Method and system for non-destructive online diagnosis of elevator brake faults and fatigue prediction

Abstract

Multiple embodiments of this specification relate to the technical field of special equipment maintenance, and specifically relate to a method and system for non-destructive online diagnosis of elevator brake faults and fatigue prediction. The method includes the steps of: online collecting the action data and braking times when a single elevator brake operates, where the action data includes sound data, vibration data, current data, braking distance, and load; when it is determined that an emergency stop state occurs, immediately extracting the action data obtained during the emergency stop phase, denoted as emergency stop action data; obtaining a fault diagnosis result based on the comparison of the emergency stop action data with a pre-established fault diagnosis model; when the fault diagnosis result is no fault, comparing the action data with a preset brake fatigue baseline to obtain a fatigue prediction result for the single elevator brake. It can perform fault diagnosis and fatigue prediction, avoid the 125% load test, has higher safety, causes no damage to the elevator and the brake, and realizes online detection.

Description

Technical Field

[0001] Multiple embodiments of this specification relate to the technical field of special equipment maintenance, and specifically relate to a method and system for non-destructive online diagnosis of elevator brake faults and fatigue prediction. Background Art

[0002] The brake is a very critical safety component in an elevator, responsible for performing various safety functions during elevator operation, such as normal parking braking, protection braking during upward overspeed, and emergency braking during unexpected car movement. The common type of brake in an elevator is a friction normally closed brake. The so-called normally closed brake here means that in the absence of power supply, the brake is default in the braking state; while when the mechanical system works, it releases the brake. The braking process is achieved by controlling the contact or separation between the brake shoe and the brake disc by the electromagnet core. To ensure the working performance of the brake, the so-called "125% load test" is usually carried out. This test requires operating the brake to stop the traction machine under the condition that the car is loaded with 125% of the rated load, and checking whether the car can be effectively braked.

[0003] However, for in-service elevators, especially old elevators, implementing the "125% load test" may cause potential damage to the traction system, wire ropes, suspension devices, etc. Therefore, researching and realizing non-destructive detection of elevator brakes has become one of the important research topics currently. On the other hand, currently, there is a lack of monitoring means between two regular maintenance periods of elevators, resulting in the operating conditions of elevators not being able to be grasped in time. Therefore, it is necessary to research online elevator brake monitoring technology to improve the safety guarantee of elevator brakes. Summary of the Invention

[0004] Multiple embodiments of this specification describe a method and system for non-destructive online diagnosis of elevator brake faults and fatigue prediction.

[0005] In a first aspect, embodiments of this specification provide a method for non-destructive online diagnosis of elevator brake faults and fatigue prediction, including the steps of:

[0006] Online collect the action data and braking times when the single elevator brake operates, where the action data includes sound data, vibration data, current data, braking distance, and load;

[0007] When it is judged that an emergency stop state occurs, immediately extract the action data in the emergency stop stage and record it as emergency stop action data;

[0008] Obtain a fault diagnosis result according to the comparison between the emergency stop action data and a pre-established fault diagnosis model;

[0009] When the fault diagnosis result is no fault, compare the action data with the preset brake fatigue baseline to obtain the fatigue degree prediction result of the single elevator brake.

[0010] In a second aspect, an embodiment of the present specification provides an elevator brake fault non-destructive online diagnosis and fatigue degree prediction system, including:

[0011] An online acquisition module that online acquires the action data and the number of braking times when the single elevator brake acts, where the action data includes sound data, vibration data, current data, braking distance, and load;

[0012] A judgment module that, when it judges that an emergency stop state occurs, immediately extracts and obtains the action data in the emergency stop stage, denoted as emergency stop action data;

[0013] A fault diagnosis module that obtains a fault diagnosis result according to the comparison between the emergency stop action data and a pre-established fault diagnosis model;

[0014] A fatigue prediction module that, when the fault diagnosis result is no fault, compares the action data with the preset brake fatigue baseline to obtain the fatigue degree prediction result of the single elevator brake.

[0015] In a third aspect, an embodiment of the present specification provides an electronic device, including a processor and a memory;

[0016] The processor is connected to the memory;

[0017] The memory is used to store executable program code;

[0018] The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the method described in any of the above aspects.

[0019] In a fourth aspect, an embodiment of the present specification provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method described in any of the above aspects is implemented.

[0020] In a fifth aspect, an embodiment of the present specification provides a computer program product, including a computer program, and when the computer program is executed by a processor, the method described in any of the above aspects is implemented.

[0021] The beneficial effects brought by the technical solutions provided by some embodiments of the present specification at least include:

[0022] In multiple embodiments of this specification, the provided method for non-destructive online diagnosis of elevator brake faults and fatigue prediction can perform fault diagnosis and fatigue prediction based on the collected action data, avoid the 125% load test, have higher safety, cause no damage to the elevator and the brake, and achieve online detection, with higher safety. By means of the established fault diagnosis model, the fault diagnosis result of the brake can be obtained, which helps to get rid of the situation where the judgment of faults needs to rely on the experience level of maintenance personnel, and the fault diagnosis has higher accuracy. Through the analysis of the action data, it is possible to divide the upward and downward movements, as well as the start stage, braking stage, and emergency stop stage, making the extracted features more targeted and improving the accuracy of fault diagnosis. By means of single elevator data and establishing a general fatigue prediction model, it is possible to generally predict the fatigue life of the brakes of single elevators with the same structure type, the same service year, the number of emergency stops, and the maintenance status, providing an important reference basis for the maintenance of single elevators.

[0023] Other features and advantages of multiple embodiments of this specification will be further revealed in the following specific implementation manners and drawings. Brief Description of the Drawings

[0024] In order to more clearly illustrate the technical solutions in the embodiments of this specification, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of this specification. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0025] Figure 1 Schematic diagram of the application of the method for non-destructive online diagnosis of elevator brake faults and fatigue prediction provided in this specification.

[0026] Figure 2 Schematic diagram of the interaction interface provided in this specification.

[0027] Figure 3 Schematic diagram of the flow of the method for non-destructive online diagnosis of elevator brake faults and fatigue prediction provided in this specification.

