Urban rail train door operation and maintenance scene classification method, device and equipment and storage medium

Abstract

The application discloses a metro train door operation and maintenance scene classification method, device and equipment and a storage medium. The method comprises the following steps: determining an operation and maintenance scene type according to a metro train door opening and closing process; defining single-sensor data operation and maintenance scenes and multi-sensor decision fusion operation and maintenance scenes according to each operation and maintenance scene type; obtaining single-sensor data operation and maintenance scene classification through fault threshold training and operation and maintenance scene classification threshold training according to the definition of the single-sensor data operation and maintenance scenes and historical data of each single sensor; and obtaining multi-sensor decision fusion operation and maintenance scene classification through fusion weight training according to each single-sensor data operation and maintenance scene classification and the definition of the multi-sensor decision fusion operation and maintenance scenes. The application can accurately capture fault characteristics and adaptively determine classification thresholds by using a large amount of unlabeled multi-source sensor data generated in the actual operation of the metro train, and has the advantages of high classification accuracy, strong adaptability and controllability.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance of urban rail trains, and in particular to a method, apparatus, equipment and storage medium for classifying operation and maintenance scenarios of urban rail train doors. Background Technology

[0002] Urban rail transit, with its safety, efficiency, stability, energy saving, and environmental friendliness, is increasingly becoming the preferred mode of transportation for the public in large and medium-sized cities. As passenger volume continues to grow, subway operators are also gradually increasing the number of urban rail trains in operation. Against this backdrop, how to reduce maintenance costs and improve maintenance efficiency while ensuring the safe and stable operation of trains has become a critical issue that urgently needs to be addressed in the urban rail transit sector.

[0003] As a critical component of urban rail trains, the door system directly impacts passenger safety and train operation safety. Due to frequent door operations and a relatively high failure rate, traditional manual maintenance methods are insufficient to meet the demands for speed and accuracy. Therefore, research on the identification and prediction of intelligent maintenance scenarios during the opening and closing of urban rail train doors is of paramount importance for achieving fault early warning and rapid decision-making, thereby providing efficient and reliable solutions and maintenance strategies for door operation and maintenance.

[0004] The inventors discovered through research that existing methods for diagnosing car door malfunctions have at least the following shortcomings:

[0005] In the actual operation of urban rail trains, due to the complexity of the environment around the doors and the multi-source and randomness of door failures, the data obtained through experiments cannot cover all failure modes. Therefore, the diagnostic and prediction models trained and constructed from experimental data have poor predictive performance.

[0006] The information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to accurately obtain the fault characteristic values ​​of each fault mode, making the classification of operation and maintenance scenarios more accurate, thereby improving the predictive performance of diagnostic and predictive models.

[0008] This invention provides a method for classifying urban rail train door maintenance scenarios, including the following steps:

[0009] S11. Determine the operation and maintenance scenario type based on the opening and closing process of urban rail train doors; the operation and maintenance scenario type includes: normal operation, fault-tolerant operation, and emergency decision-making.

[0010] S12. Define single-sensor data operation and maintenance scenarios and multi-sensor decision fusion operation and maintenance scenarios according to each of the aforementioned operation and maintenance scenario types; wherein, the definition of single-sensor data operation and maintenance scenarios includes the following formula:

[0011]

[0012] In the formula, P i Categorize and label data operation and maintenance scenarios for each single sensor; x p The raw values ​​of single sensor data; θ p λ is the fault threshold; n is the number of sampling points; λ is the fault characteristic value; λ θ The classification threshold for single-sensor data scenarios;

[0013] The definition of a multi-sensor decision fusion operation and maintenance scenario includes the following formulas:

[0014]

[0015] In the formula, P + For fusion value; w i1 w i2 w i3 , where are the fusion weights of the scene classification results for single-sensor data; s is the number of sensors;

[0016] S13. Based on the definition of the single-sensor data operation and maintenance scenario and the historical data of each single sensor, the single-sensor data operation and maintenance scenario classification is obtained through fault threshold training and operation and maintenance scenario classification threshold training; the categories of the single-sensor data operation and maintenance scenario classification include speed, torque and angle.

[0017] S14. Based on the classification of single-sensor data operation and maintenance scenarios and the definition of multi-sensor decision fusion operation and maintenance scenarios, the classification of multi-sensor decision fusion operation and maintenance scenarios is obtained through fusion weight training.

[0018] Preferably, in this embodiment of the invention, the step of obtaining the classification of single-sensor data operation and maintenance scenarios through fault threshold training and operation and maintenance scenario classification threshold training includes the following steps:

[0019] S131. Based on the historical data of the motor, including speed, torque and angle, collected during the opening and closing of the urban rail train doors, determine the best fitting curve based on the historical data of each of the single sensors.

[0020] S132. Determine the fault threshold based on the stage division of the best fitting curve; and divide the single sensor historical data into normal samples and fault samples based on the fault threshold;

[0021] S133. Extract fault feature values ​​based on the fault samples; the fault feature values ​​include: mean overshoot deviation, overshoot duration, and overshoot rate of change;

[0022] S134. Based on the fault feature values, obtain the operation and maintenance scenario classification threshold through the adaptive classification threshold algorithm and the fault feature value threshold difference improvement algorithm.

[0023] Preferably, in this embodiment of the invention, determining the best fitting curve based on the historical data of each of the single sensors includes:

[0024] The mean value of the historical data from each sampling point of the single sensor is calculated using the following formula:

[0025]

[0026] In the formula, The mean of the data from each sampling point; N is the number of samples; n is the number of sensor sampling points during the door opening and closing process; x pi Let be the value of the p-th sampling point and the i-th sample.

