Elevator acceleration fault detection method and device based on improved meta-learning

Abstract

The present invention discloses an elevator acceleration fault detection method and device based on improved meta-learning, comprising: obtaining and preprocessing the original acceleration response data sequence of a target elevator; randomly selecting a support set and a query set after preprocessing; constructing and training an acceleration fault detection model; extracting features from the support set and the query set; performing prototype calculation on the extracted features to obtain a loss, wherein an intra-class distance regularization term is introduced in the prototype calculation; obtaining a difficult sample library; testing the acceleration fault detection model on a test set to obtain the model's accuracy; adjusting the support set and the query set based on the accuracy; repartitioning the adjusted support set and query set based on the difficult sample library; and retraining the acceleration fault detection model using the retrained acceleration fault detection model. This method solves the technical problems of few-sample generalization and class imbalance in related technologies.

Description

Technical Field

[0001] The present application relates to the technical field of elevator acceleration fault detection, and in particular to an elevator acceleration fault detection method and device based on improved meta-learning. Background Art

[0002] There are three major challenges in implementing intelligent fault diagnosis methods based on machine learning or deep learning. First, it is often difficult to obtain sufficient fault data in actual elevator scenarios, and training deep learning fault diagnosis models typically requires a large amount of labeled data, which limits the practical application of such methods. Second, existing elevator fault diagnosis methods based on small samples of elevator operation data under complex conditions require manual intervention in the configuration of support and query sets. For example, understanding relevant meta-learning methods and comparing 1-shot, 5-shot, and 10-shot experiments are required to obtain results, which complicates the experiments and limits the model's generalization ability. Finally, the performance of the prototype network depends heavily on the number and quality of samples in the support set, and can be significantly affected by class-imbalanced datasets. Summary of the Invention

[0003] The purpose of the embodiments of the present application is to provide an elevator acceleration fault detection method and device based on improved meta-learning to solve the technical problems of few-sample generalization and category imbalance in related technologies.

[0004] According to a first aspect of an embodiment of the present application, a method for detecting elevator acceleration faults based on improved meta-learning is provided, comprising:

[0005] Obtaining the original acceleration response data sequence of the target elevator;

[0006] Preprocessing the original acceleration response data sequence to obtain a two-dimensional image containing time information;

[0007] randomly selecting a support set and a query set from the two-dimensional image;

[0008] Constructing and training an accelerated fault detection model based on the support set and the query set, wherein the accelerated fault detection model is used to extract features from the support set and the query set, and then perform prototype calculation on the extracted features to obtain a loss, wherein an intra-class distance regularization term is introduced in the prototype calculation;

[0009] Using the trained accelerated fault detection model to perform classification prediction, when a sample is misclassified, it is added to the difficult library of the corresponding category to obtain a difficult sample library, wherein the query set consists of the difficult sample library and common samples;

[0010] Testing the trained accelerated fault detection model using a test set to obtain the accuracy of the model, and adjusting the support set and the query set based on the accuracy;

[0011] Re-dividing the adjusted support set and query set according to the difficult sample library, and retraining the accelerated fault detection model using the re-divided support set and query set to predict the operating state of the elevator;

[0012] The preprocessed acceleration response data sequence is input into the retrained acceleration fault detection model to obtain the fault detection results.

[0013] According to a second aspect of an embodiment of the present application, there is provided an elevator acceleration fault detection device based on improved meta-learning, comprising:

[0014] An acquisition module, used for acquiring an original acceleration response data sequence of a target elevator;

[0015] A preprocessing module, configured to preprocess the original acceleration response data sequence to obtain a two-dimensional image containing time information;

[0016] A data set selection module, configured to randomly select a support set and a query set from the two-dimensional image;

[0017] a model training module, configured to construct and train an accelerated fault detection model based on the support set and the query set, wherein the accelerated fault detection model is configured to extract features from the support set and the query set, and then perform prototype calculation on the extracted features to obtain a loss, wherein an intra-class distance regularization term is introduced in the prototype calculation;

[0018] A sample acquisition module is used to perform classification prediction using the trained accelerated fault detection model. When a sample is misclassified, it is added to the difficult library of the corresponding category to obtain a difficult sample library. The query set consists of the difficult sample library and common samples.