[0028] Figure 4 Schematic diagram of the curve of the action data in the downward emergency stop stage.

[0029] Figure 5 Schematic diagram of the curve of the action data in the upward emergency stop stage.

[0030] Figure 6 Schematic diagram of the curve of the action data in the upward emergency stop stage using the STO technology.

[0031] Figure 7Schematic diagram of the method flow for establishing a fault diagnosis model provided in this specification.

[0032] Figure 8 Schematic diagram of the method flow for establishing a brake fatigue baseline provided in this specification.

[0033] Figure 9 Another schematic diagram of the method flow for establishing a brake fatigue baseline provided in this specification.

[0034] Figure 10 Schematic diagram of the brake fatigue baseline provided in this specification.

[0035] Figure 11 Schematic diagram of the method flow for establishing a general fatigue prediction model provided in this specification.

[0036] Figure 12 Another schematic diagram of the method flow for fatigue prediction provided in this specification.

[0037] Figure 13 Schematic diagram of the elevator brake fault non-destructive online diagnosis and fatigue prediction system provided in this specification.

[0038] Figure 14 Schematic diagram of the electronic device provided in this specification. Specific implementation manners

[0039] The technical solutions of the embodiments of this specification will be explained and described below with reference to the accompanying drawings of the embodiments of this specification. However, the following embodiments are only the preferred embodiments of this specification, not all of them. Based on the embodiments in the implementation manners, other embodiments obtained by those skilled in the art without creative efforts all fall within the protection scope of this specification.

[0040] Terms such as "first", "second", "third", etc. in the specification, claims and the above accompanying drawings of this specification are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or devices.

[0041] In the following description, terms indicating orientation or positional relationships such as "inside", "outside", "above", "below", "left", "right", etc. are only for the convenience of describing the embodiments and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation to this specification.

[0042] The data involved in this application are all information and data authorized by users or fully authorized by all parties, and the collection of relevant data complies with the relevant laws, regulations and standards of relevant countries and regions.

[0043] Before introducing the technical solutions described in this specification, the application scenarios and related technologies of the technical solutions are introduced.

[0044] Special equipment refers to boilers, pressure vessels, elevators, cranes, special-purpose motor vehicles, etc. that involve life safety and are relatively dangerous. These devices have specific working conditions and technical requirements, and require special management and maintenance. An elevator belongs to special equipment and is used to quickly transport people between building floors. A traction elevator drives the traction rope through an electric motor, so that the car 12 runs on the guide rail to realize movement between floors. The elevator is in a state of reciprocating motion for a long time. In order to ensure its normal operation and safety, regular manual maintenance is required.

[0045] The maintenance work of the elevator is usually carried out by professional technicians. The maintenance personnel enter the elevator hoistway, machine room, car top, etc. to inspect each component of the elevator. This includes cleaning the pulleys, guide shoes and the interior of the car 12; inspecting and adjusting the door system; detecting the electrical control system; calibrating the speed limiter and safety gear device; and testing functions such as emergency lighting and alarm phones. In addition, it is also necessary to lubricate the mechanical part of the elevator.

[0046] The brake 11 is a key component in the elevator safety system and is responsible for quickly braking the car 12 in the event of a power outage or abnormal situation to prevent a falling accident. For the maintenance of the elevator brake 11, it is first necessary to ensure that the brake 11 can respond to commands and act immediately, that is, when the control system issues a stop signal, the brake 11 can quickly and effectively stop the elevator. During the maintenance process, it is necessary to check whether the spring force of the brake 11 is sufficient. At the same time, it is also necessary to check whether the gap between the brake lining and the brake disc is appropriate. Too small a gap will cause unnecessary wear, while too large a gap will result in poor braking effect. The surface of the brake 11 should be kept clean and free of oil. The maintenance personnel will also judge whether there is an abnormal situation with the brake 11 based on the action and sound of the brake 11.

[0047] The manual maintenance method has some drawbacks. Due to the tight time for maintenance work, there may be a risk of undetected problems. Some faults or hidden dangers may only appear under specific conditions, and these problems are difficult to capture through regular maintenance inspections. On the other hand, manual maintenance depends on the experience and technical level of maintenance personnel, and there may be differences among different personnel, which will also affect the consistency of maintenance quality. Moreover, with the acceleration of urbanization, there are more and more high-rise buildings and the number of elevators has increased sharply, bringing huge challenges to the maintenance work. Facing the huge maintenance task volume, how to ensure that each elevator can be maintained in a timely and effective manner has become an urgent problem to be solved.

[0048] To improve the maintenance efficiency and quality of the elevator brake 11, this embodiment provides a method and system for fault diagnosis and fatigue prediction of the elevator brake 11.

[0049] Please refer to the appendix Figure 1 In the application of this embodiment, a current sensor 21, a vibration sensor 22 and a sound sensor 23 are installed on the brake 11, and a vibration sensor 22 and a sound sensor 23 are installed on the top of the car 12. After obtaining the action data, the action data is uploaded to the server 40 through the handheld terminal 30, and the server 40 gives the fault diagnosis and fatigue prediction results. In the case of edge computing on the handheld terminal 30, part or all of the fault diagnosis and fatigue prediction can also be performed on the handheld terminal 30.

[0050] The method provided in this application is applied to the system architecture as Figure 1 shown. The system architecture includes a server 40 and a handheld terminal 30. An interactive interface 31 is set on the handheld terminal 30. Please refer to the appendix Figure 2, the interactive interface 31 can run on the handheld terminal 30 in the form of a browser, or can also run on the handheld terminal 30 in the form of an independent application (APP), etc. For the specific display form of the interactive interface 31, no limitation is made here. The server 40 involved in this application can be an independent physical server, or can also be a server cluster or distributed system composed of multiple physical servers, and can also be a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms. The handheld terminal 30 can be a smart phone, a tablet computer, a laptop computer, a handheld computer, a personal computer, a smart speaker, a smart TV, a smart watch, a vehicle-mounted device, a wearable device, etc., but is not limited thereto. The handheld terminal 30 and the server 40 can be directly or indirectly connected through wired or wireless communication methods, and no limitation is made in this application. The number of the server 40 and the handheld terminal 30 is also not limited. The solution provided by this application can be completed independently by the handheld terminal 30, or can also be completed independently by the server 40, or can also be completed by the cooperation of the handheld terminal 30 and the server 40. In this regard, no specific limitation is made in this application.

