Rail transit air conditioning system intelligent monitoring method and device

Abstract

The present application belongs to the technical field of intelligent monitoring of rail transit, and particularly relates to a kind of intelligent monitoring method and device for rail transit air conditioning system.The present application uses a sensor array with adaptive sampling frequency to collect multi-modal operating data, then pre-processes the collected multi-modal operating data, and establishes a feature degradation trajectory atlas, providing strong data support for subsequent fault warning and diagnosis, introduces historical abnormal events to dynamically correct the health status evaluation baseline, ensuring the timeliness and accuracy of the evaluation baseline.In the comparative analysis of real-time operating data and dynamic evaluation baseline, the health score and dynamic threshold mechanism are used to realize the quantitative evaluation of the health status of the air conditioning system.By analyzing the propagation path and timing correlation of abnormal parameters in the air conditioning system, combined with the equipment topology graph, a decision graph containing fault diagnosis and predictive maintenance suggestions is generated to improve the efficiency of fault troubleshooting and repair.

Description

Technical Field

[0001] This invention belongs to the field of intelligent monitoring technology for rail transit, specifically relating to an intelligent monitoring method and device for rail transit air conditioning systems. Background Technology

[0002] With the rapid development of rail transit, the stability and reliability of the air conditioning system, as an important component of rail transit vehicles, are crucial to ensuring passenger comfort and driving safety. As trains operate under high load and variable environmental conditions for extended periods, the air conditioning system is prone to various malfunctions, affecting the normal operation of the train and the passenger experience. Therefore, ensuring the stable operation of the air conditioning system and promptly identifying and addressing potential malfunctions are of paramount importance.

[0003] While existing technologies offer methods for monitoring the operational status of air conditioning systems, most rely on manual inspections or periodic maintenance, which are not only inefficient but also make it difficult to detect and address potential faults in a timely manner. Furthermore, although some sensor-based monitoring systems exist, they often only issue alarms after a fault occurs, lacking predictability and intelligence. Moreover, due to the complexity of air conditioning system networks, once a fault point is detected, the status of other related equipment cannot be determined in a timely manner, leading to low efficiency in fault diagnosis and repair. Based on this, this solution proposes an intelligent monitoring method for rail transit air conditioning systems to address the aforementioned problems. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent monitoring method and device for rail transit air conditioning systems, which can monitor the operating status of the air conditioning system in real time, improve the efficiency of fault detection and handling, and realize intelligent and predictive maintenance.

[0005] The specific technical solution adopted by this invention is as follows:

[0006] A method for intelligent monitoring of rail transit air conditioning systems, comprising:

[0007] Multimodal operating data of the air conditioning system are collected in real time through a distributed sensor array, including temperature parameters, electrical parameters and mechanical status parameters;

[0008] Dimensionality reduction is performed on multimodal operational data to eliminate redundant information, key feature vectors are extracted, and feature degradation trajectory maps are established.

[0009] A health status assessment baseline for the air conditioning system is constructed based on historical operating data, and historical abnormal events are introduced to dynamically correct the health status assessment baseline, outputting a dynamic assessment baseline.

[0010] By comparing and analyzing real-time operational data with dynamic evaluation baselines, a comprehensive evaluation report containing health level and fault warning information is generated.

[0011] Obtain the propagation path and temporal correlation of abnormal parameters within the air conditioning system, and combine this with the equipment topology diagram within the air conditioning system to generate a decision map that includes fault diagnosis and predictive maintenance suggestions.

[0012] In a preferred embodiment, the distributed sensor array employs an adaptive sampling frequency when collecting multimodal operational data, dynamically adjusting the sampling interval according to the changing trend of the multimodal operational data. The specific steps are as follows:

[0013] Preset the initial sampling frequency for each sensor in the sensor array;

[0014] Establish a sliding time window and calculate the rate of change gradient of the multimodal running data in real time within each sliding time window;

[0015] Compare the rate of change gradient with a preset tolerance threshold;

[0016] When the rate of change gradient exceeds the tolerance threshold, the automatic adjustment mechanism of the sampling interval is triggered, and the sampling interval of the corresponding sensor is adjusted according to the adjustment result.

[0017] If, after the automatic adjustment mechanism for the sampling interval is triggered, the rate of change gradient within N consecutive sliding time windows still exceeds the tolerance threshold, the sampling frequency will be directly increased to the preset upper limit.

[0018] In a preferred embodiment, the automatic sampling interval adjustment mechanism includes the following steps:

[0019] Collect the difference between the rate of change gradient and the corresponding tolerance threshold, and record it as an adjustment condition parameter;

[0020] The sampling interval of the corresponding sensor is calculated based on the adjustment condition parameters, and when the adjustment amount of the sampling interval exceeds the adjustment threshold, the sampling frequency of the corresponding sensor is directly increased to the preset upper limit.

[0021] Adjustment signals are sent to similar sensors in the same area to synchronously adjust the sampling frequency of similar sensors.

[0022] In a preferred embodiment, the steps of performing dimensionality reduction processing on the multimodal operational data, eliminating redundant information, extracting key feature vectors, and establishing a feature degradation trajectory map include:

[0023] Multimodal operation data is segmented by using a sliding time window to generate data segments that include parameter correlations;

[0024] The data segments are processed using standard methods, and then a hybrid dimensionality reduction method combining principal component analysis and kernel density estimation is used to calculate the principal component scores and kernel density distribution of the data segments, and to screen key feature vectors.

[0025] The key feature vectors are time-aligned, and then a feature degradation trajectory map is constructed. The feature degradation trend and key inflection points are marked in the degradation trajectory map.