[0019] A data set adjustment module is used to test the trained accelerated fault detection model using a test set to obtain the accuracy of the model, and adjust the support set and query set according to the accuracy;

[0020] a model retraining module, configured to re-divide the adjusted support set and query set according to the difficult sample library, and retrain the accelerated fault detection model using the re-divide support set and query set to predict the operating status of the elevator;

[0021] The fault detection module is used to input the preprocessed acceleration response data sequence into the retrained acceleration fault detection model to obtain the fault detection result.

[0022] According to a third aspect of the embodiments of the present application, there is provided an electronic device, including:

[0023] one or more processors;

[0024] a memory for storing one or more programs;

[0025] When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in the first aspect.

[0026] According to a third aspect of an embodiment of the present application, a computer-readable storage medium is provided, on which computer instructions are stored. When the instructions are executed by a processor, the steps of the method described in the first aspect are implemented.

[0027] The technical solutions provided by the embodiments of the present application may have the following beneficial effects:

[0028] As can be seen from the above embodiments, the present application innovatively integrates three core technologies: dynamic meta-training mechanism, adaptive sample scheduling, and regularized prototype network, and systematically solves key problems such as scarcity of fault samples, imbalanced category distribution, and insufficient generalization ability of diagnostic models in industrial scenarios. The trained model can accurately identify high-risk faults such as sudden stops and severe vibrations of elevators under extreme conditions with only a small number of samples / classes, and output high-reliability fault classification results, meeting the stringent requirements of industrial intelligent operation and maintenance systems for fault diagnosis under complex working conditions. This solution not only breaks through the performance bottleneck of traditional deep learning models in small sample scenarios, but also forms a systematic technological breakthrough in dynamic task construction, difficult sample mining, unsupervised feature alignment, etc., significantly improving the practicality and engineering capabilities of elevator fault diagnosis in data-scarce environments.

[0029] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0031] Figure 1 The present invention is a flowchart of an elevator acceleration fault detection method based on improved meta-learning according to an exemplary embodiment.

[0032] Figure 2 It is a general framework diagram of an elevator acceleration fault detection method based on improved meta-learning according to an exemplary embodiment.

[0033] Figure 3is a diagram showing the length distribution of original acceleration response data sequences according to an exemplary embodiment.

[0034] Figure 4 4 is a diagram showing the number of categories of a unified 400-length acceleration fault data set according to an exemplary embodiment.

[0035] Figure 5 FIG. 4 is a diagram showing a network model structure of MetaRes-DMT-AS according to an exemplary embodiment.

[0036] Figure 6 FIG. 4 is a diagram showing an iterative process of comparing various models according to an exemplary embodiment.

[0037] Figure 7 3 is a comparison chart showing the running time, accuracy and model parameter quantity of each model according to an exemplary embodiment.

[0038] Figure 8 is a confusion matrix diagram of each model according to an exemplary embodiment.

[0039] Figure 9 The present invention is a block diagram of an elevator acceleration fault detection device based on improved meta-learning according to an exemplary embodiment. DETAILED DESCRIPTION

[0040] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0041] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. As used in this application and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0042] It should be understood that although the terms first, second, third, etc. may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0043] like Figure 1 and Figure 2 As shown, an embodiment of the present invention provides an elevator acceleration fault detection method based on improved meta-learning, which may include the following steps:

[0044] S1: Obtain the original acceleration response data sequence of the target elevator;

[0045] Specifically, the experimental environment was set up in elevators in daily use areas such as residential communities and commercial buildings. A gyroscope sensor installed above the elevator captured the three-axis acceleration of the elevator during operation, capturing a total of 16,794 operation processes. The complete operation process was automatically captured during the operation of the elevator, i.e., a complete operation process was considered as one process from the elevator door closing, the elevator running, to the elevator door opening. Five categories of raw acceleration response data sequences were screened using platform detection data (elevator reset operation, severe car vibration, emergency stop failure, normal operation, and overspeed). After obtaining the raw acceleration response data sequence, it was also cleaned, as errors occasionally occur during the elevator data collection process, resulting in data anomalies (0 and excessive acceleration). This was done to ensure data quality.

[0046] In one embodiment, the cleaning is to remove the running process containing incomplete and missing data in the data. In addition, the time length corresponding to each running state is different, such as Figure 3 As shown, most of the overspeed intervals are between 400 and 600, most of the reset operation intervals are above 800, most of the emergency stop intervals are between 100 and 400, most of the violent car oscillation intervals are between 100 and 400, and most of the normal operation intervals are between 100 and 400. Finally, the most corresponding intervals in each different operating state are taken and processed with a uniform length of 400. The insufficient intervals are interpolated with cubic spline, and the excessive intervals are extracted.