[0027] Preferably, in this embodiment of the invention, the stage division of the best-fit curve includes:

[0028] The acceleration phase, the steady phase, the deceleration phase, and the arrival phase.

[0029] Preferably, in this embodiment of the invention, determining the fault threshold includes:

[0030] Using the raw values ​​of the historical data from each individual sensor as the basis for fault diagnosis, the standard deviation of each sampling point of the historical data from each individual sensor is calculated using the following formula:

[0031]

[0032] In the formula, σ p is the standard deviation of the sampling points; N is the sample size; x p The original data is from a single sensor; n is the number of sensor sampling points during the door opening and closing process; x pi Let be the value of the p-th sampling point and the i-th sample.

[0033] Preferably, in this embodiment of the invention, extracting fault feature values ​​based on the fault sample includes:

[0034] The mean overshoot deviation is calculated using the following formula:

[0035]

[0036] In the formula, a is the starting point of consecutive sampling points exceeding the threshold; b is the ending point of consecutive sampling points exceeding the threshold.

[0037] The overshoot duration is calculated using the following formula:

[0038] t = b - a + 1

[0039] The overshoot rate is calculated using the following formula:

[0040]

[0041] In the formula, f is the sampling point where the overshoot deviation first reaches the mean or a set value; x f The value of sampling point f; σ is the mean of all samples at sampling point f; f Let f be the standard deviation of each sample at sampling point f.

[0042] Preferably, in this embodiment of the invention, the classification of operation and maintenance scenarios obtained by multi-sensor decision fusion through fusion weight training includes:

[0043] The fusion weights are obtained by classifying the single-sensor data operation and maintenance scenarios using a fusion weighting algorithm.

[0044] Based on the fusion weights, a classification of operation and maintenance scenarios for multi-sensor decision fusion is obtained.

[0045] In another aspect of the present invention, a classification device for the operation and maintenance scenarios of urban rail train doors is also provided, comprising:

[0046] The operation and maintenance scenario type classification unit is used to determine the operation and maintenance scenario type based on the opening and closing process of urban rail train doors; the operation and maintenance scenario types include: normal operation, fault-tolerant operation, and emergency decision-making.

[0047] The operation and maintenance scenario definition unit is used to define single-sensor data operation and maintenance scenarios and multi-sensor decision fusion scenarios according to each of the operation and maintenance scenario types; wherein, the definition of single-sensor data operation and maintenance scenarios includes the following formula:

[0048]

[0049] In the formula, P i Classify and label the data scenarios for each single sensor; x p The raw values ​​of single sensor data; θ p λ is the fault threshold; n is the number of sampling points; λ is the fault characteristic value; λ θ The classification threshold for single-sensor data scenarios;

[0050] The definition of a multi-sensor decision fusion operation and maintenance scenario includes the following formulas:

[0051]

[0052] In the formula, P + For fusion value; w i1 w i2 w i3 , where are the fusion weights of the scene classification results for single-sensor data; s is the number of sensors;

[0053] A single-sensor data operation and maintenance scenario classification construction unit is used to obtain a single-sensor data operation and maintenance scenario classification based on the definition of the single-sensor data operation and maintenance scenario and the historical data of each single sensor, through fault threshold training and operation and maintenance scenario classification threshold training; the categories of the single-sensor data operation and maintenance scenario classification include speed, torque and angle;

[0054] The multi-sensor decision fusion operation and maintenance scenario classification construction unit is used to obtain the multi-sensor decision fusion operation and maintenance scenario classification by training through fusion weights based on the operation and maintenance scenario classification of each single sensor data and the definition of the multi-sensor decision fusion operation and maintenance scenario.

[0055] In another aspect of this invention, a classification device for the operation and maintenance scenarios of urban rail train doors is also provided. The urban rail train door operation and maintenance scenario classification device includes a computer program stored on a medium. The computer program includes program instructions. When the program instructions are executed by the computer, the computer performs the methods described in the above aspects and achieves the same technical effect.

[0056] In another aspect of this invention, a storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the various steps of the urban rail train door maintenance scenario classification method as described in any of the preceding claims.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] In this invention, the operation and maintenance scenario type is first determined based on the door opening and closing process of urban rail trains. This scenario type includes normal operation, fault-tolerant operation, and emergency decision-making. This scenario type meets the requirements for speed and accuracy of operation and maintenance services during the door opening and closing process of urban rail trains. Single-sensor data operation and maintenance scenarios and multi-sensor decision fusion operation and maintenance scenarios are defined according to each scenario type. Then, based on the definition of the single-sensor data operation and maintenance scenarios and the historical data of each single sensor, a single-sensor data operation and maintenance scenario classification is obtained through fault threshold training and operation and maintenance scenario classification threshold training. Finally, based on the single-sensor data operation and maintenance scenario classifications and the definitions of the multi-sensor decision fusion operation and maintenance scenarios, a multi-sensor decision fusion operation and maintenance scenario classification is obtained through fusion weight training. In this invention, the large amount of unlabeled multi-source sensor data generated during the actual operation of urban rail trains is integrated into scenario classification to accurately capture fault characteristics and adaptively determine classification thresholds. This results in accurate classification, strong adaptability, and controllability. The fusion decision-making of multi-source heterogeneous sensor data in this invention further improves the accuracy of operation and maintenance scenario classification, thereby also improving the predictive performance of diagnostic and predictive models.

[0059] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it according to the contents of the specification, and to make the above and other objects, technical features and advantages of the present invention easier to understand, one or more preferred embodiments are listed below and described in detail with reference to the accompanying drawings. Attached Figure Description

[0060] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in 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.