[0051] In view of the fact that this application will involve some professional terms, therefore, the following will first introduce this part of professional terms.

[0052] Machine learning model

[0053] A machine learning model refers to a method that enables a computer system to learn from data through algorithms and statistical models, so as to improve its performance or make predictions without explicit programming. Machine learning is a subfield of artificial intelligence that allows a computer to learn from experience and improve its functions without being explicitly programmed. Machine learning models can be classified into the following categories: supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning.

[0054] Clustering algorithm

[0055] Clustering algorithms are unsupervised learning methods that aim to divide the objects in a dataset into several clusters such that the objects within a cluster are similar to each other, while the objects between different clusters are quite different. Clustering algorithms are widely used in many fields such as data analysis, pattern recognition, market segmentation, social network analysis, etc. By measuring the similarity or distance between data points, the dataset is divided into several groups, and the data points in each group share certain features. The goal of clustering is to make the data points within a cluster as similar as possible, while the differences between clusters are as large as possible. Clustering algorithms include partitioning clustering such as K-means, which is one of the most commonly used clustering algorithms. It divides the data into a preset number of clusters through an iterative process, and the center point (centroid) of each cluster represents the average attributes of the cluster. K-medoids is similar to K-means, but the centroid must be an actual data point in the cluster. Hierarchical clustering such as agglomerative hierarchical clustering merges the nearest clusters bottom-up to form a tree-like structure. Divisive hierarchical clustering divides the dataset top-down into smaller and smaller clusters. Density-based clustering, such as DBSCAN, discovers clusters of arbitrary shapes by measuring the density of the neighborhood around a point and can identify noise points. OPTICS extends the idea of DBSCAN and provides a hierarchical density clustering method. Grid-based clustering, such as STING, divides the space into cells and performs statistical analysis at the cell level. CLIQUE combines the advantages of grid partitioning and hierarchical clustering and can efficiently process large-scale datasets. Model-based clustering such as the EM algorithm uses a mixture model (such as Gaussian mixture model) for clustering and estimates the probability distribution by iteratively optimizing parameters.

[0056] Autoencoder model

[0057] The autoencoding model is the autoencoder. The autoencoder is an unsupervised learning algorithm mainly used for data dimensionality reduction, feature extraction, and data reconstruction. It achieves this process through two main parts - the encoder and the decoder. The autoencoder consists of an encoder, a latent space, and a decoder. Among them, the encoder compresses the input data into a low-dimensional latent space representation. This process is usually achieved through several layers of neural networks, aiming to extract the key features or information in the input data. The latent space is the low-dimensional representation output by the encoder, also known as the code or bottleneck. The representation in this space is a compressed form of the input data, containing the key information required to reconstruct the original data. The decoder then decodes the low-dimensional latent space representation back to the dimension of the original data. The decoder is also composed of several layers of neural networks, and its goal is to reconstruct the original input data as accurately as possible. The autoencoder is trained and used in two stages. In the training stage, the input data x is mapped to the latent space z through the encoder, expressed as z = f(x). Then, the latent space representation z is reconstructed into the original data x' through the decoder, expressed as x' = g(z) = g(f(x)). The training objective is to minimize the reconstruction error, that is, the difference between x and x'. The commonly used loss function is the mean square error (MSE). In the usage stage, after training, the encoder can be used to map new data to the low-dimensional latent space for feature extraction or dimensionality reduction; the decoder can generate data from the latent space representation and be applied to tasks such as generative models.

[0058] This specification first provides a method for non-destructive online diagnosis of elevator brake faults and fatigue prediction. Please refer to the appendix Figure 3 , including the steps:

[0059] Step S101) Online collect the action data and braking times when a single elevator brake operates. The action data includes sound data, vibration data, current data, braking distance, and load.

[0060] The sound data can be collected using an omnidirectional MEMS audio sensor or a bone conduction sound sensor. The omnidirectional MEMS audio sensor is a miniaturized microphone that can capture sounds from all directions. It has high sensitivity and can effectively pick up low-volume sounds. However, it is easily interfered by ambient noise, and the airflow generated during elevator operation can also produce noise. The bone conduction sound sensor mainly senses the vibration generated by the part where the sound is produced by contacting that part and converts it into an electrical signal for processing. The bone conduction sound sensor can reduce ambient noise interference. However, there may be some distortion, especially in the high-frequency part, so there may be a risk of missing faults that generate high-frequency sounds. The current data is the current data of the brake electromagnetic coil. The current data can be collected using a current sensor 21 connected in series to the electromagnetic coil or a Hall current sensor 21 to obtain the detection result of the current without invading the electromagnetic coil circuit.

[0061] The vibration sensor 22 is a six-axis vibration sensor 22. The six-axis vibration sensor 22 can actually not only detect vibrations but also directly reflect the six-axis acceleration itself. The total vibration after synthesis by the six-axis vibration sensor 22 constitutes vibration data. When counted separately by six axes, it can be used as six-axis acceleration data.

[0062] In this embodiment, the relevant sensors and devices for collecting action data are long-term deployed at positions related to the elevator and the brake, enabling continuous online collection of action data. And by setting up a concentrator, the collected action data is centralized, and a communication connection is established with a server or a cloud server to regularly report the action data. An edge computing device can also be deployed on the concentrator or on a device connected to the concentrator to achieve online real-time diagnosis of single elevator brake faults and prediction of fatigue. Achieving online collection of action data and using edge computing technology to achieve real-time non-destructive fault diagnosis and fatigue prediction are relatively more significant new technical effects of this embodiment. At the same time, in this embodiment, the faults can be diagnosed and the fatigue can be predicted through the action data, without any damage to the elevator and the brake, and it has relatively higher safety.

