Rail transit hub equipment fault intelligent diagnosis and early warning method

Abstract

The application discloses a kind of rail transit hub equipment fault intelligent diagnosis and early warning method, it is related to rail transit equipment fault diagnosis and predictive maintenance technical field, including acquisition multimodal data, and based on equipment topological relationship and time stamp, space-time alignment is carried out to multimodal data, generates multidimensional fault feature data;Multidimensional fault feature data is input parallel feature extraction module, and generates fault feature vector;Fault feature vector is input fault diagnosis model, and preliminary classification is carried out based on ResNet-18 network main structure, and the corresponding fault type and confidence are output;According to the remaining useful life of equipment calculated according to fault development trend prediction result, generate early warning information, and combine the historical early warning statistical data of confidence and equipment real-time load rate, dynamically adjust confidence, realize the hierarchical management of early warning notice.The application not only enhances the accuracy of fault prediction, but also realizes the intelligentization and refinement of operation and maintenance strategy.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis and predictive maintenance technology for rail transit equipment, and in particular to a method for intelligent fault diagnosis and early warning of rail transit hub equipment. Background Technology

[0002] Currently, urban heating systems commonly use heat exchange stations as the heat exchange link between the heat source and the user end, achieving efficient heat energy transmission and distribution. Due to the wide distribution and complex operating environment of heat exchange stations, traditional manual monitoring methods are not only costly in terms of manpower, but also suffer from slow response times in emergencies and poor remote monitoring capabilities, easily leading to problems such as unstable heating quality, high energy consumption, and delayed equipment maintenance. With the development of smart cities and automated control technologies, the intelligent transformation of heat exchange stations has become an urgent need, especially in terms of how to achieve remote monitoring, automatic adjustment, and fault alarms, where key technological bottlenecks still exist.

[0003] CN103438503A discloses an intelligent control method and control system for unattended heat exchange stations. This system uses a programmable logic controller (PLC) as the core control unit. The PLC's input terminals are connected to an outdoor temperature compensator, temperature and pressure sensors at the heat exchanger inlet and outlet, and a level sensor in the makeup water tank. Its output terminals are connected to actuators such as electric regulating valves, solenoid valves, and frequency converters, enabling variable frequency control of the circulating pump and makeup water pump. This allows for automatic constant temperature and pressure heating, water replenishment, and power outage restart operations. It also supports real-time transmission of field data to a remote control terminal, providing users with the ability to remotely access real-time data. However, this technology primarily addresses the automatic adjustment and remote monitoring of heating parameters. It does not yet address predictive maintenance of equipment operating status, energy efficiency analysis, or multi-point collaborative control, making it difficult to adapt to the ever-changing urban heating demands.

[0004] CN102985063A discloses a remote monitoring method for thermal systems based on configuration software. By constructing a SCADA platform, it centrally collects and graphically displays data from multiple heat exchange stations, enabling centralized monitoring of operational status and fault alerts. However, this method has high requirements for the real-time performance and stability of data transmission. Unstable network conditions or data loss may lead to misjudgments and delayed responses. Furthermore, it lacks the autonomous capability of local control logic, still relying on manual intervention in practical applications to handle faults and formulate adjustment strategies. Therefore, this method cannot achieve fully automated intelligent operation and early warning decision-making for complex thermal systems. Summary of the Invention

[0005] In view of the problems of insufficient multi-source data fusion, high false alarm rate of fault warning and lack of dynamic adjustment mechanism of operation and maintenance strategy in existing equipment monitoring methods, this invention is proposed.

[0006] Therefore, the problem to be solved by this invention is how to achieve accurate alignment and intelligent analysis of multimodal data of rail transit hub equipment, improve the accuracy of fault diagnosis, and establish an adaptive hierarchical early warning mechanism, thereby optimizing the allocation efficiency of operation and maintenance resources and the reliability management of equipment.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] In a first aspect, embodiments of the present invention provide a method for intelligent diagnosis and early warning of faults in rail transit hub equipment, comprising,

[0009] Multimodal data of various equipment in the rail transit hub are collected, and the multimodal data is spatiotemporally aligned based on the equipment topology and timestamps to generate multidimensional fault feature data.

[0010] The multidimensional fault feature data is input into the parallel feature extraction module to generate a fault feature vector.

[0011] The fault feature vector is input into the fault diagnosis model, and preliminary classification is performed based on the ResNet-18 network backbone structure, outputting the corresponding fault type and confidence level.