[0061] Figure 1 This is a flowchart illustrating the steps of the urban rail train door maintenance scenario classification method described in this invention;

[0062] Figure 2 This is a flowchart illustrating the operation and maintenance scenario classification method based on multi-source sensor data fusion described in this invention.

[0063] Figure 3 This is a schematic diagram of the best-fit curve stage division described in this invention;

[0064] Figure 4 This is a flowchart illustrating the optimal classification evaluation index algorithm based on the greedy idea and sample distribution assumptions described in this invention.

[0065] Figure 5This is a flowchart of the improved adaptive scene classification threshold algorithm described in this invention.

[0066] Figure 6 This is a flowchart illustrating the fusion weight algorithm described in this invention;

[0067] Figure 7 This is a schematic diagram of the urban rail train door maintenance scenario classification device described in this invention;

[0068] Figure 8 This is a structural schematic diagram of the urban rail train door maintenance scenario classification device described in this invention. Detailed Implementation

[0069] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings, but it should be understood that the scope of protection of the present invention is not limited to the specific embodiments.

[0070] Unless otherwise expressly stated, throughout the specification and claims, the term "comprising" or its variations such as "including" or "comprises" shall be understood to include the stated elements or components without excluding other elements or other components.

[0071] In this document, the terms "first," "second," etc., are used to distinguish two different elements or parts, and are not used to define specific positions or relative relationships. In other words, in some embodiments, the terms "first," "second," etc., can also be used interchangeably.

[0072] Example 1

[0073] To make the classification of operation and maintenance scenarios more accurate, such as Figure 1 and Figure 2 As shown, this embodiment of the invention provides a method for classifying urban rail train door maintenance scenarios, including the following steps:

[0074] S11. Determine the operation and maintenance scenario type based on the opening and closing process of urban rail train doors; the operation and maintenance scenario type includes: normal operation, fault-tolerant operation, and emergency decision-making.

[0075] The inventive concept of this invention includes the following: In the actual operation of urban rail trains, due to the complexity of the environment around the doors and the multi-source and random nature of door failures, the data obtained through experiments cannot cover all fault modes. Diagnostic and predictive models trained on experimental data are difficult to quickly determine the fault degradation trend under complex operating conditions, resulting in poor model performance. This invention utilizes a large amount of unlabeled multi-source sensor historical data generated by the frequent opening and closing of doors for operation and maintenance scenario classification. It can accurately capture fault characteristics by incorporating the classification requirements of operation and maintenance scenarios and adaptively determine the classification threshold, resulting in better classification accuracy, adaptability, and controllability. This also improves the predictive performance of diagnostic and predictive models.

[0076] In this embodiment of the invention, operation and maintenance scenario types are defined from the perspective of operation and maintenance requirements, including: normal operation, fault-tolerant operation, and emergency decision-making, which are three typical operation and maintenance scenario types. Among them, the normal operation and maintenance scenario corresponds to the correct and complete opening and closing of the door, and no disposal measures are required in this operation and maintenance scenario. The emergency decision-making operation and maintenance scenario corresponds to faults that are sudden and of high severity or endanger personal safety. Such faults develop rapidly and can cause great harm to equipment operation or personal safety. Therefore, this operation and maintenance scenario requires rapid decision-making and allows for a certain sacrifice of accuracy. The fault-tolerant operation and maintenance scenario corresponds to other fault situations. Such faults may develop slowly and have monotonous characteristics. Their degradation trend can be predicted and the harm is relatively small. Fault-tolerant operation and fault diagnosis can be performed. This operation and maintenance scenario requires precise decision-making and allows for a certain sacrifice of response time.

[0077] In this embodiment of the invention, while retaining sufficient fault samples, the following sample distribution assumptions are made for the three types of operation and maintenance scenarios: In the overall sample, the operation and maintenance scenarios with normal operation are the majority; in the fault samples, the operation and maintenance scenarios with fault tolerance are the majority. The overall sample proportion range can be designed as follows: 70% < normal operation scenarios ≤ 99%; 0.7% < fault tolerance operation and maintenance scenarios < 29.7%; 0.01% ≤ emergency decision-making operation and maintenance scenarios < 9%.

[0078] S12. Define single-sensor data operation and maintenance scenarios and multi-sensor decision fusion operation and maintenance scenarios according to each of the aforementioned operation and maintenance scenario types; wherein, the definition of single-sensor data operation and maintenance scenarios includes the following formula:

[0079]

[0080] In the formula, P i Categorize and label data operation and maintenance scenarios for each single sensor; x p The raw values ​​of single sensor data; θ p λ is the fault threshold; n is the number of sampling points; λ is the fault characteristic value; λ θ The classification threshold for single-sensor data scenarios;

[0081] The definition of a multi-sensor decision fusion operation and maintenance scenario includes the following formulas:

[0082]

[0083] In the formula, P + For fusion value; w i1 w i2 w i3 , where are the fusion weights of the scene classification results for single-sensor data; s is the number of sensors.

[0084] In this embodiment of the invention, P i =1 indicates normal operation, P i =2 indicates fault-tolerant operation, P i =3 indicates an emergency decision.

[0085] Based on the formulas defined above for multi-sensor decision fusion operation and maintenance scenarios, the following formula can be obtained:

[0086]

[0087] In the formula, p is the classification label for multi-sensor decision fusion scenarios, p=1 represents a normal operation and maintenance scenario, p=2 represents a fault-tolerant operation and maintenance scenario, and p=3 represents an emergency decision-making operation and maintenance scenario.