[0026] In a preferred embodiment, the step of constructing a health status assessment baseline for the air conditioning system based on historical operating data includes:

[0027] Historical operating data is cleaned and standardized to remove abnormal operating condition data, forming a standardized training dataset containing temperature, electrical, and mechanical parameters.

[0028] Based on the train operation mode and real-time environmental parameters, the standardized training dataset is divided into data subsets under multiple operating conditions;

[0029] Cluster analysis was performed on each data subset to extract health status features under each working condition, and the health status features were time-series aligned to construct a health status assessment baseline.

[0030] In a preferred embodiment, the step of dynamically correcting the health status assessment baseline by introducing historical abnormal events and outputting the dynamic assessment baseline includes:

[0031] Extract the abnormal feature vectors from historical abnormal events and match them with the feature vectors corresponding to the health status assessment baseline to output the feature deviation under the abnormal events;

[0032] A time decay function is constructed based on feature deviation, and the impact weights of historical anomalous events are dynamically allocated.

[0033] The dynamically allocated influence weights are integrated with the historical health status assessment baseline to correct the health status assessment baseline and generate a dynamic assessment baseline.

[0034] In a preferred embodiment, the step of comparing and analyzing real-time operational data with a dynamic evaluation baseline to generate a comprehensive evaluation report containing health level and fault warning information includes:

[0035] The real-time running data is standardized to generate a comparative feature vector with the same dimensions as the dynamic evaluation baseline;

[0036] The similarity between the comparative feature vector and the dynamic evaluation baseline is calculated, and the calculation result is output as a health score. The health score is then compared with the preset health evaluation threshold to identify the operating status.

[0037] If the health score is lower than the health assessment threshold, an abnormal warning mechanism is triggered. Based on the difference between the comparative feature vector and the dynamic assessment baseline, the health level is determined, and fault warning information is generated according to the health level, and a comprehensive assessment report is output.

[0038] If the health score is higher than the health assessment threshold, the system will maintain normal operation, record current operating data, and continuously monitor changes in health.

[0039] The health assessment threshold is a dynamic threshold, which is adaptively adjusted based on fluctuations in real-time operational data and the frequency of historical abnormal events. The specific adjustment process is as follows:

[0040] Construct a short-term sliding window and calculate the short-term volatility coefficient based on the fluctuation range of real-time running data within the short-term sliding window;

[0041] Construct a long-term sliding window, and count the frequency of historical abnormal events within the long-term sliding window. Calculate a frequency correction factor based on the frequency of historical abnormal events within the long-term sliding window.

[0042] By combining the short-term volatility coefficient and the frequency correction factor, the data is input into the threshold adjustment function to generate a dynamically adjusted health assessment threshold.

[0043] In a preferred embodiment, the step of obtaining the propagation path and temporal correlation of abnormal parameters within the air conditioning system, and generating a decision map containing fault diagnosis and predictive maintenance suggestions by combining it with the equipment topology diagram within the air conditioning system, includes:

[0044] Based on the spatial distribution of characteristic deviation, analyze the propagation path of abnormal parameters between devices and determine the temporal correlation.

[0045] By comparing the time series differences of multi-source sensor data, the propagation delay of abnormal events between different devices can be identified, forming a time-stamped fault propagation chain;

[0046] By combining the connection tightness of the equipment topology diagram, the key equipment and potential impact range related to abnormal parameters are identified, and the key equipment is marked as a potential fault source to generate fault diagnosis results.

[0047] Based on the time markers of the fault propagation chain, the failure development trend of equipment within the potential impact range is predicted, preventive maintenance strategies are formulated, and a decision map is formed.

[0048] The present invention also provides an intelligent monitoring system for rail transit air conditioning systems, using the above-mentioned intelligent monitoring method for rail transit air conditioning systems, comprising:

[0049] The parameter acquisition module is used to collect multimodal operating data of the air conditioning system in real time through a distributed sensor array, including temperature parameters, electrical parameters and mechanical status parameters;

[0050] The parameter preprocessing module is used to reduce the dimensionality of multimodal operating data, eliminate redundant information, extract key feature vectors, and establish a feature degradation trajectory map.

[0051] The baseline optimization module is used to construct a health status assessment baseline for the air conditioning system based on historical operating data, and to dynamically correct the health status assessment baseline by incorporating historical abnormal events, and output a dynamic assessment baseline.

[0052] The evaluation module is used to compare and analyze real-time operating data with dynamic evaluation baselines to generate a comprehensive evaluation report that includes health level and fault warning information.

[0053] The fault diagnosis module is used to obtain the propagation path and temporal correlation of abnormal parameters in the air conditioning system, and generate a decision map containing fault diagnosis and predictive maintenance suggestions by combining the equipment topology diagram in the air conditioning system.

[0054] And, an intelligent monitoring device for a rail transit air conditioning system, the monitoring device comprising:

[0055] At least one processor;

[0056] and a memory communicatively connected to the at least one processor;

[0057] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the above-described intelligent monitoring method for rail transit air conditioning systems.