[0012] The remaining service life of the equipment is calculated based on the fault type and confidence level, and early warning information is generated. The confidence level is dynamically adjusted by combining historical early warning statistics and real-time load rate of the equipment to achieve hierarchical management of early warning notifications.

[0013] As a preferred embodiment of the intelligent fault diagnosis and early warning method for rail transit hub equipment described in this invention, the remaining service life of the equipment is calculated based on the fault type and confidence level, and early warning information is generated, including:

[0014] The fault type and confidence level are matched with typical fault cases in the historical fault database to construct a fault similarity score matrix, extract the degradation curve template of the corresponding fault type, and calculate the remaining service life.

[0015] Based on the reliability of typical failure cases, a dynamic early warning classification rule is established to classify early warning information into different levels and collect the real-time load rate of the equipment.

[0016] The remaining service life is adjusted based on the real-time load rate. When the real-time load rate is higher than the rated load rate, the remaining service life is shortened according to the load ratio.

[0017] The accuracy of the statistical early warning levels was calculated, and an early warning confidence correction coefficient was constructed.

[0018] The warning confidence correction coefficient is multiplied by the current confidence level to obtain the corrected confidence level value, and the warning information is then pushed out in a tiered manner.

[0019] As a preferred embodiment of the intelligent fault diagnosis and early warning method for rail transit hub equipment described in this invention, the method involves matching fault types and confidence levels with typical fault cases in a historical fault database, including:

[0020] When the confidence level of the fault type is higher than the first confidence threshold and the historical case matching degree is greater than the first matching threshold, the fault case is marked as a first-level reliable matching case.

[0021] When the confidence level of the fault type is between the second confidence threshold and the first confidence threshold, and the historical case matching degree is between the second matching threshold and the first matching threshold, the fault case is marked as a second-level reliable matching case.

[0022] When the confidence level of the fault type is less than the second confidence level threshold or the historical case matching degree is less than the second matching degree threshold, the fault case is marked as a level 3 trusted matching case.

[0023] As a preferred embodiment of the intelligent fault diagnosis and early warning method for rail transit hub equipment described in this invention, it further includes:

[0024] For the first-level trusted matching case: if the remaining lifetime is less than the first time threshold and the device load rate is greater than the first load threshold, a first-level warning is triggered; if the remaining lifetime is between the first time threshold and the second time threshold, and the device load rate is between the second load threshold and the first load threshold, a second-level warning is triggered; if the remaining lifetime is between the second time threshold and the third time threshold, and the device load rate is less than the second load threshold, a third-level warning is triggered.

[0025] For the aforementioned Level 2 trusted matching case: if the remaining lifetime is less than the fourth time threshold and the device load rate is greater than the third load threshold, a Level 1 warning is triggered; if the remaining lifetime is between the fourth and fifth time thresholds and the device load rate is between the fourth and third load thresholds, a Level 2 warning is triggered; if the remaining lifetime is between the fifth and sixth time thresholds and the device load rate is less than the fourth load threshold, a Level 3 warning is triggered.

[0026] For the aforementioned three-level trusted matching case: if the remaining lifetime is less than the seventh time threshold and the device load rate is greater than the fifth load threshold, a first-level warning is triggered; if the remaining lifetime is between the seventh and eighth time thresholds and the device load rate is between the sixth and fifth load thresholds, a second-level warning is triggered; if the remaining lifetime is between the eighth and ninth time thresholds and the device load rate is less than the sixth load threshold, a third-level warning is triggered.

[0027] As a preferred embodiment of the intelligent fault diagnosis and early warning method for rail transit hub equipment described in this invention, the method for obtaining the fault type and confidence level is as follows:

[0028] A fault diagnosis model based on the ResNet-18 architecture is constructed, wherein the fault diagnosis model includes a feature extraction layer, a domain discrimination layer, and a fault classification layer;

[0029] The fault feature vector is input into the feature extraction layer to extract a high-dimensional feature representation, wherein the high-dimensional feature representation includes deep semantic information of the device status;

[0030] In the domain discrimination layer, the maximum average difference metric between the source domain features and the target domain features is calculated, and the maximum average difference metric is used as the optimization objective of the domain adaptation loss function.

[0031] The parameters of the feature extraction layer are trained using labeled samples of rail transit hub equipment in the target domain, combined with the domain adaptation loss function, so that the features have domain invariance.

[0032] The features trained by domain adaptation are input into the fault classification layer to calculate the probability distribution of each fault type, and output the fault type with the highest probability and the confidence value.