[0088] S13. Based on the definition of the single-sensor data operation and maintenance scenario and the historical data of each single sensor, the single-sensor data operation and maintenance scenario classification is obtained through fault threshold training and operation and maintenance scenario classification threshold training; the categories of the single-sensor data operation and maintenance scenario classification include speed, torque and angle.

[0089] The process of obtaining the classification of single-sensor data operation and maintenance scenarios through fault threshold training and operation and maintenance scenario classification threshold training includes the following steps:

[0090] S131. Based on the historical data of the motor, including speed, torque and angle, collected during the opening and closing of the urban rail train doors, determine the best fitting curve based on the historical data of each of the single sensors.

[0091] In this embodiment of the invention, historical data from three types of sensors of the motor are collected simultaneously during the opening and closing of the urban rail train doors. These sensors are speed, torque, and angle. The mean value of each sampling point is calculated using the overall sample data of the historical data from a single sensor, and the best fitting curve of each sensor data during the opening and closing process is obtained.

[0092] The mean of the data at each sampling point is calculated using the following formula:

[0093]

[0094] In the formula, The mean of the data from each sampling point; N is the number of samples; n is the number of sensor sampling points during the door opening and closing process; x pi Let be the value of the p-th sampling point and the i-th sample.

[0095] S132. Determine the fault threshold based on the stage division of the best fitting curve; and divide the single sensor historical data into normal samples and fault samples based on the fault threshold;

[0096] In this embodiment of the invention, the sensor data collected during door opening and closing is a set of time-series data. Different time periods may contain different fault modes. To capture more fault modes, the best-fit curve of the door opening and closing process is divided into stages based on the rotational speed, such as... Figure 3 As shown, it includes: acceleration phase, steady phase, deceleration phase, and arrival phase. Figure 3 The horizontal and vertical axes in the figure represent the timing sampling points and the values ​​of speed, torque, and angle, respectively.

[0097] In this embodiment of the invention, determining the fault threshold includes:

[0098] Using the raw values ​​of the historical data from each individual sensor as the basis for fault diagnosis, the standard deviation of each sampling point of the historical data from each individual sensor is calculated using the following formula:

[0099]

[0100] In the formula, σ p x is the standard deviation of the sampling points, and N is the sample size; p The original data is from a single sensor; n is the number of sensor sampling points during the door opening and closing process; x pi Let be the value of the p-th sampling point and the i-th sample.

[0101] By taking ±3σ of the best-fit curve p and ±4σ p When used as a fault threshold, the failure rates of the obtained data are shown in Table 1:

[0102] Table 1:

[0103]

[0104] Referring to the sample distribution assumption in step S11, ±4σ is selected. p The fault threshold is defined as follows: when the data from a single sensor satisfies the following formula, it is considered a fault sample.

[0105]

[0106] In the formula, θ p1 The fault threshold for the sampling point; θp2 The fault threshold for the sampling point.

[0107] S133. Extract fault feature values ​​based on the fault samples; the fault feature values ​​include: mean overshoot deviation, overshoot duration, and overshoot rate of change;

[0108] In this embodiment of the invention, after extracting fault samples, in order to distinguish between operation and maintenance scenario samples with sudden and severe emergency decisions and operation and maintenance scenario samples representing other situations with fault-tolerant operation, three types of fault feature values ​​are extracted from the portion where the raw value of a single sensor data continuously exceeds the fault threshold. These three fault feature values ​​are the mean overshoot deviation, the overshoot duration, and the overshoot change rate; the overshoot change rate is the first time the overshoot deviation reaches the mean or 6σ. p The slope of the straight line drawn between the time and the zero point.

[0109] The calculation of the three types of fault characteristic values ​​includes:

[0110] The mean overshoot deviation represents the magnitude of the exceedance threshold and is calculated using the following formula:

[0111]

[0112] In the formula, a is the starting point of consecutive sampling points exceeding the threshold, and b is the ending point of consecutive sampling points exceeding the threshold.

[0113] The overshoot duration is calculated using the following formula, expressed as the number of sampling points exceeding the fault threshold:

[0114] t = b - a + 1

[0115] Offset the starting point of the deviation to the right by one sampling point on the coordinate axis, and then calculate the slope of the straight line drawn between the zero point and the first point when the overshoot deviation reaches the mean. This represents the rate of change of the fault development.

[0116]

[0117] In the formula, f is the sampling point where the overshoot deviation first reaches the mean or a set value; x f The value of sampling point f; σ is the mean of all samples at sampling point f; f Let f be the standard deviation of each sample at sampling point f.

[0118] The three types of fault characteristic values ​​are based on the classification needs of actual scenarios and have clear fault characterization meanings. The severity of the fault is characterized by the mean overshoot deviation and the duration of overshoot deviation; the severity is characterized by the first time the overshoot deviation reaches the mean or 6σ. pThe slope of the straight line drawn between time and zero represents the suddenness of the fault. Since the fault characterization meaning of these three types of fault characteristic values ​​needs to correspond with the fault scenario classification, they can be used to accurately classify the operation and maintenance scenario samples with suddenness and high severity of emergency decision-making from the fault samples.

[0119] S134. Based on the fault feature values, obtain the operation and maintenance scenario classification threshold through the adaptive classification threshold algorithm and the fault feature value threshold difference improvement algorithm.

[0120] The three types of fault feature values ​​extracted in this embodiment of the invention need to be extracted from all historical fault samples to identify data that simultaneously meet the criteria of numerous exceedances, long exceedance periods, and high rates of change as fault samples for emergency decision-making and maintenance scenarios. To determine the thresholds for the three types of fault feature values ​​as the scenario classification standard, an optimal classification evaluation index algorithm based on a greedy approach and sample distribution assumptions is designed. The algorithm flow is as follows: Figure 4 As shown, when determining the threshold of fault feature value, a threshold is selected with a certain step size and the classification evaluation index under different threshold classification is calculated. The greedy algorithm seeks optimization with the goal of maximizing the classification evaluation index. The search stops when the sample distribution condition is met, and the scene classification threshold of each fault feature value is adaptively determined.

