Training of fault detection models and their fault detection methods, equipment and storage media

Abstract

This invention discloses a fault detection model training method, equipment, and storage medium. The method includes: collecting raw operating data and fault status of the elevator car door motor during the opening and closing operation; preprocessing the raw operating data based on the opening and closing stroke of the car door to obtain target operating data; extracting trend features related to the fault status from the target operating data; performing self-verification on the trend features to filter out trend features generated when the user interferes with the car door; if the self-verification is completed, training the fault detection model for the car door motor based on the trend features and fault status, so that the fault detection model can be used to calculate the total probability of the car door motor experiencing a fault status in the next time period. This embodiment effectively improves the quality of trend features for fault prediction, enhances the predictability of faults, comprehensively improves the accuracy of fault detection, and enhances the guidance significance for elevator maintenance.

Description

Technical Field

[0001] This invention relates to the technical field of elevators, and in particular to a fault detection model training method, equipment and storage medium for fault detection. Background Technology

[0002] The elevator car doors are operated by motors, and the car door motors are one of the most prone to failure among the many components of an elevator.

[0003] Currently, sensors are installed in the motor of the car door to collect its operating data, and a Seq2Seq neural network is used to predict the state at the next moment, thereby predicting whether the car door motor will malfunction.

[0004] However, most failures of the car door motor are largely related to human factors. The data collected by the sensors is subject to significant interference and has a low correlation with the failures, resulting in low accuracy in predicting whether the car door motor will fail.

[0005] Furthermore, the state of the car door motor at the next moment is instantaneous, which has limited guiding significance for elevator maintenance. Summary of the Invention

[0006] In view of this, the present invention provides a fault detection model training method, equipment and storage medium for fault detection, in order to improve the accuracy of predicting the state of the car door motor over a longer period of time, thereby improving the guidance significance for maintenance.

[0007] A first aspect of the present invention provides a method for training a fault detection model, comprising:

[0008] When the elevator car door is opening and closing, the original operating data and fault status of the car door motor are collected.

[0009] The original operating data is preprocessed based on the opening and closing stroke of the car door to obtain the target operating data;

[0010] Extract trend features related to the fault state from the target operating data;

[0011] The trend characteristics are self-verified based on the fault status to filter out the trend characteristics generated when the user interferes with the car door;

[0012] If self-verification is completed, a fault detection model for the car door motor is trained based on the trend characteristics and the fault state, so that the fault detection model can be used to calculate the total probability of the car door motor experiencing a fault state in the next time period.

[0013] A second aspect of the present invention provides a fault detection method, comprising:

[0014] Load the fault detection model trained according to the method described in the first aspect;

[0015] When the elevator car door is opening and closing, the original operating data and fault status of the car door motor are collected.

[0016] The original operating data is preprocessed based on the opening and closing stroke of the car door to obtain the target operating data;

[0017] Extract trend features related to fault states from the target operational data;

[0018] The trend characteristics are self-verified based on the fault status to filter out the trend characteristics generated when the user interferes with the car door;

[0019] If self-verification is completed, the trend characteristics within the current time period are input into the fault detection model to calculate the total probability of the car door motor malfunctioning in the next time period.

[0020] A third aspect of the present invention provides a training apparatus for a fault detection model, comprising:

[0021] The data acquisition module is used to collect raw operating data and fault status of the elevator car door motor when the elevator car door performs the opening and closing operation;

[0022] The preprocessing module is used to preprocess the original operating data based on the opening and closing stroke of the car door to obtain the target operating data;

[0023] A trend feature extraction module is used to extract trend features related to the fault state from the target operating data;

[0024] A trend feature self-verification module is used to self-verify the trend feature based on the fault state, so as to filter out the trend feature generated when the user interferes with the car door;

[0025] The fault detection model training module is used to train a fault detection model for the car door motor based on the trend characteristics and the fault state after self-verification is completed, so that the fault detection model can be used to calculate the total probability of the car door motor experiencing a fault state in the next time period.

[0026] A fourth aspect of the present invention provides a fault detection device, comprising:

[0027] The fault detection model loading module is used to load the fault detection model trained according to the method described in Embodiment 1.

[0028] The data acquisition module is used to collect raw operating data and fault status of the elevator car door motor when the elevator car door performs the opening and closing operation;

[0029] The preprocessing module is used to preprocess the original operating data based on the opening and closing stroke of the car door to obtain the target operating data;

[0030] A trend feature extraction module is used to extract trend features related to the fault state from the target operating data;

[0031] A trend feature self-verification module is used to self-verify the trend feature based on the fault state, so as to filter out the trend feature generated when the user interferes with the car door;

[0032] The total probability calculation module is used to calculate the total probability of the car door motor failing in the next time period by inputting the trend characteristics in the current time period into the fault detection model if self-verification is completed.

[0033] A fifth aspect of the present invention provides an electronic device comprising:

[0034] At least one processor; and

[0035] A memory communicatively connected to the at least one processor; wherein,

[0036] The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform a training method for a fault detection model as described in the first aspect above or a fault detection method as described in the second aspect above.

[0037] A sixth aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements a training method for a fault detection model as described in the first aspect above or a fault detection method as described in the second aspect above.

[0038] A seventh aspect of the present invention provides a computer program product comprising a computer program that, when executed by a processor, implements a training method for a fault detection model as described in the first aspect above or a fault detection method as described in the second aspect above.

[0039] In this embodiment, when the elevator car door performs the opening and closing operation, raw operating data and fault status of the car door motor are collected. Preprocessing is performed on the raw operating data based on the opening and closing stroke of the car door to obtain target operating data. Trend features related to the fault status are extracted from the target operating data. Self-verification is performed on the trend features to filter out trend features generated when users interfere with the car door. If self-verification is completed, a fault detection model is trained on the car door motor based on the trend features and fault status, so that the fault detection model can be used to calculate the total probability of the car door motor experiencing a fault status in the next time period. This embodiment, by preprocessing based on the characteristics of the car door and eliminating user interference, can effectively improve the quality of trend features for fault prediction, improve fault predictability, and improve the generalization of fault detection using the fault rate over a time period, thus comprehensively improving the accuracy of fault detection and enhancing the guidance significance for elevator maintenance.