[0033] As a preferred embodiment of the intelligent fault diagnosis and early warning method for rail transit hub equipment described in this invention, the method for obtaining the fault feature vector is as follows:

[0034] Multidimensional fault characteristic data are categorized into vibration characteristic channel, temperature characteristic channel, and current characteristic channel according to data type.

[0035] Wavelet packet decomposition is performed on the data of the vibration feature channel to extract the energy features, kurtosis features and skewness features of each frequency band, forming vibration feature sub-vectors;

[0036] Statistical analysis is performed on the data of the temperature feature channels to calculate the temperature mean, standard deviation, and rate of change, forming a temperature feature sub-vector;

[0037] Fourier transform is performed on the data of the current characteristic channel to extract the current harmonic ratio, phase difference and waveform distortion rate, forming a current characteristic sub-vector;

[0038] The vibration feature vector, the temperature feature vector, and the current feature vector are fused together using feature importance weights to generate a fault feature vector.

[0039] As a preferred embodiment of the intelligent fault diagnosis and early warning method for rail transit hub equipment described in this invention, the method for obtaining the multi-dimensional fault feature data is as follows:

[0040] Collect multimodal data of various equipment within the rail transit hub, including vibration sensor data and equipment operating condition data;

[0041] The multimodal data is labeled with device numbers according to the collection location, and a spatial association matrix between devices is established based on the topology diagram of rail transit hub equipment;

[0042] The multimodal data is timestamped and aligned, and data with different sampling frequencies are unified to the same time scale by linear interpolation to generate a multimodal data sequence.

[0043] Based on the spatial correlation matrix, spatial correlation analysis is performed on the data of adjacent devices in the multimodal data sequence to calculate the fault propagation coefficient between devices;

[0044] The multimodal data sequence and the fault propagation coefficient are combined to form multidimensional fault feature data.

[0045] Secondly, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any step of the above-described intelligent diagnosis and early warning method for faults in rail transit hub equipment.

[0046] Thirdly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the above-described intelligent diagnosis and early warning method for faults in rail transit hub equipment.

[0047] Compared with existing technologies, the advantages of this invention are as follows: Through multimodal data acquisition and spatiotemporal alignment technology, it achieves full-dimensional monitoring of equipment status, effectively solving the challenges of heterogeneous sensor data synchronization and cross-device fault correlation analysis; based on a parallel feature extraction module, it integrates multi-source features such as vibration, temperature, and current, significantly improving the ability to identify complex faults; combined with a dynamic confidence correction mechanism, it adaptively adjusts the warning threshold based on historical warning accuracy and real-time load rate, greatly reducing the false alarm rate and optimizing the warning timeliness; through a hierarchical push strategy, it implements differentiated warning responses based on cases matching different confidence levels, ensuring priority processing of high-reliability warnings while rationally allocating maintenance resources; this method not only enhances the accuracy of fault prediction but also realizes the intelligence and refinement of maintenance strategies, providing an efficient and reliable solution for the health management of rail transit hub equipment. Attached Figure Description

[0048] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0049] Figure 1 A flowchart for intelligent diagnosis and early warning methods for equipment faults in rail transit hubs. Detailed Implementation

[0050] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0051] 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.

[0052] Secondly, the term "one 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 one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0053] This invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not adhering to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0054] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0055] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0056] Example 1

[0057] Reference Figure 1 This is the first embodiment of the present invention, which provides a method for intelligent diagnosis and early warning of faults in rail transit hub equipment, including:

[0058] S1: Collect multimodal data of various equipment in the rail transit hub, and perform spatiotemporal alignment of the multimodal data based on equipment topology and timestamps to generate multidimensional fault feature data.

[0059] S1.1: Collect multimodal data of various equipment in the rail transit hub, including vibration sensor data and equipment operating condition data.

[0060] It should be noted that the data includes multimodal vibration sensor data and equipment operating condition data; the vibration sensor data includes the equipment's X-axis vibration acceleration, Y-axis vibration acceleration, and Z-axis vibration acceleration; the equipment operating condition data includes bearing temperature data, motor current data, and speed data.

[0061] S1.2: Label the multimodal data with device numbers according to the collection location, and establish a spatial correlation matrix between devices based on the topology diagram of rail transit hub equipment.

[0062] It should be noted that the element values ​​in the spatial correlation matrix represent the physical distance between devices.

[0063] S1.3: Timestamp alignment of multimodal data is performed, and data with different sampling frequencies are unified to the same time scale through linear interpolation to generate multimodal data sequences.

[0064] S1.4: Based on the spatial correlation matrix, perform spatial correlation analysis on the data of adjacent devices in the multimodal data sequence and calculate the fault propagation coefficient between devices.