[0121] When using a greedy algorithm to update the threshold, each step can obtain a local optimum for the classification evaluation index. However, the final classification threshold may not correspond to the optimal classification evaluation index. Considering that the three types of fault feature values ​​have similar distributions, the difference between the classification evaluation index and the threshold step is fused as the optimization objective of the greedy algorithm. The improved algorithm flow is as follows: Figure 5 As shown, the improved algorithm significantly enhances the classification evaluation metrics, especially the DB value.

[0122] The improved algorithm is used to obtain the scene classification thresholds for three types of fault feature values. For a single-sensor fault sample, if the following formula is satisfied in the fault part, it is classified as an emergency decision-making fault sample; otherwise, it is classified as a fault-tolerant operation fault sample:

[0123]

[0124] In the formula, t θ k θ The threshold values ​​are the scene classification thresholds for the three types of fault characteristic values.

[0125] S14. Based on the classification of single-sensor data operation and maintenance scenarios and the definition of multi-sensor decision fusion operation and maintenance scenarios, the classification of multi-sensor decision fusion operation and maintenance scenarios is obtained through fusion weight training.

[0126] In this embodiment of the invention, for three types of multi-source heterogeneous sensor data—speed, torque, and angle—after classifying the operation and maintenance scenarios of the single sensor data separately, a fusion weighting algorithm is used to obtain a fusion weighting value. The fusion weighting algorithm is as follows: Figure 6 As shown, the operation and maintenance scenario classification of multi-source sensor (multi-sensor) fusion is then obtained based on the fusion weights.

[0127] In summary, this embodiment of the invention first determines the operation and maintenance scenario type based on the door opening and closing process of urban rail trains. The operation and maintenance scenario types include: normal operation, fault-tolerant operation, and emergency decision-making. This operation and maintenance scenario type meets the requirements for speed and accuracy of operation and maintenance services during the door opening and closing process of urban rail trains. Single-sensor data operation and maintenance scenarios and multi-sensor decision fusion operation and maintenance scenarios are defined according to each of the aforementioned operation and maintenance scenario types. Then, based on the definition of the single-sensor data operation and maintenance scenario and the historical data of each single sensor, a single-sensor data operation and maintenance scenario classification is obtained through fault threshold training and operation and maintenance scenario classification threshold training. Finally, based on the single-sensor data operation and maintenance scenario classifications and the definitions of the multi-sensor decision fusion operation and maintenance scenarios, a multi-sensor decision fusion operation and maintenance scenario classification is obtained through fusion weight training. In this invention, the large amount of unlabeled multi-source sensor data generated during the actual operation of urban rail trains is integrated into scenario classification to accurately capture fault characteristics and adaptively determine classification thresholds. This results in accurate classification, strong adaptability, and controllability. The fusion decision-making of multi-source heterogeneous sensor data in this invention further improves the accuracy of operation and maintenance scenario classification, thereby also improving the predictive performance of diagnostic and predictive models.

[0128] Example 2

[0129] Corresponding to the method embodiment, another aspect of the present invention provides a classification device for urban rail train door maintenance scenarios. Figure 7 This diagram illustrates the structure of a classification device for urban rail train door maintenance scenarios provided in an embodiment of the present invention. The urban rail train door maintenance scenario classification device is... Figure 1 The device corresponding to the urban rail train door maintenance scenario classification method described in the corresponding embodiment is implemented through a virtual device. Figure 1 In the corresponding embodiment of the urban rail train door maintenance scenario classification method, each virtual module constituting the urban rail train door maintenance scenario classification device can be executed by electronic devices, such as network devices, terminal devices, or servers. Specifically, the urban rail train door maintenance scenario classification device in this embodiment of the invention includes:

[0130] The operation and maintenance scenario type classification unit 01 is used to determine the operation and maintenance scenario type based on the opening and closing process of urban rail train doors; the operation and maintenance scenario types include: normal operation, fault-tolerant operation, and emergency decision-making.

[0131] In this embodiment of the invention, operation and maintenance scenario types are defined from the perspective of operation and maintenance requirements, including: normal operation, fault-tolerant operation, and emergency decision-making, which are three typical operation and maintenance scenario types. While retaining sufficient fault samples, the following sample distribution assumptions are made for the three types of operation and maintenance scenarios: in the overall sample, normal operation and maintenance scenarios constitute the majority; in the fault samples, fault-tolerant operation and maintenance scenarios constitute the majority. The overall sample proportion range can be designed as follows: 70% < normal operation scenarios ≤ 99%; 0.7% < fault-tolerant operation and maintenance scenarios < 29.7%; 0.01% ≤ emergency decision-making operation and maintenance scenarios < 9%.

[0132] Operation and maintenance scenario definition unit 02 is used to define single-sensor data operation and maintenance scenarios and multi-sensor decision fusion scenarios according to each of the operation and maintenance scenario types; wherein, the definition of single-sensor data operation and maintenance scenarios includes the following formula:

[0133]

[0134] In the formula, P i Classify and label the data scenarios for each single sensor; x p The raw values ​​of single sensor data; θ p λ is the fault threshold; n is the number of sampling points; λ is the fault characteristic value; λ θ The classification threshold for single-sensor data scenarios;

[0135] The definition of a multi-sensor decision fusion operation and maintenance scenario includes the following formulas:

[0136]

[0137] In the formula, P + For fusion value; w i1 w i2 w i3 , where are the fusion weights of the scene classification results for single-sensor data; s is the number of sensors.