[0040] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0042] Figure 1 This is a flowchart of a training method for a fault detection model provided in Embodiment 1 of the present invention.

[0043] Figure 2 This is an example diagram of raw operating data provided in Embodiment 1 of the present invention.

[0044] Figure 3 This is a feature example diagram provided in Embodiment 1 of the present invention when there is no user interference.

[0045] Figure 4 This is a feature example diagram provided in Embodiment 1 of the present invention when there is user interference.

[0046] Figure 5 This is a flowchart of a fault detection method provided in Embodiment 2 of the present invention.

[0047] Figure 6 This is a schematic diagram of the structure of a training device for a fault detection model provided in Embodiment 3 of the present invention.

[0048] Figure 7 This is a schematic diagram of the structure of a fault detection device provided in Embodiment 4 of the present invention.

[0049] Figure 8 This is a schematic diagram of the structure of an electronic device provided in Embodiment 5 of the present invention. Detailed Implementation

[0050] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0051] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be used interchangeably where appropriate so that the embodiments of the invention described herein can cover implementations in sequences other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0052] Example 1

[0053] See Figure 1 This diagram illustrates a flowchart of a fault detection model training and fault detection method according to Embodiment 1 of the present invention. This method can be executed by a fault detection model training and fault detection device, which can be implemented in hardware and / or software. This device can be configured in electronic devices, particularly a cluster server providing services to elevators. The fault detection model is periodically trained and updated using the accumulated raw operating data and fault status of each elevator's car doors, or it can be trained and updated in real-time based on the fault detection model's on-site processing performance. The fault detection model is then distributed to each elevator for deployment. Figure 1 As shown, the method includes:

[0054] Step 101: When the elevator car door is opening and closing, collect the original operating data and fault status of the car door motor.

[0055] Different types of buildings, especially high-rise buildings, have different transportation needs for people, pets, and goods. Therefore, different types of elevators can be deployed in buildings according to different transportation needs, such as passenger elevators, freight elevators, sightseeing elevators, etc. The method in this embodiment can be applied to various types of elevators.

[0056] This embodiment can be applied to elevators. An elevator is a complex system, and the structure of an elevator varies in different types of elevators.

[0057] In one example, a certain type of elevator is configured with: a controller (also known as an elevator control system), call buttons distributed on each floor, a car (including car doors), a motor for pulling the car (also known as a traction machine), a control cabinet, a speed governor, a door operator, a car frame, car doors, counterweight guide rails, car guide rails, guide rail supports, traveling cables, a counterweight device, a compensating chain (cable), landing doors, a guide device for the compensating chain (cable), buffers, etc.

[0058] The car door is equipped with a motor, which drives the car door to open and close.

[0059] In some types of elevators, the traction machine, control cabinet, speed governor, traveling cable, etc., can be omitted.

[0060] These devices can be divided into different sets according to their functions, thus forming various subsystems that support the operation of the elevator. The controller is connected to multiple systems of the elevator via wired means such as serial port or serial clock line (SCL). The controller monitors each system and controls the operation of each subsystem, so that the car moves in the hoistway and reaches each floor of the building.

[0061] In one example, the controller includes a door system, a frequency conversion system, a call system, and a traction system. The door system controls the elevator doors. The car is equipped with a car door, and the elevator has hall doors on each floor. The elevator doors include the car door and the hall doors on each floor. The car door and the hall door are of the same type and open / close simultaneously. The frequency conversion system controls the frequency converter. The call system controls the logic of internal call (calling the elevator from inside the car) and external call (calling the elevator from the hall). The traction system controls the car to move vertically (vertically upward or vertically downward) in the hoistway.

[0062] The elevator manager or owner can choose whether to install an edge computing node on the elevator based on factors such as the elevator's load status. If no edge computing node is installed, the controller maintains the original control logic and does not affect the normal operation of the elevator. If an edge computing node is installed, a suitable computing device can be selected as the elevator's edge computing node according to the needs. The edge computing node is combined with the original controller to form a new controller and redefine the elevator's control logic.

[0063] Generally, edge computing nodes are computing devices with strong computing capabilities, such as computers, servers, or embedded devices. In addition, depending on the different intelligent services, edge computing nodes can be equipped with graphics processing units (GPUs) or embedded neural network processors (NPUs).

[0064] Edge computing nodes refer to new business platforms built at the network edge near elevators, providing storage, computing, and network resources. This allows some critical business applications to be offloaded to the edge of the access network, reducing bandwidth and latency losses caused by network transmission and multi-level forwarding. Located between the user and the cloud (server), edge computing nodes are closer to the user (data source) than traditional cloud computing, featuring miniaturization, distribution, and user-friendliness. Massive amounts of data (such as audio data) no longer need to be uploaded to the cloud for processing; data processing can be performed at the network edge, reducing request response time, reducing network bandwidth, and ensuring data security and privacy.

[0065] In addition, edge computing nodes can implement algorithm functions and model inference, communicate with the original controller, and provide the original controller with artificial intelligence (AI) and complex computing capabilities; edge computing can also communicate with the cloud to realize algorithm functions and model updates, and relay the original controller function calls, etc.

[0066] When the elevator car door performs opening and closing operations (including opening and closing), on the one hand, raw operating data related to the fault of the car door motor is collected at a certain frequency, such as current data, voltage data, speed data, etc.

[0067] On the other hand, the fault status of the car door motor is collected, that is, whether a fault has occurred.