[0065] S1.5: Combine multimodal data sequences and fault propagation coefficients to form multidimensional fault feature data.

[0066] It should be noted that the multidimensional fault feature data includes the equipment's own state characteristics and the fault correlation characteristics between equipment.

[0067] S2: Input the multidimensional fault feature data into the parallel feature extraction module to generate fault feature vectors.

[0068] S2.1: The multidimensional fault characteristic data is divided into vibration characteristic channel, temperature characteristic channel and current characteristic channel according to the data type.

[0069] It should be noted that the vibration characteristic channel includes vibration data in the X, Y, and Z axes; the temperature characteristic channel includes bearing temperature data; and the current characteristic channel includes motor current data.

[0070] S2.2: Perform wavelet packet decomposition on the vibration feature channel data to extract the energy features, kurtosis features and skewness features of each frequency band, forming vibration feature sub-vectors.

[0071] S2.3: Perform statistical analysis on the data of the temperature feature channel, calculate the temperature mean, standard deviation and rate of change, and form a temperature feature sub-vector.

[0072] S2.4: Perform Fourier transform on the data of the current characteristic channel to extract the current harmonic ratio, phase difference and waveform distortion rate, and form the current characteristic sub-vector.

[0073] S2.5: The vibration feature vector, temperature feature vector, and current feature vector are fused by feature importance weights to generate a fault feature vector.

[0074] It should be noted that the feature importance weights are obtained through training on historical fault samples.

[0075] S3: Input the fault feature vector into the fault diagnosis model, perform preliminary classification based on the ResNet-18 network backbone structure, and output the corresponding fault type and confidence level.

[0076] S3.1: Construct a fault diagnosis model based on the ResNet-18 structure, which includes a feature extraction layer, a domain discrimination layer, and a fault classification layer.

[0077] It should be noted that the feature extraction layer is pre-trained using historical data from the source domain devices.

[0078] In an optional implementation, the establishment of the fault diagnosis model includes: collecting operating data of source domain devices, labeling the operating data according to fault type, and constructing a source domain training dataset; extracting time domain features, frequency domain features, and time-frequency features from the source domain training dataset to generate a feature sample matrix.

[0079] In an optional implementation, the fault diagnosis model is trained in stages: the first stage: the feature extraction layer and the fault classification layer are pre-trained using the source domain training dataset; the second stage: the parameters of the feature extraction layer are frozen, and the domain discrimination layer is trained separately; the third stage: the feature extraction layer, the domain discrimination layer and the fault classification layer are jointly optimized.

[0080] It should be noted that after each training phase, the classification accuracy on the validation set is calculated. When the improvement in classification accuracy reaches the maximum training rounds, the next training phase begins.

[0081] Specifically, the feature extraction layer contains four residual blocks, each containing two convolutional layer combination units and one skip connection path; the domain discrimination layer adopts a three-layer fully connected network structure; and the fault classification layer uses a softmax classifier.

[0082] Furthermore, the fault diagnosis model also includes convolutional layer combination units and skip connection paths; the convolutional layer combination unit is composed of two 3×3 convolutional layers connected in series, and each convolutional layer is followed by a batch normalization layer and a ReLU activation function; the skip connection path uses a 1×1 convolutional layer to perform dimensionality reduction transformation on the input features.

[0083] S3.2: Input the fault feature vector into the feature extraction layer to extract high-dimensional feature representations, where the high-dimensional feature representations include deep semantic information about the equipment status.

[0084] S3.3: Calculate the maximum average difference metric between source domain features and target domain features in the domain discrimination layer, and use the maximum average difference metric as the optimization objective of the domain adaptation loss function.

[0085] The preferred formula for the maximum average difference measure is as follows:

[0086]

[0087] Among them, D max X is the maximum average difference measure, where N is the total number of samples, T is the signal sampling period, and X is the maximum average difference measure. i (t) represents the time-domain signal value of the i-th sample at time t, X j (t) represents the time-domain signal value of the j-th sample at time t, K is the total number of frequency domain feature dimensions, and F i (m) represents the amplitude of the m-th frequency component of the i-th sample, F j (m) represents the amplitude of the m-th frequency component of the j-th sample, and α is the weighting coefficient for balancing the time-domain and frequency-domain features.

[0088] S3.4: Using labeled samples of rail transit hub equipment in the target domain, the parameters of the feature extraction layer are trained by combining the domain adaptation loss function to make the features domain invariant.

