filtered equipment status information and extracting the changing trend characteristics of the functional module working status parameters within a continuous operating cycle as equipment performance correlation characteristics includes: Arrange the functional module working status parameters in the device status information according to the timestamp order to generate a time series data chain; The time series data chain is subjected to sliding window segmentation to obtain multiple sub-data windows containing a preset time span; Perform trend fitting processing on the working status parameters in each sub-data window to generate trend type labels that reflect whether the parameters are rising, falling, or stable. The frequency of different trend type identifiers in continuous sub-data windows is statistically analyzed, and the persistence characteristics of trend types are extracted as the stability distribution pattern of state parameters. Detect the change range of working status parameters between adjacent sub-data windows, record the trigger time point when the change range exceeds the preset threshold and the corresponding device operation scenario, and generate a description of the trigger condition for the trend change. The stability distribution pattern and triggering conditions are combined to form the device performance correlation characteristics.

4. The highway electromechanical autonomous decision-making method based on big data and knowledge graphs according to claim 2, characterized in that, The process of performing multi-source data coupling analysis on the screened environmental perception information, and extracting the synchronous change characteristics of temperature and humidity fluctuation records and equipment status information as environmental impact coupling characteristics, includes: The temperature and humidity fluctuation records in the environmental sensing information are aligned with the functional module working status parameters in the equipment status information according to the timestamp, and a time-synchronized multi-source data pair is generated. For each pair of multi-source data, correlation calculation is performed, and the consistency ratio between the temperature and humidity fluctuation amplitude and the change direction of the functional module's working status parameters is statistically analyzed to obtain the correlation calculation results. Extract the probability of abnormal occurrence of functional module working status parameters when the temperature and humidity fluctuation range reaches the set range, and generate a mapping relationship table between fluctuation range and abnormal probability; Analyze the degree of matching between the periodic characteristics of temperature and humidity fluctuations and the periodic changes of equipment status parameters, and extract the synchronization index between the fluctuation period and the status period. The correlation calculation results, mapping relationship table, and synchronization index are combined to form the environmental impact coupling characteristics.

5. The highway electromechanical autonomous decision-making method based on big data and knowledge graphs according to claim 2, characterized in that, The process of performing response delay analysis on the filtered operation instruction records, and extracting the time difference features between the instruction type sequence and the changes in equipment status parameters as operation response hysteresis features, includes: Traverse the sequence of instruction types in the operation instruction record and mark the sending timestamp of each operation instruction; Locate the start timestamp of the status parameter change corresponding to each operation command in the device status information, and calculate the difference between the sending timestamp and the start timestamp of the change as the response delay time of a single operation command; The response delay duration is classified and statistically analyzed according to the instruction type. The average response delay duration and the standard deviation of the delay duration for the same type of operation instruction are calculated as the delay fluctuation range. Analyze the relationship between the time interval between consecutively sent similar operation commands and the response delay duration, and extract the characteristics of the impact of command sending frequency on response delay; The average response delay duration, delay fluctuation range, and influencing characteristics are combined to form the operational response hysteresis characteristics.

6. The autonomous decision-making method for highway electromechanical systems based on big data and knowledge graphs according to claim 1, characterized in that, The construction of the adjacency network for device state nodes based on the aforementioned association relationships, influence weights, and delayed association edge attribute values ​​includes: Using the mapping node of the current device status as the central node, collect all temperature and humidity fluctuation nodes and instruction type nodes connected by associated edges as adjacent nodes; An influence weight is assigned to each associated edge as the edge's weight value. The influence weight is derived from the correlation calculation results of the environmental influence coupling characteristics. For each delayed associated edge, a delayed associated edge attribute value is set as the edge's delay attribute, wherein the delay attribute is derived from the average response delay duration of the operation response hysteresis characteristic; The adjacent nodes are hierarchically divided, with directly connected temperature and humidity fluctuation nodes and instruction type nodes as first-order adjacent nodes, and other environmental parameter nodes or instruction classification nodes connected through first-order adjacent nodes as second-order adjacent nodes. Based on the hierarchy of adjacent nodes and the weight and delay attributes of edges, an adjacency network containing node hierarchy and edge attribute information is constructed.

7. The autonomous decision-making method for highway electromechanical systems based on big data and knowledge graphs according to claim 1, characterized in that, The determination of the current operating status assessment results and potential risk prediction results of the highway electromechanical system based on the decision correlation information includes: Analyze the set of device state evolution paths in the decision-related information, count the frequency of the current device state node in the evolution path, and generate a stability assessment score for the current state. Extract the types of subsequent state nodes starting from the current state node in the evolution path, and count the proportion of abnormal state nodes as the probability of potential risks. Based on the strength of the main factors in the description of the relationships between influencing factors, assess the contribution of each influencing factor to the stability of the current state, and generate a ranking of the importance of the influencing factors; The stability assessment score and potential risk probability are normalized to generate a current operating status assessment result that includes a stability level identifier and a risk level identifier. Based on the changing trends of subsequent state nodes in the device state evolution path, the types of abnormal states that may occur within a preset time period and their occurrence time range are predicted, generating potential risk prediction results that include abnormal type identifiers and time range descriptions.

8. A highway electromechanical autonomous decision-making system based on big data and knowledge graphs, characterized in that, The system includes a processor and a memory, the memory being connected to the processor. The memory is used to store programs, instructions, or code, and the processor is used to execute the programs, instructions, or code in the memory to implement the autonomous decision-making method for highway electromechanical systems based on big data and knowledge graphs as described in any one of claims 1-7.

Images