[0068] In practical applications, the car door motor may experience various faults, such as "acceleration overcurrent", "deceleration overcurrent", "constant speed overcurrent", "acceleration overvoltage", "deceleration overvoltage", "constant speed overvoltage", "undervoltage", "output phase loss", "hardware fault", "motor overload protection", "EEPROM abnormality", "encoder disconnection detection protection", "encoder UVW disconnection", "angle correction error", "encoder UVW error", "non-level door opening fault", "encoder abnormal pulse count", "door width self-learning error", "pulse slippage detection", "simultaneous validity of door opening and closing limit signals", "motor speed abnormality", "light curtain output circuit fault", etc.

[0069] Some of the faults are sudden, such as "hardware failure" or "encoder UVW disconnection", while others are trending faults that can be predicted using data analysis, such as "acceleration overcurrent" or "deceleration overcurrent".

[0070] Therefore, the fault status collected for the car door motor mainly refers to trend faults, and the raw operating data collected for the car door motor is correlated with trend faults.

[0071] In this embodiment, as Figure 2 As shown, the start timestamps of each opening and closing operation of the elevator car door can be located. Figure 2 The first dashed line in the middle) and the end timestamp ( Figure 2 The second dotted line in the middle represents the raw operating data of the car door motor located between the start and end timestamps. Figure 2 The data between the two dashed lines in the middle) and the fault status.

[0072] Step 102: Perform preprocessing on the original operating data based on the opening and closing stroke of the car door to obtain the target operating data.

[0073] In this embodiment, the original operating data can be preprocessed in conjunction with the characteristics of the car door to obtain target operating data, making the target operating data more compatible with the characteristics of the car door, thereby improving the performance of the fault detection model in detecting faults in the car door motor.

[0074] In practical applications, at least one of the following preprocessing operations—noise reduction, anomaly removal, and interpolation—is performed on the original operating data based on the opening and closing stroke of the car door to obtain the target operating data.

[0075] On the one hand, since the original operating data contains electrical signals, noise reduction is performed on the original operating data. Therefore, the noise reduction operation includes:

[0076] The car door is divided into multiple opening and closing stages, which typically include an opening acceleration stage, a door constant speed stage, a door deceleration stage, a door holding stage, a closing acceleration stage, a closing constant speed stage, a closing deceleration stage, a closing holding constant speed stage, and so on.

[0077] The original operating data is labeled with the operating segments under each door opening and closing stage, and bandpass filtering is performed independently on each operating segment.

[0078] On the other hand, different elevators have different historical data sets and publication records. Therefore, anomaly removal operations are adaptively performed on the original operating data. These anomaly removal operations include:

[0079] Query historical operating data; historical operating data is the operating data collected from the car door motor when the car door was opened and closed in the past, which includes normal operating data and operating data when a fault occurred.

[0080] Based on the changing trends of historical operating data (such as upward or downward trends), abnormal intervals are constructed using methods such as the 3σ criterion. That is, abnormal intervals are constructed for historical operating data under different changing trends, making the abnormal intervals dynamic.

[0081] Identify the changing trends of the original operating data, add abnormal intervals to the original operating data that match the changing trends, and remove operating parameters that are outside the abnormal intervals from the original operating data.

[0082] On the other hand, removing abnormal operating parameters results in missing operating parameters at corresponding positions in the original operating data. To maintain the continuity of each operating parameter in the original operating data, interpolation is performed on the original operating data. The interpolation operation includes:

[0083] When an operating parameter is missing at a certain location in the original operating data, the preset operating mechanism model is queried; the operating mechanism model is the standard operating data presented by the elevator car door motor when the elevator car door performs the opening and closing operation.

[0084] Algorithms such as Mean Absolute Error (MAE) are used to calculate the operational similarity between the original operational data and the operational mechanism model.

[0085] Insert a certain running parameter at a position in the original running data to increase the running similarity.

[0086] Of course, the above preprocessing is only an example. When implementing this embodiment, other preprocessing can be set according to the actual situation, and this embodiment does not limit this. In addition, besides the above preprocessing, those skilled in the art can also use other preprocessing as needed, and this embodiment does not limit this either.

[0087] Step 103: Extract trend features related to the fault state from the target operating data.

[0088] In this embodiment, feature engineering is performed on the target operating data to extract features that are related to the fault state and show a trend, which are denoted as trend features.

[0089] In practice, a window matching the fault status can be added to the target running data.

[0090] Generally, the width of a window is fixed. When designing a window, the width of the window is positively correlated with the duration of the fault state (usually the average of the durations of multiple fault states), and the width of the window is greater than the duration of the fault state. That is, the longer the duration of the fault state, the wider the window, and vice versa. In addition, the width of the window is slightly greater than the duration of the fault state.

[0091] Each time the window slides according to the preset step size, various statistically significant basic features are calculated for the various operating parameters within the window, thereby achieving the purpose of feature dimensionality reduction.

[0092] For example, basic characteristics include mean, variance, standard deviation, peak value, peak-to-peak value, root mean square, kurtosis, skewness, peak value of the spectrum, center frequency of the spectrum, bandwidth of the spectrum, entropy of the spectrum, etc.

[0093] Various basic features of the same operating parameter (such as current data, voltage data, speed data, etc.) are concatenated into a feature sequence in chronological order.

[0094] Furthermore, the fault states can be spliced ​​into a fault sequence according to time sequence, and the sequence similarity between the feature sequence and the fault sequence can be calculated using methods such as Pearson correlation coefficient. The sequence similarity can then be compared with a preset similarity threshold.

[0095] If the sequence similarity is greater than or equal to the preset similarity threshold, and the sequence similarity between the feature sequence and the fault sequence is high, then the feature sequence is retained.