[0089] S3.5: Input the features trained by domain adaptation into the fault classification layer, calculate the probability distribution of each fault type, and output the fault type with the highest probability and the confidence value.

[0090] The preferred formulas for the fault type with the highest probability and the confidence level are as follows:

[0091]

[0092]

[0093] Where C is the confidence level value, γ(d i ) is based on sample distance d i The exponential decay function, where S is the similarity score, σ 2 The variance of the current feature set. The maximum permissible variance threshold is δ, where δ is the confidence correction coefficient, and P(F) is the maximum permissible variance threshold. k |X) represents the fault type F under given characteristic X. k The posterior probability of occurrence, where M is the total number of features, w i Let be the weight coefficient of the i-th feature. Let λ be the mapping value of the i-th feature extraction function to a given feature X. k Let X be the feature vector of the k-th type of fault, B be the total number of fault types, H(X) be the information entropy of the current feature X, and N be the total number of historical samples. n Let be the feature value of the nth historical sample.

[0094] S4: Calculate the remaining service life of the equipment based on the fault type and confidence level, generate early warning information, and dynamically adjust the confidence level by combining historical early warning statistics and real-time load rate of the equipment to achieve hierarchical management of early warning notifications.

[0095] S4.1: Match the fault type and confidence level with typical fault cases in the historical fault database, construct a fault similarity scoring matrix, extract the degradation curve template corresponding to the fault type, and calculate the remaining service life of the equipment.

[0096] Preferably, the specific formula for the fault similarity scoring matrix is ​​as follows:

[0097]

[0098] Where S is the similarity score between the i-th current feature vector and the j-th historical case, d is the Euclidean distance, θ is the angle between the feature vectors, ρ is the Pearson correlation coefficient, and β and γ are weight coefficients.

[0099] In an optional implementation, when the confidence level of the fault type is higher than the first confidence threshold and the historical case matching degree is greater than the first matching threshold, the fault case is marked as a first-level reliable matching case; when the confidence level of the fault type is between the second confidence threshold and the first confidence threshold, and the historical case matching degree is between the second matching threshold and the first matching threshold, the fault case is marked as a second-level reliable matching case; when the confidence level of the fault type is less than the second confidence threshold or the historical case matching degree is less than the second matching threshold, the fault case is marked as a third-level reliable matching case.

[0100] It should be noted that the first confidence threshold is based on the 85th percentile of the diagnostic accuracy in historical fault diagnosis results; the second confidence threshold is based on the 70th percentile of the diagnostic accuracy in historical fault diagnosis results; the first matching threshold is based on the 80th percentile of the similarity calculation results of historical fault cases; and the second matching threshold is based on the 75th percentile of the similarity calculation results of historical fault cases.

[0101] In an optional implementation, for a first-level trusted matching case: if the remaining lifetime is less than a first time threshold and the device load rate is greater than a first load threshold, a first-level warning is triggered; if the remaining lifetime is between the first time threshold and a second time threshold, and the device load rate is between the second load threshold and the first load threshold, a second-level warning is triggered; if the remaining lifetime is between the second time threshold and a third time threshold, and the device load rate is less than the second load threshold, a third-level warning is triggered.

[0102] In an optional implementation, for the second-level trusted matching case: if the remaining lifetime is less than the fourth time threshold and the device load rate is greater than the third load threshold, a first-level warning is triggered; if the remaining lifetime is between the fourth and fifth time thresholds and the device load rate is between the fourth and third load thresholds, a second-level warning is triggered; if the remaining lifetime is between the fifth and sixth time thresholds and the device load rate is less than the fourth load threshold, a third-level warning is triggered.

[0103] In an optional implementation, for a three-level trusted matching case: if the remaining lifetime is less than the seventh time threshold and the device load rate is greater than the fifth load threshold, a first-level warning is triggered; if the remaining lifetime is between the seventh and eighth time thresholds and the device load rate is between the sixth and fifth load thresholds, a second-level warning is triggered; if the remaining lifetime is between the eighth and ninth time thresholds and the device load rate is less than the sixth load threshold, a third-level warning is triggered.

[0104] Furthermore, the specific formula for the remaining useful life is as follows:

[0105]

[0106] Where RUL is the remaining useful life, P is the number of historical cases selected, and w a w is the weight of the a-th case. a For the total lifetime of the a-th case, L a For the total lifetime of the a-th case, t a S is the time when the a-th case reaches its current degradation level. c S represents the current degradation state value. a Let be the degradation state value of the a-th case.