[0058] The technical effects achieved by this invention are as follows:

[0059] This invention utilizes an adaptive sampling frequency sensor array to dynamically adjust the sampling strategy based on data change trends. This ensures data comprehensiveness while effectively reducing data volume and improving data processing efficiency. By performing dimensionality reduction and feature extraction on the collected multimodal operational data, a feature degradation trajectory map is established, providing strong data support for subsequent fault early warning and diagnosis. Historical abnormal events are introduced to dynamically correct the health status assessment baseline, ensuring the timeliness and accuracy of the assessment baseline. In the comparative analysis of real-time operational data and the dynamic assessment baseline, a health score and dynamic threshold mechanism are used to achieve a quantitative assessment of the health status of the air conditioning system, providing a reference for fault early warning and decision-making. By analyzing the propagation path and temporal correlation of abnormal parameters within the air conditioning system, combined with the equipment topology diagram, a decision map containing fault diagnosis and predictive maintenance suggestions is generated. This provides maintenance personnel with intuitive and comprehensive fault information, thereby improving the efficiency of fault diagnosis and repair. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0061] Figure 2 This is a schematic diagram of the system modules of the present invention;

[0062] Figure 3 This is a schematic diagram of the monitoring device of the present invention. Detailed Implementation

[0063] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0064] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0065] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in a preferred embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that mutually excludes other embodiments.

[0066] Please see Figure 1 As shown, the present invention provides an intelligent monitoring method for rail transit air conditioning systems, comprising:

[0067] S1. Real-time acquisition of multimodal operating data of the air conditioning system through a distributed sensor array, including temperature parameters, electrical parameters and mechanical status parameters;

[0068] In step S1, during the real-time data monitoring of the rail transit air conditioning system, multimodal operating data of the air conditioning system is first collected in real time using a distributed sensor array. This multimodal operating data covers multiple aspects, including temperature parameters, electrical parameters, and mechanical state parameters, ensuring comprehensive monitoring and providing information for subsequent data analysis and fault diagnosis. The distributed sensor array employs an adaptive sampling frequency when collecting multimodal operating data, dynamically adjusting the sampling interval based on the changing trends of the multimodal operating data. The specific steps are as follows:

[0069] Preset the initial sampling frequency for each sensor in the sensor array;

[0070] Establish a sliding time window and calculate the rate of change gradient of the multimodal running data in real time within each sliding time window;

[0071] Compare the rate of change gradient with a preset tolerance threshold;

[0072] When the rate of change gradient exceeds the tolerance threshold, the automatic adjustment mechanism of the sampling interval is triggered, and the sampling interval of the corresponding sensor is adjusted according to the adjustment result.

[0073] If, after the automatic adjustment mechanism for the sampling interval is triggered, the rate of change gradient still exceeds the tolerance threshold within N consecutive sliding time windows, the sampling frequency will be directly increased to the preset upper limit.

[0074] Specifically, when collecting multimodal operating data under the air conditioning system, an initial sampling frequency is first set for each sensor in the sensor array. This initial sampling frequency is preset based on the normal operating state of the air conditioning system and monitoring requirements. Then, a sliding time window is established. The size of the sliding time window can be set according to actual monitoring needs. This window is used to calculate the rate of change gradient of the multimodal operating data within each time period. The rate of change gradient reflects how quickly the data changes over a short period and is used in this embodiment as a basis for judging the trend of data change. The calculated rate of change gradient is then compared with a preset tolerance threshold. This tolerance threshold is set based on the normal operating fluctuation range of the air conditioning system and is used to determine whether the change in the multimodal operating data is within an acceptable range. When the rate of change gradient exceeds the tolerance threshold, it indicates that the corresponding data has an anomaly. When the rate of change fluctuates, an automatic sampling interval adjustment mechanism is triggered. This mechanism adjusts the sampling interval of the corresponding sensor based on the magnitude and duration of the rate of change gradient. If the rate of change gradient continuously exceeds the tolerance threshold, the sampling interval is gradually reduced to increase the data acquisition density and thus more accurately capture the data change trend. Conversely, if the rate of change gradient gradually decreases and returns to within the tolerance threshold, the sampling interval is gradually increased to reduce the data acquisition frequency and save system resources. After the automatic sampling interval adjustment mechanism is triggered, if the rate of change gradient still exceeds the tolerance threshold within N consecutive sliding time windows, the sampling frequency of the corresponding sensor will be directly increased to the preset upper limit. This ensures that data changes can be quickly captured in the event of abnormal data fluctuations, providing timely and accurate information for subsequent fault diagnosis and predictive maintenance.

[0075] Secondly, the automatic sampling interval adjustment mechanism includes the following steps:

[0076] Collect the difference between the rate of change gradient and the corresponding tolerance threshold, and record it as an adjustment condition parameter;

[0077] The sampling interval of the corresponding sensor is calculated based on the adjustment condition parameters, and when the adjustment amount of the sampling interval exceeds the adjustment threshold, the sampling frequency of the corresponding sensor is directly increased to the preset upper limit.

[0078] Send adjustment signals to similar sensors in the same area to synchronously adjust the sampling frequency of similar sensors;

[0079] The automatic sampling interval adjustment mechanism is activated automatically when abnormal data fluctuations are detected. It quantifies the degree of data fluctuation by collecting the difference between the rate of change gradient and the corresponding tolerance threshold, and records this as an adjustment condition parameter. Then, based on the adjustment condition parameter, a preset calculation function is used to calculate the adjusted sampling interval for the corresponding sensor. The expression for the calculation function is as follows: In the formula, Indicates the amount of adjustment to the sampling interval. Indicates the current sampling interval. Indicates the adjustment factor. This represents the difference between the gradient and the tolerance threshold. This represents the attenuation factor, primarily used to control the attenuation rate of the adjustment amplitude. The duration of abnormal fluctuations is indicated. It is important to clarify that if the adjustment amount of the sampling interval exceeds the preset adjustment threshold, it indicates that the corresponding data fluctuation is large, and more frequent sampling is required to capture the data changes. In this case, the sampling frequency of the corresponding sensor will be directly increased to the preset upper limit. In addition, in order to ensure the consistency and accuracy of monitoring, the automatic adjustment mechanism of the sampling interval will also send adjustment signals to similar sensors in the same area to synchronize their sampling frequencies. This can avoid the distortion of monitoring data caused by the inconsistency of sampling frequencies of different sensors, and provide more comprehensive information support for subsequent data analysis and fault diagnosis.