[0096] If the sequence similarity is less than the preset similarity threshold, the sequence similarity between the feature sequence and the fault sequence is low. In this case, the feature sequence is deleted to reduce the amount of invalid data. This not only reduces the amount of computation but also improves the accuracy of fault detection.

[0097] The car door is divided into multiple opening and closing stages. According to the opening and closing stages, the characteristic sequences of the selected time period covering the maintenance cycle are divided into multiple trend segments. That is, the trend segments are the characteristic sequences located within each opening and closing stage.

[0098] If the maintenance cycle is 14 days, then the time period is 17 days.

[0099] By traversing each trend segment, trend characteristics can be calculated within each trend segment.

[0100] For example, trend characteristics include at least one of the following:

[0101] 1. Time series trend: The direction and magnitude of changes in the basic characteristics of different stages of a single door opening and closing within the sample period.

[0102] 2. Growth / Decline Rate: The rate of change of the basic characteristics at different stages of a single door opening and closing within the sample period.

[0103] 3. Volatility: The degree of fluctuation of the basic characteristics of different stages of a single door opening and closing within the sample period, above and below the trend line.

[0104] 4. Inflection Point: The point at which the trend of the basic characteristics at different stages of a single door opening and closing changes direction within the sample period.

[0105] In this embodiment, a sliding window approach is used to extract basic features from the target operating data, and then trend features are extracted from the basic features. This feature fusion technology for car door motors can effectively help predict the probability of long-cycle door opening and closing failures.

[0106] Step 104: Perform self-verification of trend features based on the fault status to filter out trend features generated when users interfere with the car door.

[0107] In this embodiment, there are significant differences between the trend characteristics of car door malfunctions when there is user interference and the trend characteristics of car door malfunctions when there is no user interference.

[0108] For example, such as Figure 3 As shown, when the car door is currently being opened or closed, a fault is reported without user interference. The fault code is E001 (door circuit fault). As indicated by the arrow, the trend of previous opening and closing operations shows an upward trend, forming an abnormal range. Figure 4As shown, when the car door is currently being opened or closed, a fault is reported if there is user interference (such as obstructing the car door from closing). The fault code is E001 (door circuit fault). As indicated by the arrow, there is no change in the trend characteristics of other previous opening and closing operations.

[0109] Therefore, trend features can be self-verified based on fault status, and user interference can be filtered out from all trend features. Trend features generated when the car door malfunctions make the trend features more predictable.

[0110] In the specific implementation, if the current door opening / closing operation is in a fault state, then before the current door opening / closing operation, a specified number of other door opening / closing operations are searched.

[0111] Iterate through all other door opening and closing operations. If the trend characteristics corresponding to other door opening and closing operations change (such as rising, falling, etc.), then add a second quantity to the other door opening and closing operations.

[0112] Upon completion of the traversal, the second quantity is compared with a preset quantity threshold. The first quantity and its threshold are generated by dividing the trend features corresponding to historical door opening and closing operations into trends when there is user interference with the car door and trends when there is no user interference, using unsupervised clustering.

[0113] If the second quantity is greater than or equal to the preset quantity threshold, it is determined that there is no user interference with the car door during the current door opening and closing operation, and the trend characteristics corresponding to the current door opening and closing operation are retained.

[0114] If the second quantity is less than the preset quantity threshold, it is determined that the current door opening and closing operation is interfering with the car door, and the trend features corresponding to the current door opening and closing operation are filtered out.

[0115] Step 105: If self-verification is completed, train the fault detection model for the car door motor based on trend characteristics and fault status, so that the fault detection model can be used to calculate the total probability of the car door motor experiencing a fault status in the next time period.

[0116] During the self-verification process, the fault detection model of the car door motor configuration is trained in a supervised manner using the remaining trend features as samples and the fault state as a label. This enables the fault detection model of the car door motor configuration to calculate the total probability of the car door motor experiencing a fault state in the next time period.

[0117] The fault detection model can be a machine learning model, such as Support Vector Machine (SVM), decision tree, random forest, etc., or a deep learning model, such as Convolutional Neural Networks (CNN), Long Short Term Memory (LSTM), Transformer (a neural network architecture based on self-attention mechanism), etc.

[0118] For deep learning models, the structure of fault detection models is not limited to manually designed neural networks. It can also be a neural network optimized by model quantization methods, a neural network searched for the characteristics of the car door motor by NAS (Neural Architecture Search) methods, and so on. This embodiment does not impose any restrictions on this.

[0119] In one design, the car door can be divided into multiple opening and closing stages. According to the opening and closing stages, each trend feature is divided into multiple trend segments. The trend segments are marked with fault status, that is, the trend segments are marked with 1 (fault exists) or 0 (fault does not exist).

[0120] Within each time period (e.g., 17 days), a dataset is constructed from trend segments under the same opening and closing phase. The dataset is then divided into a training set, a validation set, and a test set in a 6:2:2 ratio.

[0121] Without distinguishing between fault types, the sub-probabilities of fault states are statistically analyzed in each dataset, which can improve the generalization of the fault detection model.

[0122] Wherein, the sub-probability of the fault state = the number of times the fault state is 1 (i.e., a fault exists) / the total number of effective opening and closing times of the door.

[0123] Random forest models are trained separately for each dataset and each sub-probability to form a fault detection model for the car door motor. That is, the fault detection model contains multiple random forest models, and the number of random forest models is the same as the number of opening and closing stages of the car door.

[0124] The random forest model has two parameters: numTrees (the number of decision trees in the forest) and maxDepth (the maximum depth of a single decision tree). To reduce the risk of overfitting, a grid search can be performed to select the optimal settings for the number of trees and the maximum depth.