[0138] In this embodiment of the invention, P i =1 indicates normal operation, P i =2 indicates fault-tolerant operation, P i =3 indicates an emergency decision.

[0139] Based on the formulas defined above for multi-sensor decision fusion operation and maintenance scenarios, the following formula can be obtained:

[0140]

[0141] In the formula, P is the classification label for multi-sensor decision fusion scenarios. P=1 represents a normal operation and maintenance scenario, P=2 represents a fault-tolerant operation and maintenance scenario, and P=3 represents an emergency decision-making operation and maintenance scenario.

[0142] The single-sensor data operation and maintenance scenario classification construction unit 03 is used to obtain a single-sensor data operation and maintenance scenario classification based on the definition of the single-sensor data operation and maintenance scenario and the historical data of each single sensor through fault threshold training and operation and maintenance scenario classification threshold training; the categories of the single-sensor data operation and maintenance scenario classification include speed, torque and angle;

[0143] The process of obtaining the classification of single-sensor data operation and maintenance scenarios through fault threshold training and operation and maintenance scenario classification threshold training includes the following steps:

[0144] S131. Based on the historical data of the motor, including speed, torque and angle, collected during the opening and closing of the urban rail train doors, determine the best fitting curve based on the historical data of each of the single sensors.

[0145] In this embodiment of the invention, historical data from three types of sensors of the motor are collected simultaneously during the opening and closing of the urban rail train doors. These sensors are speed, torque, and angle. The mean value of each sampling point is calculated using the overall sample data of the historical data from a single sensor, and the best fitting curve of each sensor data during the opening and closing process is obtained.

[0146] The mean of the data at each sampling point is calculated using the following formula:

[0147]

[0148] In the formula, The mean of the data from each sampling point; N is the number of samples; n is the number of sensor sampling points during the door opening and closing process; x pi Let be the value of the p-th sampling point and the i-th sample.

[0149] S132. Determine the fault threshold based on the stage division of the best fitting curve; and divide the single sensor historical data into normal samples and fault samples based on the fault threshold;

[0150] In this embodiment of the invention, the sensor data collected during door opening and closing is a set of time-series data. Different time periods may contain different fault modes. To capture more fault modes, the best-fit curve of the door opening and closing process is divided into stages based on the rotational speed, such as... Figure 3 As shown, it includes: acceleration phase, steady phase, deceleration phase and arrival phase.

[0151] In this embodiment of the invention, determining the fault threshold includes:

[0152] Using the raw values ​​of the historical data from each individual sensor as the basis for fault diagnosis, the standard deviation of each sampling point of the historical data from each individual sensor is calculated using the following formula:

[0153]

[0154] In the formula, σ p is the standard deviation of the sampling points; N is the sample size; x p The original data is from a single sensor; n is the number of sensor sampling points during the door opening and closing process; x pi Let be the value of the p-th sampling point and the i-th sample.

[0155] By taking ±3σ of the best-fit curve p and ±4σ p When using this as the fault threshold, the fault rate of the obtained data is selected based on the sample distribution assumption in step S11, choosing ±4σ. p The fault threshold is defined as follows: when the data from a single sensor satisfies the following formula, it is considered a fault sample.

[0156]

[0157] In the formula, θ p1 θ is the fault threshold for the sampling points. p2 The fault threshold for the sampling point.

[0158] S133. Extract fault feature values ​​based on the fault samples; the fault feature values ​​include: mean overshoot deviation, overshoot duration, and overshoot rate of change;

[0159] In this embodiment of the invention, the calculation of the three types of fault characteristic values ​​includes:

[0160] The mean overshoot deviation represents the magnitude of the exceedance threshold and is calculated using the following formula:

[0161]

[0162] In the formula, a is the starting point of consecutive sampling points exceeding the threshold; b is the ending point of consecutive sampling points exceeding the threshold.

[0163] The overshoot duration is calculated using the following formula, expressed as the number of sampling points exceeding the fault threshold:

[0164] t = b - a + 1

[0165] Offset the starting point of the deviation to the right by one sampling point on the coordinate axis, and then calculate the slope of the straight line drawn between the zero point and the first point when the overshoot deviation reaches the mean. This represents the rate of change of the fault development.

[0166]

[0167] In the formula, f is the sampling point where the overshoot deviation first reaches the mean or a set value; x f The value of sampling point f; σ is the mean of all samples at sampling point f; fLet f be the standard deviation of each sample at sampling point f.

[0168] S134. Based on the fault feature values, obtain the operation and maintenance scenario classification threshold through the adaptive classification threshold algorithm and the fault feature value threshold difference improvement algorithm.

[0169] The multi-sensor decision fusion operation and maintenance scenario classification construction unit 04 is used to obtain the multi-sensor decision fusion operation and maintenance scenario classification by training through fusion weights based on the single-sensor data operation and maintenance scenario classification and the definition of the multi-sensor decision fusion operation and maintenance scenario.

[0170] In this embodiment of the invention, for three types of multi-source heterogeneous sensor data—speed, torque, and angle—after classifying the operation and maintenance scenarios of the single sensor data separately, a fusion weighting algorithm is used to obtain a fusion weighting value. The fusion weighting algorithm is as follows: Figure 6 As shown, the operation and maintenance scenario classification of multi-source sensor (multi-sensor) fusion is then obtained based on the fusion weights.

[0171] It should be noted that the specific implementation method and technical effects of the urban rail train door maintenance scenario classification device in the embodiments of the present invention can be referred to Figure 1 The corresponding classification methods for the operation and maintenance scenarios of urban rail train doors will not be elaborated here.