[0080] S2. Perform dimensionality reduction processing on the multimodal operation data to eliminate redundant information, extract key feature vectors, and establish a feature degradation trajectory map;

[0081] In step S2, after the multimodal operation data acquisition is completed, the acquired multimodal operation data will be subjected to dimensionality reduction processing to eliminate redundant information in the multimodal operation data. Simultaneously, key feature vectors will be extracted, and a feature degradation trajectory map will be established based on this, visually displaying the changing trend of the air conditioning system's state. The steps of performing dimensionality reduction processing on the multimodal operation data, eliminating redundant information, extracting key feature vectors, and establishing a feature degradation trajectory map include:

[0082] Multimodal operation data is segmented by using a sliding time window to generate data segments that include parameter correlations;

[0083] The data segments are processed using standard methods, and then a hybrid dimensionality reduction method combining principal component analysis and kernel density estimation is used to calculate the principal component scores and kernel density distribution of the data segments, and to screen key feature vectors.

[0084] The key feature vectors are time-series aligned, and then a feature degradation trajectory map is constructed. The feature degradation trend and key inflection points are marked in the degradation trajectory map.

[0085] Specifically, when extracting key feature vectors and establishing corresponding feature degradation trajectory maps, the multimodal data is first segmented using a sliding time window to generate data fragments containing parameter correlations. This data fragment generation helps in a more detailed analysis of data variation patterns. Then, the data fragments undergo standardization to unify the data's dimensions and distribution, improving the accuracy of subsequent dimensionality reduction. Based on this standardization, a hybrid dimensionality reduction approach using principal component analysis (PCA) and kernel density estimation (KD) is employed to reduce the dimensionality of the data fragments. PCA extracts the principal components, i.e., key feature vectors, while KD estimates the data's distribution characteristics. The dimensionality reduction is achieved by calculating the principal components of the data fragments. The kernel density distribution can be used to screen out key feature vectors that have a significant impact on the state changes of the air conditioning system. For example, key feature vectors can include abnormal fluctuations in temperature parameters, abrupt changes in electrical parameters, and degradation of mechanical state parameters, which can reflect potential problems in the operation of the air conditioning system. After extracting the key feature vectors, the selected key feature vectors are time-series aligned to ensure their consistency in time series. On this basis, a feature degradation trajectory map is constructed to visualize the changing trend of the air conditioning system state. The degradation trend and key inflection points are marked in the degradation trajectory map to identify the state change patterns and potential fault points of the air conditioning system.

[0086] S3. Construct a health status assessment baseline for the air conditioning system based on historical operating data, and introduce historical abnormal events to dynamically correct the health status assessment baseline, and output the dynamic assessment baseline.

[0087] In step S3, the health status assessment baseline serves as a reference for assessing the health status of the air conditioning system and is used to determine the current health status of the system. Furthermore, historical abnormal events are incorporated to dynamically correct the health status assessment baseline, ensuring its accuracy and real-time performance. The dynamic assessment baseline is then output to provide a basis for real-time monitoring. The step of constructing the air conditioning system's health status assessment baseline based on historical operating data includes:

[0088] Historical operating data is cleaned and standardized to remove abnormal operating condition data, forming a standardized training dataset containing temperature, electrical, and mechanical parameters.

[0089] Based on the train operation mode and real-time environmental parameters, the standardized training dataset is divided into data subsets under multiple operating conditions;

[0090] Cluster analysis was performed on each data subset to extract health status features under each working condition, and the health status features were time-series aligned to construct a health status assessment baseline.

[0091] Specifically, when constructing a health status assessment baseline, the historical operating data is first cleaned and standardized to remove abnormal operating condition data, ensuring the accuracy and consistency of the historical operating data. The processed historical operating data forms a standardized training dataset containing temperature, electrical, and mechanical parameters, providing a reliable foundation for subsequent analysis. Then, based on the train operating mode and real-time environmental parameters, the standardized training dataset is divided into data subsets under multiple operating conditions. This allows for analysis of the air conditioning system's operating status under different conditions. Afterward, cluster analysis is performed on each data subset. Cluster analysis groups similar data points into one category, thereby... Health status characteristics under various operating conditions are extracted. Cluster analysis can be performed using methods such as K-means clustering, hierarchical clustering, or density clustering, depending on the characteristics of the data and the analysis requirements. This can reflect the key parameters and changing trends of the air conditioning system under normal operating conditions. After extracting the health status characteristics, time-series alignment processing is also performed to ensure the consistency of the data over time, thereby constructing a corresponding health status assessment baseline. Specifically, a curve graph of the health status characteristics changing over time can be plotted, and the upper and lower limits of the health status assessment baseline can be determined based on the curve graph, which serves as a reference standard for assessing the health status of the air conditioning system.

[0092] Secondly, the steps for dynamically adjusting the health status assessment baseline by incorporating historical abnormal events and outputting the dynamic assessment baseline include:

[0093] Extract the abnormal feature vectors from historical abnormal events and match them with the feature vectors corresponding to the health status assessment baseline to output the feature deviation under the abnormal events;

[0094] A time decay function is constructed based on feature deviation, and the impact weights of historical anomalous events are dynamically allocated.