[0125] In this embodiment, when the elevator car door performs the opening and closing operation, raw operating data and fault status of the car door motor are collected. Preprocessing is performed on the raw operating data based on the opening and closing stroke of the car door to obtain target operating data. Trend features related to the fault status are extracted from the target operating data. Self-verification is performed on the trend features to filter out trend features generated when users interfere with the car door. If self-verification is completed, a fault detection model is trained on the car door motor based on the trend features and fault status, so that the fault detection model can be used to calculate the total probability of the car door motor experiencing a fault status in the next time period. This embodiment, by preprocessing based on the characteristics of the car door and eliminating user interference, can effectively improve the quality of trend features for fault prediction, improve fault predictability, and improve the generalization of fault detection using the fault rate over a time period, thus comprehensively improving the accuracy of fault detection and enhancing the guidance significance for elevator maintenance.

[0126] Example 2

[0127] See Figure 5 The diagram illustrates a flowchart of a fault detection method provided in Embodiment 2 of the present invention. This method can be executed by a fault detection device. The training of the fault detection model and the fault detection device can be implemented in hardware and / or software. The fault detection device can be configured in electronic devices, particularly elevator controllers. Figure 5 As shown, the method includes:

[0128] Step 501: Load the fault detection model.

[0129] In this embodiment, the fault detection model trained according to the method of Embodiment 1 can be loaded and run locally.

[0130] Step 502: When the elevator car door is opening or closing, collect the original operating data and fault status of the car door motor.

[0131] Step 503: Perform preprocessing on the original operating data based on the opening and closing stroke of the car door to obtain the target operating data.

[0132] In practice, at least one of the following preprocessing operations—noise reduction, anomaly removal, and interpolation—can be performed on the original operating data based on the opening and closing stroke of the car door to obtain the target operating data.

[0133] The noise reduction operation includes:

[0134] The car door is divided into multiple opening and closing stages; the operating segment data under each opening and closing stage is marked in the original operating data; and bandpass filtering is performed independently on each operating segment data.

[0135] The anomaly removal operation includes:

[0136] Query historical operating data; historical operating data is the operating data collected from the car door motor when the car door was opened and closed in the past; construct abnormal intervals based on the changing trend of historical operating data; add abnormal intervals to the original operating data that match the changing trend of the original operating data, and remove operating parameters that are outside the abnormal intervals from the original operating data.

[0137] Interpolation operations include:

[0138] When an operating parameter is missing at a certain location in the original operating data, a preset operating mechanism model is queried. The operating mechanism model is the standard operating data presented by the elevator car door motor when the elevator car door performs the opening and closing operation. The operating similarity between the original operating data and the operating mechanism model is calculated. The operating parameter is inserted at the location in the original operating data to increase the operating similarity.

[0139] Step 504: Extract trend features related to the fault state from the target operating data.

[0140] In the specific implementation, a window matching the fault state is added to the target operating data; the width of the window is positively correlated with the duration of the fault state, and the width of the window is greater than the duration of the fault state; each time the window is slid, various basic features are calculated for various operating parameters within the window; various basic features of the same operating parameter are concatenated into a feature sequence; the car door is divided into multiple opening and closing stages; each feature sequence is divided into multiple trend segments according to the opening and closing stages; and trend features are calculated in each trend segment.

[0141] Furthermore, the fault states are concatenated into a fault sequence according to time sequence; the sequence similarity between the feature sequence and the fault sequence is calculated; if the sequence similarity is greater than or equal to a preset similarity threshold, the feature sequence is retained; if the sequence similarity is less than the preset similarity threshold, the feature sequence is deleted.

[0142] Step 505: Perform self-verification on trend features to filter out trend features generated when users interfere with the car door.

[0143] In the specific implementation, if a fault occurs in the current door opening / closing operation, a specified first number of other door opening / closing operations are searched before the current operation. If the trend characteristics corresponding to the other door opening / closing operations change, a second number is added to the other door opening / closing operations. If the second number is greater than or equal to a preset threshold, it is determined that there is no user interference with the car door in the current door opening / closing operation, and the trend characteristics corresponding to the current door opening / closing operation are retained. If the second number is less than the preset threshold, it is determined that there is user interference with the car door in the current door opening / closing operation, and the trend characteristics corresponding to the current door opening / closing operation are filtered out.

[0144] Step 506: If self-verification is completed, input the trend characteristics within the current time period into the fault detection model to calculate the total probability of the car door motor failing in the next time period.

[0145] When completing the self-verification, the trend characteristics within the current time period (e.g., the current 17 days) can be input into the fault detection model to calculate the total probability of the car door motor failing in the next time period (e.g., the next 17 days).

[0146] When the total probability is greater than or equal to the preset fault threshold, the car door can be included in the maintenance list, so that maintenance personnel will pay attention to the car door when they inspect the elevator next time, thereby reducing the chance of the car door failing in the next time period.

[0147] In the specific implementation, the fault detection model includes multiple random forest models trained for the door opening and closing stages. Therefore, the car door can be divided into multiple door opening and closing stages, and the trend characteristics within the current time period can be divided into multiple trend segments according to the door opening and closing stages.

[0148] Each trend segment is input into the corresponding random forest model to calculate the sub-probability of the car door motor failing at each opening and closing stage in the next time period.

[0149] Based on the importance of the door opening and closing phases, the individual probabilities are merged into the total probability of the car door motor malfunctioning in the next time period.

[0150] Generally, weights representing the importance of each door opening and closing stage can be assigned. For example, the weights of the door opening constant speed stage, door opening hold stage, door closing constant speed stage, and door closing hold stage are relatively low, while the weights of the door opening acceleration stage, door opening deceleration stage, door closing acceleration stage, and door closing deceleration stage are relatively high. For each door opening and closing stage, the product of the sub-probability and the corresponding weight is summed to obtain the total probability of the car door motor malfunctioning.