[0107] It should be noted that the following time thresholds are defined: First time threshold: 25th percentile based on the statistical data of equipment failure development cycle; Second time threshold: 50th percentile based on the statistical data of equipment failure development cycle; Third time threshold: 75th percentile based on the statistical data of equipment failure development cycle; Fourth time threshold: average development cycle based on first-level confidence failure cases; Fifth time threshold: average development cycle based on second-level confidence failure cases; Sixth time threshold: average development cycle based on third-level confidence failure cases; Seventh time threshold: shortest response time based on critical failures; Eighth time threshold: shortest response time based on important failures; Ninth time threshold: shortest response time based on general failures; First load threshold: 80th percentile based on the rated load of the equipment; Second load threshold: 75th percentile based on the rated load of the equipment; Third load threshold: upper limit of the safe operating load of the equipment; Fourth load threshold: upper limit of the optimal operating load range of the equipment; Fifth load threshold: statistical results of the peak load of the equipment; Sixth load threshold: statistical results of the average load of the equipment.

[0108] S4.2: Establish dynamic early warning classification rules based on the reliability of typical failure cases, classify early warning information into levels, and collect the real-time load rate of equipment.

[0109] In an optional implementation, the warning classification rules include:

[0110] The triggering conditions for a Level 1 warning (meeting any of the following conditions): the fault type is a critical component fault and the fault evolution rate exceeds the first rate threshold; the equipment vibration amplitude exceeds the safe operation limit and affects the normal operation of adjacent equipment; the fault causes the equipment efficiency to decrease by more than the first efficiency reduction threshold and the duration exceeds the first time threshold; multiple monitoring points show abnormalities at the same time and the degree of abnormality exceeds the first abnormality threshold.

[0111] The triggering conditions for a Level 2 warning (meeting any of the following conditions): the fault type is a critical component fault and the fault evolution rate is between the second rate threshold and the first rate threshold; the equipment vibration amplitude is close to the upper limit of safe operation but has not affected adjacent equipment; the fault causes the equipment efficiency reduction to be between the second efficiency reduction threshold and the first efficiency reduction threshold; the abnormality of a single key monitoring point exceeds the second abnormality threshold and the duration exceeds the second time threshold.

[0112] The triggering conditions for a Level 3 warning (meeting any of the following conditions): the fault type is a general component fault and the fault evolution rate is lower than the second rate threshold; the equipment vibration amplitude fluctuates but does not exceed the safe operating range; the equipment efficiency reduction caused by the fault is less than the second efficiency reduction threshold; the monitoring point shows intermittent anomalies and the degree of anomaly is lower than the second anomaly threshold.

[0113] It should be noted that the first rate threshold is based on the rate of change of characteristic parameters during the rapid development stage of the fault; the second rate threshold is based on the rate of change of characteristic parameters during the stable development stage of the fault; the first efficiency reduction threshold is based on the statistical value of the inflection point of a sharp decline in equipment efficiency; the second efficiency reduction threshold is based on the statistical value of the inflection point of a slow decline in equipment efficiency; the first anomaly threshold is based on twice the standard deviation of the fluctuation range of the normal operating parameters of the equipment; the second anomaly threshold is based on 1.5 times the standard deviation of the fluctuation range of the normal operating parameters of the equipment.

[0114] S4.3: Adjust the remaining service life based on the real-time load rate. When the real-time load rate is higher than the rated load rate, shorten the remaining service life according to the load ratio.

[0115] S4.4: Calculate the accuracy of the early warning level and construct the early warning confidence correction coefficient.

[0116] It should be noted that the confidence correction coefficient for early warning increases as the accuracy of early warning increases.

[0117] S4.5: Multiply the warning confidence correction coefficient by the current confidence level to obtain the corrected confidence level value, and then push the warning information in a tiered manner.

[0118] As illustrated in the example, during the operation of a certain rail transit hub, a fault diagnosis model identifies a bearing fault type with a confidence level of 0.88, which is higher than the first confidence threshold of 0.85. This fault case is then matched against a historical database to calculate a fault similarity score matrix. The highest matching degree is 0.83, which is higher than the first matching degree threshold of 0.80, thus marking this case as a first-level reliable matching case. The degradation curve template of the bearing fault is extracted from the matched historical cases. Based on the vibration characteristics, temperature characteristics, and other parameters of the current equipment, the remaining service life is calculated to be 120 hours. Since the current real-time load rate of the equipment is 85%, which is higher than the first load threshold of 80%, and the remaining service life is less than the first time threshold of 150 hours, the system triggers a first-level warning.