[0172] Example 3

[0173] Corresponding to the method embodiments, this invention also provides a classification device for urban rail train door maintenance scenarios, such as a terminal and a server. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be a smartphone, tablet, laptop, desktop computer, etc., but is not limited to these.

[0174] An example diagram of the hardware structure block diagram of the urban rail train door maintenance scenario classification device provided in this embodiment of the invention is shown below. Figure 8 As shown, it may include:

[0175] Processor 1, communication interface 2, memory 3, and communication bus 4;

[0176] The processor 1, communication interface 2, and memory 3 communicate with each other via communication bus 4.

[0177] Optionally, communication interface 2 can be an interface of a communication module, such as the interface of a GSM module;

[0178] Processor 1 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0179] Memory 3 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0180] Specifically, processor 1 is used to execute the computer program stored in memory 3 to perform the following steps:

[0181] S11. Determine the operation and maintenance scenario type based on the opening and closing process of urban rail train doors; the operation and maintenance scenario type includes: normal operation, fault-tolerant operation, and emergency decision-making.

[0182] S12. Define single-sensor data operation and maintenance scenarios and multi-sensor decision fusion operation and maintenance scenarios according to each of the aforementioned operation and maintenance scenario types; wherein, the definition of single-sensor data operation and maintenance scenarios includes the following formula:

[0183]

[0184] In the formula, P i Categorize and label data operation and maintenance scenarios for each single sensor; x p The raw values ​​of single sensor data; θ p λ is the fault threshold; n is the number of sampling points; λ is the fault characteristic value; λ θ The classification threshold for single-sensor data scenarios;

[0185] The definition of a multi-sensor decision fusion operation and maintenance scenario includes the following formulas:

[0186]

[0187] In the formula, P + For fusion value; w i1 w i2 w i3 , where are the fusion weights of the scene classification results for single-sensor data; s is the number of sensors;

[0188] S13. Based on the definition of the single-sensor data operation and maintenance scenario and the historical data of each single sensor, the single-sensor data operation and maintenance scenario classification is obtained through fault threshold training and operation and maintenance scenario classification threshold training; the categories of the single-sensor data operation and maintenance scenario classification include speed, torque and angle.

[0189] S14. Based on the classification of single-sensor data operation and maintenance scenarios and the definition of multi-sensor decision fusion operation and maintenance scenarios, the classification of multi-sensor decision fusion operation and maintenance scenarios is obtained through fusion weight training.

[0190] The above-described products can execute the methods provided in the embodiments of the present invention, and possess the corresponding functional modules and beneficial effects of executing the methods. For technical details not described in detail in this embodiment, please refer to the classification method for urban rail train door maintenance scenarios provided in the embodiments of the present invention.

[0191] Example 4

[0192] In this embodiment of the invention, a storage medium is also provided, which can store a program suitable for execution by a processor, the program being used for:

[0193] S11. Determine the operation and maintenance scenario type based on the opening and closing process of urban rail train doors; the operation and maintenance scenario type includes: normal operation, fault-tolerant operation, and emergency decision-making.

[0194] S12. Define single-sensor data operation and maintenance scenarios and multi-sensor decision fusion operation and maintenance scenarios according to each of the aforementioned operation and maintenance scenario types; wherein, the definition of single-sensor data operation and maintenance scenarios includes the following formula:

[0195]

[0196] In the formula, P i Categorize and label data operation and maintenance scenarios for each single sensor; x p The raw values ​​of single sensor data; θ p λ is the fault threshold; n is the number of sampling points; λ is the fault characteristic value; λ θ The classification threshold for single-sensor data scenarios;

[0197] The definition of a multi-sensor decision fusion operation and maintenance scenario includes the following formulas:

[0198]

[0199] In the formula, P + For fusion value; w i1 w i2 w i3 , where are the fusion weights of the scene classification results for single-sensor data; s is the number of sensors;

[0200] S13. Based on the definition of the single-sensor data operation and maintenance scenario and the historical data of each single sensor, the single-sensor data operation and maintenance scenario classification is obtained through fault threshold training and operation and maintenance scenario classification threshold training; the categories of the single-sensor data operation and maintenance scenario classification include speed, torque and angle.

[0201] S14. Based on the classification of single-sensor data operation and maintenance scenarios and the definition of multi-sensor decision fusion operation and maintenance scenarios, the classification of multi-sensor decision fusion operation and maintenance scenarios is obtained through fusion weight training.

[0202] Optionally, the refined and extended functions of the program can be found in the description above.

[0203] The above-described product can execute the methods provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the methods. Technical details not described in detail in this embodiment can be found in the methods provided in other embodiments of the present invention.

[0204] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0205] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. Furthermore, the couplings or direct couplings or communication connections shown or discussed may be indirect couplings or communication connections through interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0206] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0207] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0208] It should be understood that in the embodiments of this application, the claims, various embodiments, and features can be combined with each other to solve the aforementioned technical problems.