[0151] In this embodiment, a fault detection model is loaded; when the elevator car door performs an opening and closing operation, raw operating data of the car door motor is collected; the raw operating data is preprocessed according to the opening and closing stroke of the car door to obtain target operating data; trend features related to the fault state are extracted from the target operating data; the trend features are self-verified to filter out trend features generated when the car door is disturbed by the user; if the self-verification is completed, the trend features in the current time period are input into the fault detection model to calculate the total probability of the car door motor malfunctioning in the next time period. This embodiment preprocesses based on the characteristics of the car door and eliminates user interference with the car door, which can effectively improve the quality of trend features for fault prediction, improve the predictability of faults, and improve the generalization of fault detection by the fault rate of a time period, which can comprehensively improve the accuracy of fault detection and enhance the guiding significance of elevator maintenance.

[0152] Example 3

[0153] See Figure 6 The diagram shows a structural schematic of a training device for a fault detection model provided in Embodiment 3 of the present invention. Figure 6 As shown, the device includes:

[0154] The data acquisition module 601 is used to collect raw operating data and fault status of the elevator car door motor when the elevator car door performs the opening and closing operation;

[0155] Preprocessing module 602 is used to preprocess the original operating data based on the opening and closing stroke of the car door to obtain target operating data;

[0156] Trend feature extraction module 603 is used to extract trend features related to the fault state from the target operating data;

[0157] The trend feature self-verification module 604 is used to self-verify the trend feature based on the fault state, so as to filter out the trend feature generated when the user interferes with the car door;

[0158] The fault detection model training module 605 is used to train a fault detection model for the car door motor based on the trend characteristics and the fault state if self-verification is completed, so that the fault detection model can be used to calculate the total probability of the car door motor experiencing a fault state in the next time period.

[0159] In one embodiment of the present invention, the preprocessing module 602 includes:

[0160] The target operation data survival module is used to perform at least one of the following preprocessing operations on the original operation data based on the opening and closing stroke of the car door: noise reduction, anomaly removal, and interpolation, to obtain the target operation data.

[0161] The noise reduction operation includes:

[0162] The car door is divided into multiple opening and closing stages;

[0163] The original operational data is used to mark operational segment data under each of the door opening and closing stages;

[0164] Bandpass filtering is performed independently on each of the aforementioned runtime data segments;

[0165] The anomaly removal operation includes:

[0166] Query historical operating data; the historical operating data refers to the operating data collected from the motor of the car door when the car door was historically performing opening and closing operations;

[0167] Anomaly intervals are constructed based on the changing trends of the historical operational data;

[0168] Add abnormal intervals that match the changing trend of the original operating data to the original operating data, and remove operating parameters that are outside the abnormal intervals from the original operating data;

[0169] The interpolation operation includes:

[0170] When an operating parameter is missing at a certain location in the original operating data, a preset operating mechanism model is queried; the operating mechanism model is the standard operating data presented by the motor of the elevator car door when the elevator car door performs the opening and closing operation;

[0171] Calculate the operational similarity between the original operational data and the operational mechanism model;

[0172] Insert running parameters at the specified positions in the original running data to increase the running similarity.

[0173] In one embodiment of the present invention, the trend feature extraction module 603 includes:

[0174] A window adding module is used to add a window matching the fault state to the target running data; the width of the window is positively correlated with the duration of the fault state, and the width of the window is greater than the duration of the fault state;

[0175] The basic feature extraction module is used to calculate various basic features for various operating parameters within the window each time the window is slid;

[0176] The feature sequence splicing module is used to splice various basic features of the same operating parameter into a feature sequence;

[0177] The door opening / closing phase division module is used to divide the car door into multiple door opening / closing phases;

[0178] A trend segmentation module is used to segment each of the feature sequences into multiple trend segments according to the door opening and closing stages;

[0179] A trend feature calculation module is used to calculate trend features in each of the trend segments.

[0180] In one embodiment of the present invention, the trend feature extraction module 603 further includes:

[0181] The fault sequence splicing module is used to splice the fault states into a fault sequence according to time order;

[0182] A sequence similarity calculation module is used to calculate the sequence similarity between the feature sequence and the fault sequence;

[0183] The feature sequence retention module is used to retain the feature sequence if the sequence similarity is greater than or equal to a preset similarity threshold.

[0184] The feature sequence deletion module is used to delete the feature sequence if the sequence similarity is less than a preset similarity threshold.

[0185] In one embodiment of the present invention, the trend feature self-verification module 604 includes:

[0186] The door opening / closing operation lookup module is used to look up a specified number of other door opening / closing operations before the current door opening / closing operation if the current door opening / closing operation is in the fault state.

[0187] The quantity accumulation module is used to accumulate a second quantity for other door opening and closing operations if the trend characteristics corresponding to other door opening and closing operations change;

[0188] The trend feature retention module is used to determine that there is no user interference with the car door in the current door opening and closing operation if the second quantity is greater than or equal to a preset quantity threshold, and to retain the trend feature corresponding to the current door opening and closing operation.

[0189] The trend feature filtering module is used to determine that if the second quantity is less than a preset quantity threshold, the current door opening and closing operation is interfering with the car door, and to filter out the trend feature corresponding to the current door opening and closing operation.

[0190] In one embodiment of the present invention, the fault detection model training module 605 includes:

[0191] The door opening / closing phase division module is used to divide the car door into multiple door opening / closing phases;

[0192] A trend segmentation module is used to segment each of the trend features into multiple trend segments according to the door opening and closing stages;

[0193] The fault status labeling module is used to label the fault status of the trend segment;

[0194] A dataset construction module is used to construct datasets from the trend segments under the same door opening and closing phase within various time periods;

[0195] The sub-probability statistics module is used to calculate the sub-probability of the fault state in each of the datasets.

[0196] The random forest model training module is used to train random forest models independently based on each dataset and each sub-probability to form a fault detection model for the car door motor, so that the fault detection model can be used to calculate the total probability of the car door motor malfunctioning in the next time period.