[0119] As illustrated in the example, the bearing failure evolution rate is 0.15 / hour, exceeding the first rate threshold of 0.12 / hour, and the equipment vibration amplitude has reached 1.8 times the safe operating limit, affecting the normal operation of adjacent equipment. Simultaneously, equipment efficiency decreases by 15%, exceeding the first efficiency reduction threshold of 12%, and this state has lasted for more than the first time threshold. The anomalies at multiple monitoring points all exceed the first anomaly threshold, validating the rationale for the Level 1 warning.

[0120] As illustrated in the example, historical early warning statistics show that the accuracy rate of Level 1 early warnings for this type of fault is 92%, with a corresponding early warning confidence correction coefficient of 1.05. Multiplying this correction coefficient by the current confidence level of 0.88 yields a corrected confidence value of 0.924. Based on the final confidence value and early warning level, the system pushes a Level 1 early warning message to equipment managers and maintenance personnel, recommending equipment maintenance within 24 hours.

[0121] In summary, this invention achieves comprehensive monitoring of equipment status through multimodal data acquisition and spatiotemporal alignment technology, effectively solving the challenges of data synchronization from heterogeneous sensors and cross-device fault correlation analysis. Based on a parallel feature extraction module, it integrates multi-source features such as vibration, temperature, and current, significantly improving the ability to identify complex faults. Combined with a dynamic confidence correction mechanism, it adaptively adjusts the warning threshold based on historical warning accuracy and real-time load rate, greatly reducing the false alarm rate and optimizing the timeliness of warnings. Through a hierarchical push strategy, it implements differentiated warning responses based on cases matched with different confidence levels, ensuring priority processing of high-reliability warnings while rationally allocating operation and maintenance resources. This method not only enhances the accuracy of fault prediction but also realizes the intelligence and refinement of operation and maintenance strategies, providing an efficient and reliable solution for the health management of rail transit hub equipment.

[0122] This embodiment also provides a computer device applicable to the intelligent diagnosis and early warning method for faults in rail transit hub equipment, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the intelligent diagnosis and early warning method for faults in rail transit hub equipment as proposed in the above embodiment.

[0123] This embodiment also provides an electronic device, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. The processor of this computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used for wired or wireless communication with external terminals. Wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a multi-task edge computing resource scheduling method. The display screen of the computer device can be a liquid crystal display screen or an e-ink display screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad located on the casing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0124] This embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method proposed in the above embodiments.

[0125] The storage medium proposed in this embodiment belongs to the same inventive concept as the method proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0126] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the method of the embodiments of the present invention.

[0127] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0128] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented using various computer languages.

[0129] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0130] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0131] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0132] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0133] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for intelligent diagnosis and early warning of faults in rail transit hub equipment, characterized in that: include, Multimodal data of various equipment in the rail transit hub are collected, and the multimodal data is spatiotemporally aligned based on the equipment topology and timestamps to generate multidimensional fault feature data. The multidimensional fault feature data is input into the parallel feature extraction module to generate a fault feature vector. The fault feature vector is input into the fault diagnosis model, and preliminary classification is performed based on the ResNet-18 network backbone structure, outputting the corresponding fault type and confidence level. The remaining service life of the equipment is calculated based on the fault type and confidence level, and early warning information is generated. The confidence level is dynamically adjusted by combining historical early warning statistics and real-time load rate of the equipment to achieve hierarchical management of early warning notifications. The fault type and confidence level are matched with typical fault cases in the historical fault database to construct a fault similarity score matrix, extract the degradation curve template of the corresponding fault type, and calculate the remaining service life. Based on the reliability of typical failure cases, a dynamic early warning classification rule is established to classify early warning information into different levels and collect the real-time load rate of the equipment. The remaining service life is adjusted based on the real-time load rate. When the real-time load rate is higher than the rated load rate, the remaining service life is shortened according to the load ratio. The accuracy of the statistical early warning levels was calculated, and an early warning confidence correction coefficient was constructed. The warning confidence correction coefficient is multiplied by the current confidence level to obtain the corrected confidence level value, and the warning information is then pushed out in a tiered manner.

2. The intelligent diagnosis and early warning method for faults in rail transit hub equipment as described in claim 1, characterized in that: Match fault types and confidence levels with typical fault cases in the historical fault database, including: When the confidence level of the fault type is higher than the first confidence threshold and the historical case matching degree is greater than the first matching threshold, the fault case is marked as a first-level reliable matching case. When the confidence level of the fault type is between the second confidence threshold and the first confidence threshold, and the historical case matching degree is between the second matching threshold and the first matching threshold, the fault case is marked as a second-level reliable matching case. When the confidence level of the fault type is less than the second confidence level threshold or the historical case matching degree is less than the second matching degree threshold, the fault case is marked as a level 3 trusted matching case.