[0063] There are various ways to obtain the braking distance, specifically including: (1) Detect the vibration of the car 12 through the vibration sensor 22. When there is no vibration or the vibration is very small, it indicates that the elevator is stationary. When the vibration is relatively regular and the amplitude change rate is less than a certain threshold, it indicates that the elevator is in a basically uniform motion state. When the amplitude change rate of the vibration exceeds a certain threshold, it indicates that the elevator is in the start-up stage or the braking stage. According to the vibration situation before the amplitude change of the vibration, the start-up stage or the braking stage can be distinguished. When it is in the braking stage and the amplitude detected by the vibration sensor 22 is small, it indicates that the braking is completed. At this time, according to the position of the elevator car 12 at the start and end times of the braking stage, the braking distance can be obtained, and the position of the elevator car 12 is obtained from the elevator controller. (2) When the current sensor 21 detects that the electromagnetic coil of the brake 11 is energized, it indicates the start of braking. At this time, the brake 11 will also vibrate due to friction. When the vibration amplitude detected by the vibration sensor 22 on the brake 11 is lower than a certain value, it indicates that the car 12 has been braked and the braking ends. According to the position of the elevator car 12 at the start and end times of the braking stage, the braking distance can be obtained. (3) According to the acceleration accumulation method detected on the vertical axis of the six-axis vibration sensor 22, the speed of the car 12 is obtained, and the start and end times of the braking stage are judged according to the speed of the car 12. The braking distance is obtained from the position of the car 12 at the start and end times of the braking stage, or the braking distance is obtained according to the integral of the speed of the car 12.

[0064] The action data is associated with the annotation of the braking action generated by the brake, and the annotation includes one or more of a stage annotation and a direction annotation. The stage annotation includes start, brake, and emergency stop, and the direction annotation includes up and down. The number of start times is equal to the sum of the number of brake times and the number of emergency stop times. That is, normal braking of the elevator belongs to the braking stage, and emergency stop belongs to the emergency stop stage.

[0065] It should be noted that the braking distance during upward travel is different from that during downward travel. The braking distances under different load conditions are also different. The role of the annotation is to group the action data, and it has higher accuracy to predict faults and fatigue degrees for each group separately. When the fatigue degree prediction results for upward and downward travel are different, the mean value of the two can be used, or the fatigue degrees for upward and downward travel can be represented separately. Exemplarily, as shown in Figure 4 the figure, it is the jerk, acceleration, and speed curves during the downward emergency stop stage. As shown in Figure 5As shown, they are the jerk, acceleration, and speed curves during the upward emergency stop. Additionally, some elevators are equipped with STO technology. The STO (Safe Torque Off) technology is a safety function used in motor drive systems, aiming to ensure that the torque output of the motor can be quickly and reliably cut off in case of an emergency or when needed. STO achieves the rapid disconnection of the motor power supply by directly controlling the safety relay or other forms of switching devices inside the motor drive through a hardware circuit. When the STO function is activated, it prevents current from flowing into the motor windings, thereby immediately braking the torque generated by the motor. This includes not only the active torque output but also any remaining torque that may be caused by inertia or other factors. As attached Figure 6 As shown, they are the jerk, acceleration, and speed curves during the upward emergency stop with the STO technology. In the attached Figures 4-6 from top to bottom are the jerk curve, acceleration curve, and speed curve respectively. Comparing the attached Figure 5 and the attached Figure 6 it can be seen that the emergency stop with the STO technology is smoother. However, whether the STO technology is adopted or not, the brake 11 fault diagnosis and fatigue prediction method disclosed in this embodiment can be applied

[0066] Step S102) When it is determined that an emergency stop state occurs, immediately extract the action data during the emergency stop phase, denoted as the emergency stop action data. The acceleration, current data, and braking distance obtained from the vibration data can all accurately identify the emergency stop phase. When the acceleration is greater than the threshold, the current exceeds the threshold, or the braking distance is less than a certain value, it can be determined that an emergency stop has occurred

[0067] The emergency stop includes both the downward emergency stop and the upward emergency stop. The upward emergency stop is the phenomenon of overshooting the top

[0068] Step S103) Obtain the fault diagnosis result based on the comparison between the emergency stop action data and the pre-established fault diagnosis model

[0069] Among them, please refer to the attached Figure 7 The method for establishing the fault diagnosis model includes

[0070] Step S201) Receive the action data related to the fault, and group the action data according to the annotation

[0071] Step S202) Extract the time-frequency domain features of the sound data, vibration data, and current data included in the action data of each group respectively

[0072] Step S203) Compare the time-frequency domain features of the action data of each group with the time-frequency domain features of the action data in the normal state of a single elevator respectively to obtain the differential action data

[0073] Among them, the method for obtaining the differential action data by comparing the time-frequency domain features of the action data with those of the action data in the normal state of a single ladder includes:

[0074] Calculate the similarity of the time-frequency domain features according to the frequency composition and its amplitude, and calculate the similarity of the action data of the associated fault and the action data in the normal state of a single ladder;

[0075] When the similarity is lower than a preset threshold, obtain the differential action data according to the action data corresponding to the time-frequency domain features.

[0076] The frequency composition and its amplitude can be expressed in the form of a vector, and calculating the similarity is to calculate the similarity between vectors. The similarity between vectors can adopt the reciprocal of the distance between vectors.

[0077] Step S204) Establish and use the time-frequency domain features of the differential action data to train an autoencoder model to obtain the feature expressions of the sound data, vibration data, and current data of the differential action data.

[0078] Step S205) After associating the feature expressions, braking distance, and load of each group, obtain the fault sample data of this group after an associated fault.

[0079] Step S206) Respectively establish and use the fault sample data of each group to train a machine learning model, and obtain the fault diagnosis model of each group according to the machine learning model.

[0080] On the other hand, the method for obtaining a fault diagnosis result by comparing the emergency stop action data with a pre-established fault diagnosis model includes:

[0081] Input the time-frequency domain features of the sound data, vibration data, and current data included in the emergency stop action data into the autoencoder model to obtain the feature expressions of the sound data, vibration data, and current data;

[0082] After associating the feature expressions of the sound data, vibration data, and current data with the corresponding braking distance and load, input them into the fault diagnosis model to obtain the fault diagnosis result.