[0197] The training device for the fault detection model provided in this embodiment of the invention can execute the training method for the fault detection model provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing the training method for the fault detection model.

[0198] Example 4

[0199] See Figure 7 The diagram shows a structural schematic of a fault detection device provided in Embodiment 4 of the present invention. Figure 7 As shown, the device includes:

[0200] The fault detection model loading module 701 is used to load the fault detection model trained according to the method described in Embodiment 1;

[0201] The data acquisition module 702 is used to collect raw operating data and fault status of the elevator car door motor when the elevator car door performs the opening and closing operation;

[0202] Preprocessing module 703 is used to preprocess the original operating data based on the opening and closing stroke of the car door to obtain target operating data;

[0203] Trend feature extraction module 704 is used to extract trend features related to fault status from the target operating data;

[0204] The trend feature self-verification module 705 is used to perform self-verification of the trend feature based on the fault state, so as to filter out the trend feature generated when the user interferes with the car door;

[0205] The total probability calculation module 706 is used to calculate the total probability of the car door motor failing in the next time period by inputting the trend characteristics in the current time period into the fault detection model if self-verification is completed.

[0206] In one embodiment of the present invention, the fault detection model includes multiple random forest models, and the total probability calculation module 706 includes:

[0207] The door opening / closing phase division module is used to divide the car door into multiple door opening / closing phases;

[0208] A trend segmentation module is used to segment the trend characteristics within the current time period into multiple trend segments according to the door opening and closing stage;

[0209] The trend segment calculation module is used to input each of the trend segments into each of the random forest models to calculate the sub-probability of the motor of the car door in each of the opening and closing stages in the next time period;

[0210] The total probability fusion module is used to fuse the various sub-probabilities into a total probability of the car door motor malfunctioning in the next time period based on the importance of the door opening and closing stage.

[0211] In one embodiment of the present invention, the preprocessing module 703 includes:

[0212] The target operation data survival module is used to perform at least one of the following preprocessing operations on the original operation data based on the opening and closing stroke of the car door: noise reduction, anomaly removal, and interpolation, to obtain the target operation data.

[0213] The noise reduction operation includes:

[0214] The car door is divided into multiple opening and closing stages;

[0215] The original operational data is used to mark operational segment data under each of the door opening and closing stages;

[0216] Bandpass filtering is performed independently on each of the aforementioned runtime data segments;

[0217] The anomaly removal operation includes:

[0218] Query historical operating data; the historical operating data refers to the operating data collected from the motor of the car door when the car door was historically performing opening and closing operations;

[0219] Anomaly intervals are constructed based on the changing trends of the historical operational data;

[0220] Add abnormal intervals that match the changing trend of the original operating data to the original operating data, and remove operating parameters that are outside the abnormal intervals from the original operating data;

[0221] The interpolation operation includes:

[0222] When an operating parameter is missing at a certain location in the original operating data, a preset operating mechanism model is queried; the operating mechanism model is the standard operating data presented by the motor of the elevator car door when the elevator car door performs the opening and closing operation;

[0223] Calculate the operational similarity between the original operational data and the operational mechanism model;

[0224] Insert running parameters at the specified positions in the original running data to increase the running similarity.

[0225] In one embodiment of the present invention, the trend feature extraction module 704 includes:

[0226] A window adding module is used to add a window matching the fault state to the target running data; the width of the window is positively correlated with the duration of the fault state, and the width of the window is greater than the duration of the fault state;

[0227] The basic feature extraction module is used to calculate various basic features for various operating parameters within the window each time the window is slid;

[0228] The feature sequence splicing module is used to splice various basic features of the same operating parameter into a feature sequence;

[0229] The door opening / closing phase division module is used to divide the car door into multiple door opening / closing phases;

[0230] A trend segmentation module is used to segment each of the feature sequences into multiple trend segments according to the door opening and closing stages;

[0231] A trend feature calculation module is used to calculate trend features in each of the trend segments.

[0232] In one embodiment of the present invention, the trend feature extraction module 704 further includes:

[0233] The fault sequence splicing module is used to splice the fault states into a fault sequence according to time order;

[0234] A sequence similarity calculation module is used to calculate the sequence similarity between the feature sequence and the fault sequence;

[0235] The feature sequence retention module is used to retain the feature sequence if the sequence similarity is greater than or equal to a preset similarity threshold.

[0236] The feature sequence deletion module is used to delete the feature sequence if the sequence similarity is less than a preset similarity threshold.

[0237] In one embodiment of the present invention, the trend feature self-verification module 705 includes:

[0238] The door opening / closing operation lookup module is used to look up a specified number of other door opening / closing operations before the current door opening / closing operation if the current door opening / closing operation is in the fault state.

[0239] The quantity accumulation module is used to accumulate a second quantity for other door opening and closing operations if the trend characteristics corresponding to other door opening and closing operations change;

[0240] The trend feature retention module is used to determine that there is no user interference with the car door in the current door opening and closing operation if the second quantity is greater than or equal to a preset quantity threshold, and to retain the trend feature corresponding to the current door opening and closing operation.

[0241] The trend feature filtering module is used to determine that if the second quantity is less than a preset quantity threshold, the current door opening and closing operation is interfering with the car door, and to filter out the trend feature corresponding to the current door opening and closing operation.

[0242] The fault detection device provided in this embodiment of the invention can execute the fault detection method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing the fault detection method.

[0243] Example 5

[0244] See Figure 8 This diagram illustrates a structural schematic of an electronic device according to an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, blade servers, mainframe computers, and other suitable computers. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0245] like Figure 8As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0246] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0247] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as training a fault detection model and its fault detection methods.