3. The intelligent fault diagnosis and early warning method for rail transit hub equipment as described in claim 2, characterized in that: It also includes, For the first-level trusted matching case: if the remaining lifetime is less than the first time threshold and the device load rate is greater than the first load threshold, a first-level warning is triggered; if the remaining lifetime is between the first time threshold and the second time threshold, and the device load rate is between the second load threshold and the first load threshold, a second-level warning is triggered; if the remaining lifetime is between the second time threshold and the third time threshold, and the device load rate is less than the second load threshold, a third-level warning is triggered. For the aforementioned Level 2 trusted matching case: if the remaining lifetime is less than the fourth time threshold and the device load rate is greater than the third load threshold, a Level 1 warning is triggered; if the remaining lifetime is between the fourth and fifth time thresholds and the device load rate is between the fourth and third load thresholds, a Level 2 warning is triggered; if the remaining lifetime is between the fifth and sixth time thresholds and the device load rate is less than the fourth load threshold, a Level 3 warning is triggered. For the aforementioned three-level trusted matching case: if the remaining lifetime is less than the seventh time threshold and the device load rate is greater than the fifth load threshold, a first-level warning is triggered; if the remaining lifetime is between the seventh and eighth time thresholds and the device load rate is between the sixth and fifth load thresholds, a second-level warning is triggered; if the remaining lifetime is between the eighth and ninth time thresholds and the device load rate is less than the sixth load threshold, a third-level warning is triggered.

4. The intelligent diagnosis and early warning method for faults in rail transit hub equipment as described in claim 2, characterized in that: The method for obtaining the fault type and confidence level is as follows: A fault diagnosis model based on the ResNet-18 architecture is constructed, wherein the fault diagnosis model includes a feature extraction layer, a domain discrimination layer, and a fault classification layer; The fault feature vector is input into the feature extraction layer to extract a high-dimensional feature representation, wherein the high-dimensional feature representation includes deep semantic information of the device status; In the domain discrimination layer, the maximum average difference metric between the source domain features and the target domain features is calculated, and the maximum average difference metric is used as the optimization objective of the domain adaptation loss function. The parameters of the feature extraction layer are trained using labeled samples of rail transit hub equipment in the target domain, combined with the domain adaptation loss function, so that the features have domain invariance. The features trained by domain adaptation are input into the fault classification layer to calculate the probability distribution of each fault type, and output the fault type with the highest probability and the confidence value.

5. The intelligent diagnosis and early warning method for faults in rail transit hub equipment as described in claim 4, characterized in that: The method for obtaining the fault feature vector is as follows: Multidimensional fault characteristic data are categorized into vibration characteristic channel, temperature characteristic channel, and current characteristic channel according to data type. Wavelet packet decomposition is performed on the data of the vibration feature channel to extract the energy features, kurtosis features and skewness features of each frequency band, forming vibration feature sub-vectors; Statistical analysis is performed on the data of the temperature feature channels to calculate the temperature mean, standard deviation, and rate of change, forming a temperature feature sub-vector; Fourier transform is performed on the data of the current characteristic channel to extract the current harmonic ratio, phase difference and waveform distortion rate, forming a current characteristic sub-vector; The vibration feature vector, the temperature feature vector, and the current feature vector are fused together using feature importance weights to generate a fault feature vector.

6. The intelligent diagnosis and early warning method for faults in rail transit hub equipment as described in claim 5, characterized in that: The method for obtaining the multidimensional fault feature data is as follows: Collect multimodal data of various equipment within the rail transit hub, including vibration sensor data and equipment operating condition data; The multimodal data is labeled with device numbers according to the collection location, and a spatial association matrix between devices is established based on the topology diagram of rail transit hub equipment; The multimodal data is timestamped and aligned, and data with different sampling frequencies are unified to the same time scale by linear interpolation to generate a multimodal data sequence. Based on the spatial correlation matrix, spatial correlation analysis is performed on the data of adjacent devices in the multimodal data sequence to calculate the fault propagation coefficient between devices; The multimodal data sequence and the fault propagation coefficient are combined to form multidimensional fault feature data.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the intelligent diagnosis and early warning method for faults in rail transit hub equipment as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent diagnosis and early warning method for faults in rail transit hub equipment as described in any one of claims 1 to 6.

Images