[0083] Faults are highly correlated with phases, which are divided into a startup phase, a braking phase, and an emergency stop phase. The action data is not differentiated by phase and does not affect fault diagnosis. For example, some faults only have abnormal action data in the startup phase. For instance, the brake 11 is jammed, that is, the brake shoe does not open completely and the elevator runs with the brake engaged. At this time, abnormal action data will appear in the startup phase, while there will be no abnormal action data in the braking phase or the emergency stop phase. When the phase of the action data is not differentiated, when the fault of the brake 11 being jammed is recognized, it will naturally be inferred that the action data corresponding to the startup phase matches the fault diagnosis model. Another example is that insufficient braking force may be caused by excessive brake shoe clearance, excessive wear, oil contamination, etc., and abnormal action data will appear in the braking phase and the emergency stop phase, while the action data in the startup phase is normal. Therefore, the characteristic expressions associated with faults may be one or more of the startup phase, the braking phase, and the emergency stop phase.

[0084] When a large amount of action data for faults is obtained, the action data for the same fault also varies. By extracting time-frequency domain features from the action data, then clustering, and screening out the time-frequency domain features with high dispersion. The time-frequency domain features of the clustered action data correspond to a fault type. Taking the time-frequency domain features of the clustering center and associating them with a fault type, a sample data can be obtained.

[0085] After obtaining the fault sample data, using the fault sample data to train a machine learning model can obtain the corresponding fault diagnosis model. It should be noted that the quantities contained in the fault sample data for different fault types are different. Therefore, it is necessary to fill in the quantities not contained in the fault sample data, that is, fill them with default values. For example, the fault of the brake being jammed only corresponds to the action data in the startup phase, and the action data in the other phases is filled with default values.

[0086] When obtaining the fault diagnosis result of the brake 11, it is necessary to extract the characteristic expression from the collected action data using the autoencoder model generated in the manner described above, and input the corresponding characteristic expression into the fault diagnosis model. It should be noted that during the process, one or more of the action data in the startup phase, the braking phase, and the emergency stop phase need to be covered with default values and input into the fault diagnosis model multiple times. For example, first cover the action data in the braking phase and the emergency stop phase with default values to check whether the brake 11 has the fault of the brake 11 being jammed. Similarly, the action data in the startup phase can also be covered with default values to determine whether the brake 11 has insufficient braking force.

[0087] Among them, there are two comparison methods. One is to take the composition and amplitude of the upward sound frequency, the composition and amplitude of the downward sound frequency, the composition and amplitude of the upward vibration frequency, the composition and amplitude of the downward vibration frequency, the upward current data value, the downward current data value, the upward braking distance value, and the downward braking distance value as an overall vector, and find the point with the closest vector distance between the brake fatigue baseline 50 and this vector. The braking times corresponding to this point are the fatigue degree prediction results. If the maximum healthy braking times of the single-ladder brake 11 are known, the fatigue life situation of the current single-ladder brake 11 can be obtained.

[0088] The other method is to separately find the braking times corresponding to the composition and amplitude of the upward sound frequency, the composition and amplitude of the downward sound frequency, the composition and amplitude of the upward vibration frequency, the composition and amplitude of the downward vibration frequency, the upward current data value, the downward current data value, the upward braking distance value, and the downward braking distance value, and average or weighted average these braking times as the fatigue degree prediction results.

[0089] On the other hand, in another embodiment, the composition and amplitude of the upward sound frequency, the composition and amplitude of the downward sound frequency, the composition and amplitude of the upward vibration frequency, the composition and amplitude of the downward vibration frequency, the upward current data value, the downward current data value, the upward braking distance value, and the downward braking distance value are used as an overall vector. This vector is multiplied by a weight coefficient vector. After adding the weights, the point with the closest vector distance between the brake fatigue baseline 50 and this vector is found. The generation method of the weight coefficient vector is to use machine learning, and the machine learning model gives the best weight coefficient vector. Such as BP neural network model, decision tree algorithm, random forest algorithm, support vector machine, deep learning algorithm. Exemplarily, the BP neural network model is used. The BP neural network model is constructed, and the feature expressions of the action data under different faults are used as input features, and the fault results are used as output results. By training the model, the influence weights of each input feature on the output target can be obtained.

[0090] Step S104) When the fault diagnosis result is no fault, compare the action data with the preset brake fatigue baseline to obtain the fatigue degree prediction result of the single-ladder brake.

[0091] Among them, please refer to the appendix Figure 8 , the method for presetting the brake fatigue baseline includes:

[0092] Step S301) Collect the action data and braking times of the brakes of multiple single-ladders operating normally.

[0093] Step S302) Extract the action data in the starting stage and braking stage.

[0094] Step S303) Extract the time-frequency domain features of the sound data, vibration data, and current data included in the action data during the startup phase and the braking phase. The extraction of time-frequency domain features is performed through publicly disclosed techniques in this field such as Fourier transform and wavelet transform. The time-frequency domain features include the frequency composition and its amplitude. The time-frequency domain features refer to the features that take into account both the time domain and frequency domain characteristics, which are already existing technologies in this field. The feature expressions of the sound data, vibration data, and current data are also the frequency composition and amplitude.

[0095] Step S304) Establish a brake fatigue baseline based on the time-frequency domain features of the sound data, vibration data, and current data, the braking distance, the load, and the number of braking times.

[0096] The method for obtaining the fatigue degree prediction result of a single-ladder brake by comparing the action data with a preset brake fatigue baseline includes:

[0097] Extract the time-frequency domain features of the sound data, vibration data, and current data of the action data, and read the braking distance and load of the action data;

[0098] Obtain the fatigue degree prediction result of the single-ladder brake based on the comparison of the time-frequency domain features of the sound data, vibration data, and current data, the braking distance, and the load with the brake fatigue baseline.