[0095] The dynamically allocated influence weights are integrated with the historical health status assessment baseline to correct the health status assessment baseline and generate a dynamic assessment baseline.

[0096] In this implementation, to improve the accuracy of the health status assessment baseline, historical anomalies are introduced for dynamic correction. First, anomaly feature vectors are extracted from these historical anomalies. These feature vectors represent the key parameter states of the air conditioning system when anomalies occurred in the past. Then, the anomaly feature vectors are matched with the feature vectors corresponding to the health status assessment baseline. The degree of deviation between the two is used to output the feature deviation under the anomaly event. The magnitude of the feature deviation directly reflects the degree of impact of the anomaly event on the health status of the air conditioning system. To determine the impact of historical anomalies on the health status assessment baseline, a time decay function is constructed based on the feature deviation to simulate the gradual weakening of the impact of the anomaly event over time. Based on this, the impact weights of historical anomalies are dynamically allocated. The expression for the time decay function is: In the formula, Indicates the impact weight of historical anomalies. Indicates the initial weights. Indicates the decay rate. Indicates the current timestamp. The timestamps of historical abnormal events are represented. Historical abnormal events that are closer to the current time will be assigned a greater impact weight, while historical events that are more distant will have a smaller impact and their corresponding impact weights will also be smaller. After obtaining the dynamically assigned impact weights, they will be fused with the historical health status assessment baseline. The fusion process is also a process of revising the health status assessment baseline. By comprehensively considering the impact of historical abnormal events, a dynamic assessment baseline that is closer to the actual health status of the current air conditioning system can be generated, thereby providing a more reliable reference for subsequent real-time monitoring and fault diagnosis.

[0097] S4. Compare and analyze the real-time operating data with the dynamic evaluation baseline to generate a comprehensive evaluation report that includes health level and fault warning information;

[0098] In step S4, during the real-time operation status monitoring of the equipment under the air conditioning system, the real-time operation data is compared and analyzed with the dynamic evaluation baseline. This allows for the timely detection of abnormalities in the operation of the air conditioning system, generating a comprehensive evaluation report containing health level and fault warning information to provide decision support for maintenance personnel. The step of comparing and analyzing real-time operation data with the dynamic evaluation baseline to generate a comprehensive evaluation report containing health level and fault warning information includes:

[0099] The real-time running data is standardized to generate a comparative feature vector with the same dimensions as the dynamic evaluation baseline;

[0100] The similarity between the comparative feature vector and the dynamic evaluation baseline is calculated, and the calculation result is output as a health score. The health score is then compared with the preset health evaluation threshold to identify the operating status.

[0101] If the health score is lower than the health assessment threshold, an abnormal warning mechanism is triggered. Based on the difference between the comparative feature vector and the dynamic assessment baseline, the health level is determined, and fault warning information is generated according to the health level, and a comprehensive assessment report is output.

[0102] If the health score is higher than the health assessment threshold, the system will maintain normal operation, record current operating data, and continuously monitor changes in health.

[0103] The health assessment threshold is a dynamic threshold, which is adaptively adjusted based on fluctuations in real-time operational data and the frequency of historical abnormal events. The specific adjustment process is as follows:

[0104] Construct a short-term sliding window and calculate the short-term volatility coefficient based on the fluctuation range of real-time running data within the short-term sliding window;

[0105] Construct a long-term sliding window, and count the frequency of historical abnormal events within the long-term sliding window. Calculate a frequency correction factor based on the frequency of historical abnormal events within the long-term sliding window.

[0106] By combining the short-term volatility coefficient and the frequency correction factor, and inputting them into the threshold adjustment function, a dynamically adjusted health assessment threshold is generated.

[0107] Specifically, when evaluating real-time operational data, the data is first standardized to ensure comparability and accuracy. Standardization includes unifying data dimensions, removing outliers, and smoothing the data, generating a comparative feature vector with the same dimensions as the dynamic evaluation baseline. This feature vector comprehensively reflects the real-time operating status of the air conditioning system, providing a foundation for subsequent similarity calculations. Next, the comparative feature vector is compared with the dynamic evaluation baseline for similarity calculation. Methods such as cosine similarity, Euclidean distance, or Manhattan distance can be used, the specific choice depending on the data characteristics and analytical needs. The result quantifies the degree of difference between the real-time operational data and the dynamic evaluation baseline, and is output as a health score. The health score directly reflects the health status of the air conditioning system. Subsequently, the health score is compared with a preset health assessment threshold to identify the operating status of the air conditioning system. This health assessment threshold is set based on the normal operating status and historical abnormal events of the air conditioning system, used to distinguish between normal and abnormal states. When the health score is lower than the health assessment threshold, it indicates that the air conditioning system is operating abnormally. This triggers an anomaly warning mechanism, which responds quickly and sends warning messages to maintenance personnel to remind them to pay attention to the system's operating status. Simultaneously, it determines the health level based on the difference between the comparative feature vector and the dynamic assessment baseline. The health level reflects the severity of the anomaly, and different health levels correspond to different maintenance strategies. Fault warning information is generated based on the health level, including a detailed description of the anomaly, possible causes, and suggested maintenance measures. This information is integrated into a comprehensive assessment report and output to the management system for feedback to maintenance personnel, enabling them to understand the system's health status and take appropriate maintenance measures to prevent the occurrence or escalation of faults. If the health score is higher than the health assessment threshold, it indicates that the air conditioning system is operating normally. In this case, it maintains normal operation, records current operating data for subsequent analysis and evaluation, and continuously monitors health changes to ensure stable operation of the air conditioning system.