[0047] After processing, it was found that Figure 4 As shown in the figure, the data of emergency stop and violent car oscillation are far less than 1,500 for each of the other three types, with only 157 and 116 respectively.

[0048] S2: Preprocessing the original acceleration response data sequence to obtain a two-dimensional image containing time information; this step includes the following sub-steps:

[0049] S21: performing normalization processing on the original acceleration response data sequence;

[0050] Specifically, after cleaning the data, Z-score normalization is performed on the one-dimensional raw acceleration response data sequence. This involves subtracting the mean and dividing by the variance of each data point in the data sequence, so that the processed data approximately conforms to a standard normal distribution (0, 1). Before training a neural network, Z-score normalization is often performed on the input features to ensure a consistent distribution for each feature, making the optimization algorithm more stable during gradient descent and less likely to fall into local optima.

[0051] S22: converting the normalized acceleration response data sequence into polar coordinate representation, where the acceleration value corresponds to the cosine value of the angle and the timestamp corresponds to the radius;

[0052] Specifically, based on the Gram matrix, the angles between vectors at different times are calculated and converted to values ​​between 0 and 1, thereby generating a new angle matrix. This converts the one-dimensional time series into a two-dimensional representation containing spatiotemporal information, essentially converting the normalized values ​​into polar coordinate angles, preserving the complete information of the original signal. This step can amplify fault characteristics: abnormal shocks (such as sudden stops) can cause sudden changes in amplitude, which manifest as sharp corners in the polar coordinate trajectory.

[0053] S23: Use the Gram matrix to reconstruct the acceleration response data sequence represented in polar coordinates into a two-dimensional image.

[0054] Specifically, the angle matrix is ​​used as the pixel values ​​of the image to generate a two-dimensional image. Each pixel value in the image corresponds to the cosine value of the angle between different moments in the time series data, thus preserving the temporal information and dynamic characteristics of the data. After converting the one-dimensional data into a two-dimensional image, image analysis methods can be easily used to extract features, such as using convolutional neural networks (CNNs) for feature extraction, to identify patterns in the time series data.

[0055] S3: randomly selecting a support set and a query set from the two-dimensional image;

[0056] Specifically, the two-dimensional image dataset is divided into a training set and a test set with a ratio of 8:2. The support set and query set are initialized to 1 from the divided training set. That is, one data set of each category is selected for training and one data set is selected for testing.

[0057] S4: constructing and training an accelerated fault detection model based on the support set and the query set, wherein the accelerated fault detection model is used to extract features from the support set and the query set, and then perform prototype calculation on the extracted features to obtain a loss, wherein an intra-class distance regularization term is introduced in the prototype calculation;

[0058] Specifically, refer to Figure 5 ,MetaRes-DMT-AS is selected as the backbone of the PNML model.,After the ResNet-18 feature extraction, the features input to the ,prototype network are used for prototype calculation.

[0059] The loss is expressed as follows:

[0060] ;

[0061] Where L is the loss, is the cross entropy loss (Cross-EntropyLoss), is the regularization coefficient and is set to 0.1, N is the number of categories, K is the number of support samples for each category, and the kth sample feature of category c is is the prototype of category c.

[0062] First, for each category c, calculate its class center , then inputs the query set sample features x and class prototypes, calculates the probability that sample x belongs to class c, and then calculates the cross-entropy loss of all query set samples and the average squared Euclidean distance between the support set samples and the class center. Finally, the weighted combined total loss is calculated. This loss function reduces the distance between the query sample and the true class prototype and increases the distance to other class prototypes, ultimately accelerating meta-training convergence.

[0063] S5: Use the trained accelerated fault detection model to perform classification prediction. When a sample is misclassified, add it to the difficult library of the corresponding category to obtain a difficult sample library;

[0064] Specifically, the trained acceleration fault detection model is used for classification prediction. If a sample is misclassified, it is added to the difficult sample library of its true category. The query set Q is composed of the difficult sample library H and common samples to improve the model's ability to identify difficult examples (such as sudden stops and severe vibrations). If an sudden stop sample is predicted to be overspeeding, it is marked as a difficult sample and added to the difficult sample library.