[0099] Among them, in step S304), according to the time-frequency domain features of the sound data, vibration data, and current data, the braking distance, the load, and the number of braking times, please refer to the appendix Figure 9 , the method for establishing a brake fatigue baseline includes:

[0100] Step S401) Divide the load into several load intervals according to a preset interval, and group the time-frequency domain features of the sound data, vibration data, and current data and the braking distance according to the load intervals;

[0101] Step S402) Use the number of braking times as a label to associate the time-frequency domain features of the sound data, vibration data, and current data and the braking distance of each group to obtain calibrated data;

[0102] Step S403) Establish a brake fatigue baseline for each group according to the calibrated data.

[0103] The brake fatigue baseline is directly obtained based on the action data and the number of braking times in the normal state of a single ladder.

[0104] The action data and braking times of the single ladder under normal conditions can be obtained under laboratory conditions or from the historical data collected during routine maintenance. When dividing the load into several load ranges, for example, 0% - 30% of the maximum load is taken as the low load range, 30% - 60% as the medium load range, and over 60% as the high load range. Or other division methods can be adopted.

[0105] Refer to the method described above to distinguish the start-up stage, braking stage, and emergency stop stage. That is, distinguish by means of the acceleration in the vertical direction, or distinguish according to the electromagnetic coil current of the brake 11 combined with vibration data. The feature expressions include sound data, vibration data, current data, and the feature expressions of braking distance. At the same time, it can also distinguish between upward and downward movements. Therefore, the feature expressions of the start-up stage finally obtained include low load feature expressions, medium load feature expressions, and high load feature expressions. The low load feature expressions, medium load feature expressions, and high load feature expressions all include upward sound feature expressions, downward sound feature expressions, upward vibration feature expressions, downward vibration feature expressions, upward current feature expressions, downward current feature expressions, upward braking distance feature expressions, and downward braking distance feature expressions. The upward current feature expression and the downward current feature expression are the current values of upward and downward movements respectively. The upward braking distance feature expression and the downward braking distance feature expression are the braking distances of upward and downward movements respectively. The feature expressions are also time-frequency domain features, that is, including the frequency composition and its amplitude.

[0106] When distinguishing between upward and downward movements based on sound data, vibration data, and current data, it will make the fault diagnosis and fatigue prediction of the elevator have more accurate results, but it will increase the data processing volume and reduce the efficiency to a certain extent. If not distinguishing between upward and downward movements, it is best to use only the upward or only the downward action data for fault diagnosis and fatigue prediction. When the upward and downward action data are used without distinction, the accuracy of the obtained fault diagnosis and fatigue prediction results is relatively lower.

[0107] On the other hand, in another embodiment, when an emergency stop occurs, the preset equivalent braking times of the emergency stop are added to the predicted braking times of the single ladder brake, and based on the predicted braking times of the single ladder brake, the fatigue prediction result of the single ladder brake is obtained.

[0108] Among them, the method for presetting the equivalent braking times of the emergency stop includes:

[0109] Under laboratory conditions, measure the brake shoe wear amount caused by N brakings and the brake shoe wear amount caused by N emergency stops at the preset load;

[0110] According to the ratio of the brake shoe wear amount caused by N brakings to the brake shoe wear amount caused by N emergency stops, obtain the equivalent braking times of the emergency stop.

[0111] On the other hand, in other embodiments, according to the characteristic expressions of the braking phase and the emergency stop phase, the equivalent braking times of the emergency stop are obtained. The specific method includes: sorting the characteristic expressions of the braking phase and the emergency stop phase along the time axis; calculating the change rate of the characteristic expressions of the adjacent braking phases before and after the emergency stop phase; calculating multiple change rates corresponding to multiple emergency stop phases to obtain an average change rate; selecting multiple braking phases to obtain the change rate of the characteristic expressions of the adjacent braking phases before and after the overall multiple braking phases; and taking the number of braking phases whose change rate is closest to the average change rate as the equivalent braking times of the emergency stop. The equivalent braking times of one emergency stop can also be obtained by calculation.

[0112] Specifically, the following characteristic expressions of the action data are extracted: the frequency composition and amplitude of the upward sound, the frequency composition and amplitude of the downward sound, the frequency composition and amplitude of the upward vibration, the frequency composition and amplitude of the downward vibration, the upward current data value, the downward current data value, the upward braking distance value, the downward braking distance value, and compared with the brake fatigue baseline 50.

[0113] Among them, the method for obtaining the fatigue prediction result of the single-ladder brake according to the action data of the single-ladder brake and the brake fatigue baseline includes:

[0114] Extracting the time-frequency domain characteristics of the sound data, vibration data, and current data included in the action data;

[0115] Obtaining the brake fatigue baseline of the corresponding load range according to the load of the action data;

[0116] Respectively comparing the time-frequency domain characteristics and braking distances of the sound data, vibration data, and current data included in the action data with the brake fatigue baseline to obtain the number of braking times that match;

[0117] The average value of all the matching braking times is used as the fatigue prediction result of the single-ladder brake.

[0118] Specifically, according to all tags, the characteristic expression reference in the starting stage of the load, and the braking distance reference, the brake fatigue baseline 50 in the starting stage is obtained. The average value of the characteristic expression refers to the average value of the frequency composition. Exemplarily, one characteristic expression includes frequencies F1, F2, and F3 with amplitudes A11, A12, and A13 respectively, and another characteristic expression includes frequencies F1, F3, F4 with amplitudes A21, A23, and A24 respectively. Then the average value of these two characteristic expressions includes frequencies F1, F2, F3, F4 with amplitudes (A11 + A21) / 2, A12 / 2, (A13 + A23) / 2, and A24 / 2 respectively. When the number of characteristic expressions participating in the calculation of the average value is large enough, if the frequency F4 only exists in one characteristic expression, then in the finally calculated average value of the characteristic expression, the amplitude of the frequency F4 can be almost ignored. Therefore, through the average value of the characteristic expression, the typical frequency composition and its amplitude with reference value can be effectively obtained.