[0108] It should be noted that when determining the health assessment threshold, the fluctuation of real-time operating data and the frequency of historical abnormal events are comprehensively considered to improve the accuracy and adaptability of the health assessment. Specifically, a short-term sliding window is constructed, and a short-term fluctuation coefficient is calculated based on the fluctuation range of real-time operating data within the short-term sliding window. The short-term fluctuation coefficient reflects the stability of the air conditioning system's short-term operating status. When the real-time operating data fluctuates significantly, the short-term fluctuation coefficient will also increase accordingly, indicating that the air conditioning system may be in an unstable state. Simultaneously, a long-term sliding window is constructed, and the frequency of historical abnormal events within the long-term sliding window is statistically analyzed. A frequency correction factor is calculated based on the frequency of historical abnormal events within the long-term sliding window. The frequency correction factor reflects the abnormal conditions during the long-term operation of the air conditioning system. When the frequency of historical abnormal events is high, the frequency correction factor will also increase accordingly, indicating that the air conditioning system has potential failure risks. Finally, the short-term fluctuation coefficient and the frequency correction factor are combined and input into the threshold adjustment function to generate a dynamically adjusted health assessment threshold. The expression for the threshold adjustment function is as follows: In the formula, This indicates the dynamically adjusted health assessment threshold. This indicates the initial health assessment threshold. and These represent the weighting coefficients of the short-term volatility coefficient and the frequency correction factor, respectively. It represents the short-term volatility coefficient (the ratio of the standard deviation of real-time operating data within a short-term sliding window to the standard deviation of the corresponding dynamic assessment baseline). The frequency correction factor (the ratio between the number of anomalies within a long-term sliding window and the total number of samples) represents the health assessment threshold. The dynamically adjusted threshold can more accurately reflect the current health status of the air conditioning system, providing a more reliable basis for subsequent real-time monitoring and fault early warning.

[0109] S5. Obtain the propagation path and temporal correlation of abnormal parameters in the air conditioning system, and generate a decision map containing fault diagnosis and predictive maintenance suggestions by combining the equipment topology diagram in the air conditioning system.

[0110] In step S5, when abnormal parameters are found during the evaluation of the air conditioning system, the propagation path and temporal correlation of the abnormal parameters within the air conditioning system are first determined. Then, combined with the equipment topology diagram within the air conditioning system, a decision map containing fault diagnosis and predictive maintenance suggestions can be generated. This provides maintenance personnel with targeted maintenance strategies to ensure the stable operation of the air conditioning system. The steps of obtaining the propagation path and temporal correlation of abnormal parameters within the air conditioning system, and generating a decision map containing fault diagnosis and predictive maintenance suggestions by combining it with the equipment topology diagram within the air conditioning system, include:

[0111] Based on the spatial distribution of characteristic deviation, analyze the propagation path of abnormal parameters between devices and determine the temporal correlation.

[0112] By comparing the time series differences of multi-source sensor data, the propagation delay of abnormal events between different devices can be identified, forming a time-stamped fault propagation chain;

[0113] By combining the connection tightness of the equipment topology diagram, the key equipment and potential impact range related to abnormal parameters are identified, and the key equipment is marked as a potential fault source to generate fault diagnosis results.

[0114] Based on the time markers of the fault propagation chain, the failure development trend of equipment within the potential impact range is predicted, preventive maintenance strategies are formulated, and a decision map is formed.

[0115] Specifically, after the abnormal parameters are output, the propagation path of the abnormal parameters among various devices in the air conditioning system and their temporal correlation are analyzed based on the spatial distribution of the characteristic deviation. Then, the time series differences of multi-source sensor data are compared, mainly to identify the possible delays in the propagation of abnormal events among different devices. Based on this, a time-stamped fault propagation chain can be constructed to determine the development trajectory of the fault in the air conditioning system. Then, combined with the device topology diagram within the air conditioning system, the key devices and potential impact range related to the abnormal parameters are further identified. During this process, key devices are marked as potential fault sources, and corresponding fault diagnosis results are generated. Finally, based on the time stamp of the fault propagation chain, the fault development trend of devices within the potential impact range is predicted, thereby enabling the formulation of targeted preventive maintenance strategies. This forms a decision map that includes fault diagnosis and predictive maintenance suggestions, providing maintenance personnel with intuitive maintenance guidance and ensuring the stable operation of the air conditioning system.

[0116] Please see Figure 2 A smart monitoring system for rail transit air conditioning systems, using the aforementioned smart monitoring method for rail transit air conditioning systems, includes:

[0117] The parameter acquisition module is used to collect multimodal operating data of the air conditioning system in real time through a distributed sensor array, including temperature parameters, electrical parameters and mechanical status parameters;

[0118] The parameter preprocessing module is used to reduce the dimensionality of multimodal operating data, eliminate redundant information, extract key feature vectors, and establish a feature degradation trajectory map.

[0119] The baseline optimization module is used to construct a health status assessment baseline for the air conditioning system based on historical operating data, and to dynamically correct the health status assessment baseline by incorporating historical abnormal events, and output a dynamic assessment baseline.

[0120] The evaluation module is used to compare and analyze real-time operating data with dynamic evaluation baselines to generate a comprehensive evaluation report that includes health level and fault warning information.