[0065] S6: Testing the accelerated fault detection model using the test set to obtain the accuracy of the model, and adjusting the support set and the query set based on the accuracy;

[0066] Specifically, if the accuracy rate increases or decreases, the support set and query set will be dynamically adjusted. The specific formula is as follows:

[0067] ;

[0068] in is the change in accuracy, p To dynamically adjust parameters, P={S,Q}, S is the support set size, Q is the query set size, t is the current training round, for p The value at round t, for p The value at round t+1, for p The upper bound constraint of for p The lower bound constraint of .

[0069] When accuracy decreases, the support and query sets are expanded to mitigate overfitting. When accuracy improves, the support and query sets are reduced to improve generalization. If accuracy does not improve for t consecutive rounds and the trigger condition is met, parameter adjustments are performed.

[0070] S7: Re-divide the adjusted support set and query set according to the difficult sample library, and retrain the accelerated fault detection model using the re-divide support set and query set; this step includes the following sub-steps:

[0071] S71: allocating the difficult sample library to the adjusted support set and query set, and retraining the accelerated fault detection model;

[0072] Specifically, the support set and query set adjusted in the previous round are used for the next round of training, and parameter adjustment and performance prediction are performed simultaneously to prepare for the next round of optimization.

[0073] First, the emergency stop samples that were incorrectly predicted in the previous round are added to the difficult sample library. When the current training begins, the emergency stop samples in the difficult sample library are preferentially added to the query set for testing after the current training is completed. The support set and query set are then randomly selected for training. Finally, after training is completed, prediction is performed. If the prediction is correct, the difficult sample library is removed. If the prediction fails, the next round of training will be carried out until the prediction is correct.

[0074] S72: After training, update the difficult sample library and perform prediction to obtain accuracy;

[0075] Specifically, each time the obtained accuracy is compared with the accuracy of the previous round to detect whether the threshold is reached.

[0076] First, after the difficult sample library is obtained in the previous round and the current training is completed, the difficult sample library will be predicted first. If the prediction is correct, the difficult sample library will be removed. If the prediction fails, the next round of training will be carried out until the prediction is correct.

[0077] S73: Adjust the support set and the query set according to the accuracy until the iteration is completed.

[0078] Specifically, cyclic training adjusts the support set, query set, and difficult sample library until the iteration is complete and the optimal result is achieved. First, set the number of epochs to 100, and train 100 tasks per epoch, with each task undergoing different training. Performance is checked every T rounds of training (e.g., T = 10), and the accuracy is calculated to determine whether tmodT = 0 has reached the detection point. The accuracy is then compared and the number of consecutive rounds without improvement is recorded. If this reaches 5 times, the threshold is reached. If the accuracy decreases, the support set or query set is increased; if it increases, the support set or query set samples are reduced.

[0079] S8: Input the preprocessed acceleration response data sequence into the retrained acceleration fault detection model to obtain a fault detection result.

[0080] In the experiments using collected elevator acceleration, a total of six models were compared, all using 100 iterations. The accuracy of the comparison experiments is shown in Table 1.

[0081] Table 1:

[0082]

[0083] like Figure 6 As shown in the figure, the convergence characteristics of each model during the training process show significant differences: the MetaRes-DMT-AS proposed in this application achieves the best recognition performance (initial accuracy of 97.85%) in the initial stage, and steadily converges to the optimal accuracy of 98.22% with the iterative process; the WDCNN model shows suboptimal initial performance (96.12%), but there is obvious learning stagnation, and the final accuracy is 96.65%; ViT, WDCNN-GRU and WDCNN-DLSTM have similar convergence trajectories, with an average initial accuracy of 90.3±0.8%, which is increased to 96.44%, 96.65% and 96.97% respectively after iterative optimization; it is worth noting that the CCT model shows a unique overtaking characteristic, with an initial accuracy of only 82.14%, but eventually reaching 97.28%.

[0084] From the perspective of computational efficiency and model complexity ( Figure 7), MetaRes-DMT-AS maintained excellent time efficiency (34 minutes) despite a large parameter count (11.18M), validating the computational optimization of its architectural design. The WDCNN model, with its streamlined parameter count (0.66M) and shortest iteration time (29 minutes), exhibited the best resource efficiency, but its accuracy lagged behind state-of-the-art models by 1.57%. WDCNN's derivatives (WDCNN-DLSTM: 0.86M / 57 minutes, WDCNN-GRU: 0.64M / 57 minutes) and ViT (2.76M / 34 minutes) formed an intermediate performance cluster, with the accuracy difference between the three less than 0.53%. The CCT model exhibited a significant efficiency bottleneck, with its iteration time (203 minutes) consistently ranking first in the experimental group, but its accuracy advantage was only 97.28%.