[0119] The characteristic expression includes the characteristic expressions of sound data, vibration data, current data, and braking distance. At the same time, it can also distinguish between upward and downward. Therefore, the finally obtained brake fatigue baseline 50 in the starting stage includes the low load baseline 51, the medium load baseline 52, and the high load baseline 53. The horizontal axes of the low load baseline 51, the medium load baseline 52, and the high load baseline 53 are all the number of braking times, that is, the fatigue degree is represented by the number of braking times, and the vertical axis is the amplitude. A point on the low load baseline 51, the medium load baseline 52, and the high load baseline 53 represents the corresponding upward sound reference frequency composition and amplitude 511, downward sound reference frequency composition and amplitude 512, upward vibration reference frequency composition and amplitude 513, downward vibration reference frequency composition and amplitude, upward current data reference value, downward current data reference value, upward braking distance reference value, downward braking distance reference value at this number of braking times, as shown in the appendix Figure 10 shown. Similarly, the low load baseline 51, the medium load baseline 52, and the high load baseline 53 can be correspondingly obtained in the braking stage and the emergency stop stage. Together, they constitute the complete brake fatigue baseline 50.

[0120] On the other hand, in another embodiment, the method for non-destructive online diagnosis and fatigue degree prediction of elevator brake faults further includes the steps of:

[0121] Sort the action data of two single elevator brakes according to the time axis, obtain the action data of the two single elevator brakes with an interval of M, and extract the time-frequency domain characteristics.

[0122] According to the action data of the two single elevator brakes and the brake fatigue baseline, obtain the predicted number of braking times of the two single elevator brakes respectively.

[0123] Exemplarily, according to the feature expression, obtaining the vibration and wear generated by one emergency stop, which is equivalent to the vibration and wear generated by several normal brakings, can be obtained by using empirical formulas, manual settings or simulation, which is a publicly disclosed technology in this field. Specifically, when using empirical formulas, the equivalent braking times of one emergency stop can be calculated based on the wear amount of the brake shoe caused by 1000 normal brakings and the wear amount of the brake shoe caused by 1000 emergency stops. That is, the equivalent braking times of one emergency stop can be obtained according to the same wear amount of the brake shoe, which can be obtained through laboratory or in-plant tests. When using manual settings, the equivalent braking times of the emergency stop can be set to 2 times. When using simulation, the wear amount of the brake shoe is set as the observed quantity, and the equivalent braking times that are the same as the wear amount of the brake shoe caused by one emergency stop are obtained through simulation. On the other hand, according to different load ratios, an equivalent braking time can be set for one emergency stop under each load.

[0124] On the other hand, in another embodiment, the difference between the predicted braking times of two single-ladder brakes is calculated as the predicted difference, and the ratio of the predicted difference to M is calculated as the proportionality coefficient. The equivalent braking times are obtained according to the proportionality coefficient.

[0125] In this embodiment, the predicted braking times of the single-ladder brake are directly used as the fatigue prediction result. When the total braking times of the single-ladder brake are known, the fatigue condition of the current single-ladder brake can be judged. The total braking times can be obtained from the factory data.

[0126] On the other hand, please refer to the append Figure 11 , the method further includes the steps:

[0127] Step S501) Receive single-ladder data, where the single-ladder data includes the usage year, braking times, emergency stop times, and maintenance status;

[0128] Step S502) Incorporate the usage year, emergency stop times, and maintenance status of the single-ladder data into the corresponding action data;

[0129] Step S503) Associate the braking times of the single-ladder data with the corresponding action data as labels to obtain sample data;

[0130] Step S504) Establish a general fatigue prediction model for the corresponding type of brake according to the sample data of the same type of brake.

[0131] Use the number of brake applications as a label. The number of brake applications includes normal braking, and an emergency stop will be converted into the number of brake applications. After obtaining sufficient single-ladder data in terms of quantity and time length, the maximum number of brake applications for a sufficient number of individual units can be obtained. Find and calculate the average of all the maximum number of brake applications of the general type of brake 11 as the maximum service life, i.e., the fatigue life, of this type of brake 11. When used for a newly installed single ladder, the fatigue life of the newly installed single ladder can be obtained based on the structure of the brake 11 for reference.

[0132] On the other hand, please refer to the appendix Figure 12 , the method further includes the steps:

[0133] Step S601) Obtain the action data of the single-ladder brake during normal operation at a preset period and extract the time-frequency domain features;

[0134] Step S602) Obtain the predicted number of brake applications of the single-ladder brake according to the action data and the brake fatigue baseline;

[0135] Step S603) Obtain the predicted fatigue degree result of the single-ladder brake according to the predicted number of brake applications of the single-ladder brake.

[0136] In a regular manner, when there is no emergency stop for the single ladder, the fatigue degree prediction result of the single-ladder brake will also be carried out through the braking data during normal operation, and the fatigue degree prediction result will be uploaded to the server, which has a guiding role for the health status and maintenance of the single ladder.

[0137] On the other hand, even after obtaining the fault diagnosis result and the fatigue degree prediction result of the single-ladder brake 11, it is necessary to judge whether the single-ladder brake 11 needs to be repaired or replaced. Currently, it still relies on the experience judgment of maintenance personnel. This specification provides a machine learning model. After associating the action data with the labels of whether to dispose and the specific disposal methods, a machine learning model is established and trained, such as a BP neural network model, decision tree algorithm, random forest algorithm, support vector machine, and deep learning algorithm. After training, input the fault diagnosis result and the fatigue degree prediction result into this machine learning model, and the result of whether to dispose and how to dispose can be obtained for reference.

[0138] On the other hand, this specification provides an elevator brake fault non-destructive on-line diagnosis and fatigue degree prediction system. Please refer to the appendix Figure 13 , including:

[0139] An on-line acquisition module 100, which on-line acquires the action data and the number of brake applications when the single-ladder brake acts. The action data includes sound data, vibration data, current data, braking distance, and load;

[0140] The determination module 200, when it determines that an emergency stop state occurs, immediately extracts and obtains the action data in the emergency stop phase, which is recorded as the emergency stop action data;

[0141] The fault diagnosis module 300 obtains a fault diagnosis result according to the comparison between the emergency stop action data and a pre-established fault diagnosis model;

[0142] The fatigue prediction module 400, when the fault diagnosis result is no fault, compares the action data with a preset brake fatigue baseline to obtain a fatigue degree prediction result of the single ladder brake.