[0121] The fault diagnosis module is used to obtain the propagation path and temporal correlation of abnormal parameters in the air conditioning system, and generate a decision map containing fault diagnosis and predictive maintenance suggestions by combining the equipment topology diagram in the air conditioning system.

[0122] In the above, the parameter acquisition module is responsible for acquiring various operational data of the air conditioning system in real time, specifically through a sensor array distributed throughout the air conditioning system, ensuring the comprehensiveness and accuracy of the data. The parameter preprocessing module processes this raw data, removing redundant information through dimensionality reduction techniques and extracting feature vectors that are crucial for assessing the health status of the air conditioning system. Based on these vectors, a feature degradation trajectory map is established. This feature degradation trajectory map can intuitively display the performance change trend of the air conditioning system over time, providing a corresponding reference for subsequent health status assessment and fault diagnosis. The baseline optimization module constructs an initial health status assessment baseline based on historical operational data, reflecting the key parameter range of the air conditioning system under normal operating conditions. However, considering that the operating status of the air conditioning system may change due to various factors, historical abnormal events are introduced to influence the health status assessment. The baseline is dynamically corrected by comprehensively considering the impact of historical anomalies to generate a dynamic assessment baseline that more closely reflects the current health status of the air conditioning system. This provides a more accurate reference for subsequent real-time monitoring and fault diagnosis. The assessment module is responsible for comparing and analyzing real-time operating data with the dynamic assessment baseline. By comparing the differences between the real-time operating data and the dynamic assessment baseline, abnormal situations in the operation of the air conditioning system can be detected in a timely manner, and a comprehensive assessment report containing health level and fault warning information can be generated to comprehensively reflect the current health status of the air conditioning system and provide decision support for maintenance personnel. The fault diagnosis module integrates the propagation path and temporal correlation of abnormal parameters with the equipment topology diagram within the air conditioning system to generate a decision map containing fault diagnosis and predictive maintenance suggestions. This provides maintenance personnel with targeted maintenance strategies to ensure the stable operation of the air conditioning system.

[0123] Please see Figure 3 An intelligent monitoring device for rail transit air conditioning systems, the monitoring device comprising:

[0124] At least one processor;

[0125] and memory that is communicatively connected to at least one processor;

[0126] The memory stores a computer program that can be executed by at least one processor, and the computer program is executed by at least one processor so that at least one processor can execute the above-mentioned intelligent monitoring method for rail transit air conditioning systems.

[0127] The processor of the aforementioned monitoring device can be a high-performance central processing unit (CPU) or graphics processing unit (GPU), and the memory can include storage devices such as random access memory (RAM), read-only memory (ROM), solid-state drive (SSD), or hard disk drive. In addition, the monitoring device may also include an arithmetic unit, input devices, output devices, and a network interface. The arithmetic unit can be a logic unit used to perform various arithmetic and logical operations to assist the processor in completing complex data processing tasks. Input devices can include keyboards, mice, touch screens, etc., used to receive user input instructions and data. Output devices can include displays, printers, etc., used to display processing results and output reports. The network interface is used to enable network communication between the monitoring device and other systems or devices for data exchange and remote monitoring.

[0128] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0129] The above description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained in this invention are implemented according to conventional methods in the art unless otherwise specified or limited.