[0085] In order to explore the robustness of the model in the few-sample scenario, this study uses the confusion matrix ( Figure 8 ) performed a fine-grained analysis of two typical few-shot anomalies (emergency stop and severe car vibration). While MetaRes-DMT-AS maintained a recognition rate of 98.2±0.6% for conventional samples, it achieved recognition rates of 94% and 96% for the few-shot categories, respectively, significantly outperforming the comparison models. Although CCT, WDCNN-GRU, and ViT performed well in multi-shot categories (>95%), their recognition rates for few-shot categories dropped to the 75-84% range, demonstrating their sensitivity to data distribution. WDCNN-DLSTM exhibited significant imbalance, with an emergency stop recognition rate of 91%, but a severe car vibration recognition rate that plummeted to 67%, revealing the limitations of the temporal feature fusion mechanism. The WDCNN model performed poorly in both few-shot tasks (81% and 71%), validating the shortcomings of shallow architectures in learning complex patterns.

[0086] It can be seen from the above embodiments that this application constructs a meta-learning framework MetaRes-DMT-AS for elevator few-sample fault diagnosis, which innovatively integrates three core technologies: dynamic meta-training mechanism, adaptive sample scheduling module, and regularized prototype network, and systematically solves key problems such as scarcity of fault samples, imbalanced category distribution, and insufficient generalization ability of diagnostic models in industrial scenarios. The trained model can accurately identify high-risk faults such as elevator emergency stops and severe vibrations under extreme conditions with only a small number of samples / classes, and output high-reliability fault classification results, meeting the stringent requirements of industrial intelligent operation and maintenance systems for fault diagnosis under complex working conditions. This solution not only breaks through the performance bottleneck of traditional deep learning models in small sample scenarios, but also forms a systematic technical breakthrough in dynamic task construction, difficult sample mining, unsupervised feature alignment, etc., significantly improving the practicality and engineering capabilities of elevator fault diagnosis in data-scarce environments.

[0087] Corresponding to the aforementioned embodiment of the elevator acceleration fault detection method based on improved meta-learning, the present application also provides an embodiment of an elevator acceleration fault detection device based on improved meta-learning.

[0088] Figure 9 1 is a block diagram of an elevator acceleration fault detection device based on improved meta-learning according to an exemplary embodiment. Figure 9 , the device comprises:

[0089] Acquisition module 1, used to obtain the original acceleration response data sequence of the target elevator;

[0090] A preprocessing module 2 is used to preprocess the original acceleration response data sequence to obtain a two-dimensional image containing time information;

[0091] A data set selection module 3 is used to randomly select a support set and a query set from the two-dimensional image;

[0092] A model training module 4 is configured to construct and train an accelerated fault detection model based on the support set and the query set, wherein the accelerated fault detection model is configured to extract features from the support set and the query set, and then perform prototype calculation on the extracted features to obtain a loss, wherein an intra-class distance regularization term is introduced in the prototype calculation;

[0093] The sample acquisition module 5 is used to perform classification prediction using the trained accelerated fault detection model. When a sample is misclassified, it is added to the difficult library of the corresponding category to obtain a difficult sample library. The query set consists of the difficult sample library and common samples.

[0094] The data set adjustment module 6 is used to test the trained accelerated fault detection model using a test set to obtain the accuracy of the model, and adjust the support set and the query set according to the accuracy;

[0095] A model retraining module 7 is configured to re-divide the adjusted support set and query set according to the difficult sample library, and retrain the accelerated fault detection model using the re-divide support set and query set to predict the operating status of the elevator;

[0096] The fault detection module 8 is used to input the pre-processed acceleration response data sequence into the retrained acceleration fault detection model to obtain a fault detection result.

[0097] Regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0098] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present application scheme. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0099] Accordingly, the present application also provides an electronic device, comprising: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned elevator acceleration fault detection method based on improved meta-learning.

[0100] Accordingly, the present application also provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the above-mentioned elevator acceleration fault detection method based on improved meta-learning.

[0101] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the contents disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered merely as exemplary, and the true scope and spirit of the present application are indicated by the claims.