[0248] In some embodiments, the training of the fault detection model and the fault detection method thereof may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the training of the fault detection model and the fault detection method thereof described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the training of the fault detection model and the fault detection method thereof by any other suitable means (e.g., by means of firmware).

[0249] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0250] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0251] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0252] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0253] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0254] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0255] Example 6

[0256] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the training of a fault detection model and a fault detection method as provided in any embodiment of this invention.

[0257] In implementing the computer program product, computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0258] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0259] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A training method for a fault detection model, characterized in that, include: When the elevator car door is opening and closing, the original operating data and fault status of the car door motor are collected. Based on the opening and closing stroke of the car door, at least one of the following preprocessing operations—noise reduction, anomaly removal, and interpolation—is performed on the original operating data to obtain the target operating data; Extract trend features related to the fault state from the target operating data; The trend characteristics are self-verified based on the fault status to filter out the trend characteristics generated when the user interferes with the car door; If self-verification is completed, a fault detection model for the car door motor is trained based on the trend characteristics and the fault state, so that the fault detection model can be used to calculate the total probability of the car door motor experiencing a fault state in the next time period. The noise reduction operation includes: The car door is divided into multiple opening and closing stages; The original operational data is used to mark operational segment data under each of the door opening and closing stages; Bandpass filtering is performed independently on each of the aforementioned runtime data segments; The anomaly removal operation includes: Query historical operating data; the historical operating data refers to the operating data collected from the motor of the car door when the car door was historically performing opening and closing operations; Anomaly intervals are constructed based on the changing trends of the historical operational data; Add abnormal intervals that match the changing trend of the original operating data to the original operating data, and remove operating parameters that are outside the abnormal intervals from the original operating data; The interpolation operation includes: When an operating parameter is missing at a certain location in the original operating data, a preset operating mechanism model is queried; the operating mechanism model is the standard operating data presented by the motor of the elevator car door when the elevator car door performs the opening and closing operation; Calculate the operational similarity between the original operational data and the operational mechanism model; Insert running parameters at the specified positions in the original running data to increase the running similarity.

2. The method according to claim 1, characterized in that, Extracting trend features related to the fault state from the target operating data includes: A window matching the fault state is added to the target running data; the width of the window is positively correlated with the duration of the fault state, and the width of the window is greater than the duration of the fault state. Each time the window is slid, various basic features are calculated for the various operating parameters within the window; The various basic features of the same operating parameter are concatenated into a feature sequence; The car door is divided into multiple opening and closing stages; Each of the aforementioned feature sequences is divided into multiple trend segments according to the door opening and closing stages; Calculate trend features in each of the trend segments.

3. The method according to claim 2, characterized in that, The step of extracting trend features related to the fault state from the target operating data further includes: The fault states are concatenated into a fault sequence in chronological order; Calculate the sequence similarity between the feature sequence and the fault sequence; If the sequence similarity is greater than or equal to a preset similarity threshold, then the feature sequence is retained; If the sequence similarity is less than a preset similarity threshold, the feature sequence is deleted.

4. The method according to claim 1, characterized in that, The step of self-checking the trend characteristics based on the fault state to filter out the trend characteristics generated when the user interferes with the car door includes: If the current door opening / closing operation experiences the fault state, then before the current door opening / closing operation, search for a specified first number of other door opening / closing operations; If the trend characteristic corresponding to the other door opening and closing operations changes, then the second quantity is added to the other door opening and closing operations; If the second quantity is greater than or equal to the preset quantity threshold, it is determined that there is no user interference with the car door in the current door opening and closing operation, and the trend feature corresponding to the current door opening and closing operation is retained; If the second quantity is less than a preset quantity threshold, it is determined that the current door opening and closing operation is interfering with the car door, and the trend feature corresponding to the current door opening and closing operation is filtered out.

5. The method according to any one of claims 1-4, characterized in that, The step of training a fault detection model for the car door motor based on the trend characteristics and the fault state, so that the fault detection model can be used to calculate the total probability of the car door motor experiencing a fault state in the next time period, includes: The car door is divided into multiple opening and closing stages; Each of the aforementioned trend characteristics is divided into multiple trend segments according to the door opening and closing stages; The fault status is labeled on the trend segment; Within each time period, a dataset is constructed from the trend segments under the same door opening and closing phase; Calculate the sub-probabilities of the fault states in each of the datasets; Random forest models are trained separately based on each dataset and each sub-probability to form a fault detection model for the car door motor, which is then used to calculate the total probability of the car door motor malfunctioning in the next time period.

6. A fault detection method, characterized in that, include: Load the fault detection model trained by the method according to any one of claims 1-5; When the elevator car door is opening and closing, the original operating data and fault status of the car door motor are collected. The original operating data is preprocessed based on the opening and closing stroke of the car door to obtain the target operating data; Extract trend features related to fault states from the target operational data; The trend characteristics are self-verified based on the fault status to filter out the trend characteristics generated when the user interferes with the car door; If self-verification is completed, the trend characteristics within the current time period are input into the fault detection model to calculate the total probability of the car door motor malfunctioning in the next time period.

7. The method according to claim 6, characterized in that, The fault detection model includes multiple random forest models. The step of inputting the trend features within the current time period into the fault detection model to calculate the total probability of the car door motor malfunctioning in the next time period includes: The car door is divided into multiple opening and closing stages; The trend characteristics within the current time period are divided into multiple trend segments according to the opening and closing stage; Each of the trend segments is input into each of the random forest models to calculate the sub-probability of the car door motor failing in each of the opening and closing stages in the next time period; Based on the importance of the door opening and closing phase, the individual sub-probabilities are merged into the total probability of the car door motor malfunctioning in the next time period.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the training method of the fault detection model as described in any one of claims 1-5 or the fault detection method as described in any one of claims 6-7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the training method for the fault detection model as described in any one of claims 1-5 or the fault detection method as described in any one of claims 6-7.

Images