Claims

1. A method for intelligent monitoring of a rail transit air conditioning system, characterized in that: The application relates to a health state evaluation method for an air conditioning system. Real-time multi-modal operation data of the air conditioning system is collected by a distributed sensor array, including temperature parameters, electrical parameters and mechanical state parameters. The multi-modal operation data is processed by dimension reduction to eliminate redundant information and extract key feature vectors, and a feature degradation trajectory atlas is established. A health state evaluation baseline of the air conditioning system is constructed according to historical operation data, and the health state evaluation baseline is dynamically modified by introducing historical abnormal events, and a dynamic evaluation baseline is output. Real-time operation data is compared with the dynamic evaluation baseline to generate a comprehensive evaluation report containing health grades and fault warning information. The propagation path and time sequence correlation of abnormal parameters in the air conditioning system are obtained, and a decision atlas containing fault diagnosis and predictive maintenance suggestions is generated in combination with a device topology graph in the air conditioning system. The health state evaluation baseline of the air conditioning system is constructed according to historical operation data, including the following steps: The historical operation data is cleaned and standardized to eliminate abnormal working condition data, and a standardized training data set containing temperature, electrical and mechanical parameters is formed. The standardized training data set is divided into data subsets under different working conditions according to train operation modes and real-time environmental parameters. Cluster analysis is performed on each data subset to extract health state features under different working conditions, and the health state features are time-aligned to construct a health state evaluation baseline. The health state evaluation baseline is dynamically modified by introducing historical abnormal events, and a dynamic evaluation baseline is output, including the following steps: Abnormal feature vectors in the historical abnormal events are extracted and matched with corresponding feature vectors of the health state evaluation baseline to output feature deviation degrees under abnormal events. A time decay function is constructed based on the feature deviation degrees, and the influence weights of the historical abnormal events are dynamically allocated. The dynamically allocated influence weights are fused with the historical health state evaluation baseline to modify the health state evaluation baseline and generate a dynamic evaluation baseline. 2.The intelligent monitoring method of the rail transit air conditioning system according to claim 1, characterized in that: The distributed sensor array adopts an adaptive sampling frequency when collecting multi-modal operation data, dynamically adjusts the sampling interval according to the change trend of the multi-modal operation data, and the specific steps are as follows: An initial sampling frequency is preset for each sensor in the sensor array. A sliding time window is established, and the change rate gradient of the multi-modal operation data in each sliding time window is calculated in real time. The change rate gradient is compared with the corresponding tolerance threshold. When the change rate gradient exceeds the tolerance threshold, the sampling interval automatic adjustment mechanism is triggered, and the sampling interval of the corresponding sensor is adjusted according to the adjustment result. When the sampling interval automatic adjustment mechanism is triggered, if the change rate gradient in the next N sliding time windows still exceeds the tolerance threshold, the sampling frequency is directly raised to the preset upper limit. 3.The intelligent monitoring method of the rail transit air conditioning system according to claim 2, characterized in that: When the sampling interval automatic adjustment mechanism is executed, the following steps are included: The difference between the change rate gradient and the corresponding tolerance threshold is collected and recorded as an adjustment condition parameter. The adjusted sampling interval of the corresponding sensor is calculated according to the adjustment condition parameter, and when the sampling interval adjustment amount exceeds the adjustment threshold, the sampling frequency of the corresponding sensor is directly raised to the preset upper limit. Adjustment signals are sent to similar sensors in the same area to synchronously adjust the sampling frequency of similar sensors. 4.The intelligent monitoring method of the rail transit air conditioning system according to claim 1, characterized in that: The steps of performing dimensionality reduction processing on multimodal operational data, eliminating redundant information, extracting key feature vectors, and establishing a feature degradation trajectory map include: Multimodal operation data is segmented by using a sliding time window to generate data segments that include parameter correlations; The data segments are processed using standard methods, and then a hybrid dimensionality reduction method combining principal component analysis and kernel density estimation is used to calculate the principal component scores and kernel density distribution of the data segments, and to screen key feature vectors. The key feature vectors are time-aligned, and then a feature degradation trajectory map is constructed. The feature degradation trend and key inflection points are marked in the degradation trajectory map. 5.The intelligent monitoring method of the rail transit air conditioning system according to claim 1, characterized in that: The step of comparing and analyzing real-time operating data with a dynamic evaluation baseline to generate a comprehensive evaluation report containing health level and fault early warning information includes: The real-time running data is standardized to generate a comparative feature vector with the same dimensions as the dynamic evaluation baseline; The similarity between the comparative feature vector and the dynamic evaluation baseline is calculated, and the calculation result is output as a health score. The health score is then compared with the preset health evaluation threshold to identify the operating status. If the health score is lower than the health assessment threshold, an abnormal warning mechanism is triggered. Based on the difference between the comparative feature vector and the dynamic assessment baseline, the health level is determined, and fault warning information is generated according to the health level, and a comprehensive assessment report is output. If the health score is higher than the health assessment threshold, the system will maintain normal operation, record current operating data, and continuously monitor changes in health. The health assessment threshold is a dynamic threshold, which is adaptively adjusted based on fluctuations in real-time operational data and the frequency of historical abnormal events. The specific adjustment process is as follows: Construct a short-term sliding window and calculate the short-term volatility coefficient based on the fluctuation range of real-time running data within the short-term sliding window; Construct a long-term sliding window, and count the frequency of historical abnormal events within the long-term sliding window. Calculate a frequency correction factor based on the frequency of historical abnormal events within the long-term sliding window. By combining the short-term volatility coefficient and the frequency correction factor, the data is input into the threshold adjustment function to generate a dynamically adjusted health assessment threshold. 6.The intelligent monitoring method of the rail transit air conditioning system according to claim 1, characterized in that: The step of obtaining the propagation path and temporal correlation of abnormal parameters within the air conditioning system, and generating a decision map containing fault diagnosis and predictive maintenance suggestions by combining it with the equipment topology diagram within the air conditioning system, includes: Based on the spatial distribution of characteristic deviation, analyze the propagation path of abnormal parameters between devices and determine the temporal correlation. By comparing the time series differences of multi-source sensor data, the propagation delay of abnormal events between different devices can be identified, forming a time-stamped fault propagation chain; By combining the connection tightness of the equipment topology diagram, the key equipment and potential impact range related to abnormal parameters are identified, and the key equipment is marked as a potential fault source to generate fault diagnosis results. Based on the time markers of the fault propagation chain, the failure development trend of equipment within the potential impact range is predicted, preventive maintenance strategies are formulated, and a decision map is formed.

7. An intelligent monitoring system for a rail transit air conditioning system, characterized in that: The intelligent monitoring method for rail transit air conditioning systems according to any one of claims 1 to 6 includes: The parameter acquisition module is used to collect multimodal operating data of the air conditioning system in real time through a distributed sensor array, including temperature parameters, electrical parameters and mechanical status parameters; The parameter preprocessing module is used to reduce the dimensionality of multimodal operating data, eliminate redundant information, extract key feature vectors, and establish a feature degradation trajectory map. The baseline optimization module is used to construct a health status assessment baseline for the air conditioning system based on historical operating data, and to dynamically correct the health status assessment baseline by incorporating historical abnormal events, and output a dynamic assessment baseline. The evaluation module is used to compare and analyze real-time operating data with dynamic evaluation baselines to generate a comprehensive evaluation report that includes health level and fault warning information. The fault diagnosis module is used to obtain the propagation path and temporal correlation of abnormal parameters in the air conditioning system, and generate a decision map containing fault diagnosis and predictive maintenance suggestions by combining the equipment topology diagram in the air conditioning system.

8. An intelligent monitoring device for a rail transit air conditioning system, characterized in that: The monitoring device includes: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the intelligent monitoring method for rail transit air conditioning system according to any one of claims 1 to 6.

Images