[0102] It should be understood that the present application is not limited to the exact structures described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. An elevator acceleration fault detection method based on improved meta-learning, characterized in that: include: Obtaining the original acceleration response data sequence of the target elevator; Preprocessing the original acceleration response data sequence to obtain a two-dimensional image containing time information; randomly selecting a support set and a query set from the two-dimensional image; Constructing and training an accelerated fault detection model based on the support set and the query set, wherein the accelerated fault detection model is used to extract features from the support set and the query set, and then perform prototype calculation on the extracted features to obtain a loss, wherein an intra-class distance regularization term is introduced in the prototype calculation; Using the trained accelerated fault detection model to perform classification prediction, when a sample is misclassified, it is added to the difficult library of the corresponding category to obtain a difficult sample library, wherein the query set consists of the difficult sample library and common samples; Testing the trained accelerated fault detection model using a test set to obtain the accuracy of the model, and adjusting the support set and the query set based on the accuracy; Re-dividing the adjusted support set and query set according to the difficult sample library, and retraining the accelerated fault detection model using the re-divided support set and query set to predict the operating state of the elevator; The preprocessed acceleration response data sequence is input into the retrained acceleration fault detection model to obtain the fault detection results.

2. The method according to claim 1, characterized in that Preprocessing the original acceleration response data sequence to obtain a two-dimensional image containing time information includes: performing normalization processing on the original acceleration response data sequence; Convert the normalized acceleration response data sequence into polar coordinate representation, where the acceleration value corresponds to the cosine value of the angle and the timestamp corresponds to the radius; The acceleration response data sequence represented in polar coordinates is reconstructed into a two-dimensional image using the Gram matrix.

3. The method according to claim 1, characterized in that The loss is expressed as follows: ； Where L is the loss, is the cross entropy loss, is the regularization coefficient, N is the number of categories, K is the number of support samples for each category, The k-th sample feature of category c, is the prototype of category c.

4. The method according to claim 1, wherein Based on the accuracy, adjust the support set and query set, including: If the accuracy rate increases or decreases, the support set and query set will be dynamically adjusted. The specific formula is as follows: ； in is the change in accuracy, p To dynamically adjust parameters, P={S,Q}, S is the support set size, Q is the query set size, t is the current training round, for p The value at round t, for p The value at round t+1, for p The upper bound constraint of for p The lower bound constraint of ; When the accuracy rate decreases, the support set and query set are expanded to alleviate overfitting; when the accuracy rate increases, the support set and query set are reduced to improve generalization ability.

5. The method according to claim 1, wherein Re-dividing the adjusted support set and query set according to the difficult sample library, and retraining the accelerated fault detection model using the re-divided support set and query set, including: Allocating the difficult sample library to the adjusted support set and query set, and retraining the accelerated fault detection model; After training, update the difficult sample library and perform prediction to obtain accuracy; According to the accuracy, the support set and the query set are adjusted until the iteration is completed.

6. The method according to any one of claims 1 to 5, characterized in that The operating status includes elevator reset operation, severe car vibration, emergency stop failure, normal operation and overspeed.

7. An elevator acceleration fault detection device based on improved meta-learning, characterized in that: include: An acquisition module, used for acquiring an original acceleration response data sequence of a target elevator; A preprocessing module, configured to preprocess the original acceleration response data sequence to obtain a two-dimensional image containing time information; A data set selection module, configured to randomly select a support set and a query set from the two-dimensional image; a model training module, configured to construct and train an accelerated fault detection model based on the support set and the query set, wherein the accelerated fault detection model is configured to extract features from the support set and the query set, and then perform prototype calculation on the extracted features to obtain a loss, wherein an intra-class distance regularization term is introduced in the prototype calculation; A sample acquisition module is used to perform classification prediction using the trained accelerated fault detection model. When a sample is misclassified, it is added to the difficult library of the corresponding category to obtain a difficult sample library. The query set consists of the difficult sample library and common samples. A data set adjustment module is used to test the trained accelerated fault detection model using a test set to obtain the accuracy of the model, and adjust the support set and query set according to the accuracy; a model retraining module, configured to re-divide the adjusted support set and query set according to the difficult sample library, and retrain the accelerated fault detection model using the re-divide support set and query set to predict the operating status of the elevator; The fault detection module is used to input the preprocessed acceleration response data sequence into the retrained acceleration fault detection model to obtain the fault detection result.

8. An electronic device, characterized in that: include: one or more processors; a memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the instruction is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images