[0209] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0210] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for classifying urban rail train door maintenance scenarios, characterized in that, Including the following steps: S11. Determine the operation and maintenance scenario type based on the opening and closing process of urban rail train doors; the operation and maintenance scenario type includes: normal operation, fault-tolerant operation, and emergency decision-making. S12. Define single-sensor data operation and maintenance scenarios and multi-sensor decision fusion operation and maintenance scenarios according to each of the aforementioned operation and maintenance scenario types; wherein, the definition of single-sensor data operation and maintenance scenarios includes the following formula: ； In the formula, Classify and label data operation and maintenance scenarios for each single sensor; This represents the raw data from a single sensor. This is the fault threshold; This represents the number of sampling points; These are fault characteristic values; The classification threshold for single-sensor data scenarios; The definition of a multi-sensor decision fusion operation and maintenance scenario includes the following formulas: ； In the formula, The fusion value; , , These are the fusion weights for scene classification results from single-sensor data; The number of sensors; S13. Based on the definition of the single-sensor data operation and maintenance scenario and the historical data of each single sensor, obtain the single-sensor data operation and maintenance scenario classification through fault threshold training and operation and maintenance scenario classification threshold training. Including the following steps: S131. Based on the historical data of the motor, including speed, torque and angle, collected during the opening and closing of the urban rail train doors, determine the best fitting curve based on the historical data of each of the single sensors. S132. Determine the fault threshold based on the stage division of the best fitting curve; and divide the single sensor historical data into normal samples and fault samples based on the fault threshold; The determination of the fault threshold includes: Using the raw values ​​of the historical data from each individual sensor as the basis for fault diagnosis, the standard deviation of each sampling point of the historical data from each individual sensor is calculated using the following formula: ； In the formula, The standard deviation of the sampling points; The number of samples; This represents the raw data value from a single sensor; n represents the number of sensor sampling points during the door opening and closing process. Let i be the value of the p-th sampling point and the i-th sample. S133. Extract fault feature values ​​based on the fault samples; the fault feature values ​​include: mean overshoot deviation, overshoot duration, and overshoot rate of change; Based on the fault samples, fault feature values ​​are extracted, including: The mean overshoot deviation is calculated using the following formula: ； In the formula, This is the starting point for consecutive sampling points exceeding the threshold. This is the end point where the sampling points continuously exceed the threshold. The overshoot duration is calculated using the following formula: ； The overshoot rate is calculated using the following formula: ； In the formula, The sampling point at which the overshoot deviation first reaches the mean or a set value; The value of sampling point f; Let f be the mean of all samples at sampling point f; Let f be the standard deviation of each sample at sampling point f; S134. Based on the fault feature values, obtain the operation and maintenance scenario classification threshold through an adaptive classification threshold algorithm and an improved algorithm for threshold differences between fault feature values; The categories for the single-sensor data operation and maintenance scenarios include speed, torque, and angle. S14. Based on the classification of single-sensor data operation and maintenance scenarios and the definition of multi-sensor decision fusion operation and maintenance scenarios, the classification of multi-sensor decision fusion operation and maintenance scenarios is obtained through fusion weight training.

2. The method for classifying urban rail train door maintenance scenarios according to claim 1, characterized in that, The step of determining the best fitting curve based on the historical data of each of the individual sensors includes: The mean value of the historical data from each sampling point of the single sensor is calculated using the following formula: ； In the formula, This represents the mean of the data from each sampling point. n is the number of samples; n is the number of sensor sampling points during the door opening and closing process. Let be the value of the p-th sampling point and the i-th sample.

3. The method for classifying urban rail train door maintenance scenarios according to claim 2, characterized in that, The stage division of the best-fit curve includes: The acceleration phase, the steady phase, the deceleration phase, and the arrival phase.

4. The method for classifying urban rail train door maintenance scenarios according to claim 1, characterized in that, The classification of operation and maintenance scenarios obtained through multi-sensor decision fusion by training with fusion weights includes: The fusion weights are obtained by classifying the single-sensor data operation and maintenance scenarios using a fusion weighting algorithm. Based on the fusion weights, a classification of operation and maintenance scenarios for multi-sensor decision fusion is obtained.

5. A classification device for the operation and maintenance scenarios of urban rail train doors, characterized in that, include: The operation and maintenance scenario type classification unit is used to determine the operation and maintenance scenario type based on the opening and closing process of urban rail train doors. The types of operation and maintenance scenarios include: normal operation, fault-tolerant operation, and emergency decision-making. The operation and maintenance scenario definition unit is used to define single-sensor data operation and maintenance scenarios and multi-sensor decision fusion scenarios according to each of the operation and maintenance scenario types; wherein, the definition of single-sensor data operation and maintenance scenarios includes the following formula: ； In the formula, Classify and label the data scenarios of each individual sensor; This represents the raw data from a single sensor. This is the fault threshold; This represents the number of sampling points; These are fault characteristic values; The classification threshold for single-sensor data scenarios; The definition of a multi-sensor decision fusion operation and maintenance scenario includes the following formulas: ； In the formula, The fusion value; , , These are the fusion weights for scene classification results from single-sensor data; The number of sensors; A single-sensor data operation and maintenance scenario classification construction unit is used to obtain a single-sensor data operation and maintenance scenario classification based on the definition of the single-sensor data operation and maintenance scenario and the historical data of each single sensor, through fault threshold training and operation and maintenance scenario classification threshold training; the categories of the single-sensor data operation and maintenance scenario classification include speed, torque and angle; The multi-sensor decision fusion operation and maintenance scenario classification construction unit is used to obtain the multi-sensor decision fusion operation and maintenance scenario classification by training through fusion weights based on the operation and maintenance scenario classification of each single sensor data and the definition of the multi-sensor decision fusion operation and maintenance scenario.

6. A classification device for the operation and maintenance scenarios of urban rail train doors, characterized in that, include: Memory, used to store computer programs; A processor is used to call and execute the computer program to implement the steps of the urban rail train door maintenance scenario classification method as described in any one of claims 1-4.

7. A storage medium, characterized in that, Includes a software program, which is adapted to be executed by a processor of the steps of the urban rail train door maintenance scenario classification method as described in any one of claims 1-4.

Images