[0143] Please refer to Figure 14 the structural schematic diagram of an electronic device provided by an embodiment of the present specification shown.

[0144] As Figure 14 shown, the electronic device 1100 may include: at least one processor 1101, at least one network interface 1104, a user interface 1103, a memory 1105, and at least one communication bus 1102. Among them, the communication bus 1102 can be used to realize the connection and communication of the above-mentioned various components. Among them, the user interface 1103 may include buttons, and the optional user interface may further include a standard wired interface and a wireless interface. Among them, the network interface 1104 may include, but is not limited to, a Bluetooth module, an NFC module, a Wi-Fi module, etc. Among them, the processor 1101 may include one or more processing cores. The processor 1101 connects various parts within the entire electronic device 1100 through various interfaces and lines, and executes various functions of the routing device and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 1105, and by calling the data stored in the memory 1105. Optionally, the processor 1101 may be implemented in at least one of the hardware forms of DSP, FPGA, and PLA. The processor 1101 may integrate one or several combinations of a CPU, a GPU, and a modem, etc. Among them, the CPU mainly processes the operating system, the user interface, and application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communication.

[0145] It can be understood that the above-mentioned modem may not be integrated into the processor 1101 and may be implemented separately through a single chip.

[0146] Among them, the memory 1105 may include RAM or ROM. Optionally, the memory 1105 includes a non-transitory computer-readable medium. The memory 1105 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 1105 may include a program storage area and a data storage area. Among them, the program storage area may store instructions for implementing the operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above method embodiments, etc.; the data storage area may store the data involved in the above method embodiments. Optionally, the memory 1105 may also be at least one storage device located far from the aforementioned processor 1101. The memory 1105 as a computer storage medium may include an operating system, a network communication module, a user interface module, and application programs. The processor 1101 may be used to call the application programs stored in the memory 1105 and execute the methods in the above multiple embodiments.

[0147] The embodiments of this specification also provide a computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are run on a computer or a processor, the computer or the processor is caused to execute multiple steps in the above embodiments. If each component module of the above electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in the computer-readable storage medium.

[0148] The embodiments of this specification also provide a computer program product, including a computer program. When the computer program is executed by a processor, multiple steps in the above embodiments are implemented.

[0149] Without conflict, the technical features in this embodiment and the implementation schemes can be combined arbitrarily.

[0150] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes a plurality of computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this specification are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, optical fiber, Digital Subscriber Line (DSL)) or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that the computer can access or a data storage device such as a server or data center that includes a plurality of available media integrated. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a Digital Versatile Disc (DVD)), or a semiconductor medium (such as a Solid State Disk (SSD)), etc.

[0151] When implemented by hardware or firmware, the foregoing method flow is programmed into a hardware circuit to obtain a corresponding hardware circuit structure and implement the corresponding function. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. The designer can program by himself to "integrate" a digital system on a PLD, without having to ask a chip manufacturer to design and manufacture a dedicated integrated circuit chip. Moreover, nowadays, instead of manually fabricating integrated circuit chips, this programming is mostly implemented using "logic compiler" software, which is similar to the software compiler used in program development and writing. The original code before compilation also has to be written in a specific programming language, which is called a Hardware Description Language (HDL), and there is not only one type of HDL, but many types. Those skilled in the art should also be clear that only by slightly logically programming the method flow with the above-mentioned several hardware description languages and programming it into an integrated circuit, it is easy to obtain a hardware circuit that implements the logic method flow.

[0152] The embodiments described above are merely described in a preferred embodiment manner of this specification, and do not limit the scope of this specification. Without departing from the design spirit of this specification, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of this specification shall fall within the protection scope determined by the claims of this specification.

Claims

1. An online non-destructive diagnosis method for elevator brake faults and a fatigue prediction method, characterized in that, Including the steps: Online collect the action data and braking times when the single-ladder brake acts. The action data includes sound data, vibration data, current data, braking distance, and load; When it is judged that an emergency stop state occurs, immediately extract the action data in the emergency stop stage and record it as the emergency stop action data; Obtain a fault diagnosis result according to the comparison between the emergency stop action data and a pre-established fault diagnosis model; When the fault diagnosis result is no fault, compare the action data with a preset brake fatigue baseline to obtain a fatigue degree prediction result of the single-ladder brake; The action data is associated with a label of the braking action generated by the brake. The label includes one or more of a stage label and a direction label. The stage label includes start, braking, and emergency stop, and the direction label includes upward and downward; The method further includes the steps: Receive single-ladder data, where the single-ladder data includes the usage year, braking times, emergency stop times, and maintenance status; Incorporate the usage year, emergency stop times, and maintenance status of the single-ladder data into the corresponding action data; Associate the braking times of the single-ladder data as a label to the corresponding action data to obtain sample data; Establish a general fatigue degree prediction model for the corresponding type of brake according to the sample data of the same type of brake, When an emergency stop occurs, superimpose a preset equivalent braking times for emergency stop on the predicted braking times of the single-ladder brake, and obtain a fatigue degree prediction result of the single-ladder brake according to the predicted braking times of the single-ladder brake; The method further includes the steps: Sort the action data of the single-ladder brake according to the time axis, obtain the action data of two single-ladder brakes with an interval of M, and extract time-frequency domain features; Obtain the predicted braking times of two single-ladder brakes respectively according to the action data of the two single-ladder brakes and the brake fatigue baseline; Calculate the difference between the predicted braking times of the two single-ladder brakes as a prediction difference, and calculate the ratio of the prediction difference to M as a proportionality coefficient; Obtain the equivalent braking times according to the proportionality coefficient.

2. The method for non-destructive online diagnosis of elevator brake faults and fatigue degree prediction according to claim 1, characterized in that The method for establishing a fault diagnosis model includes: Receive the action data associated with the fault, and group the action data according to the label; Extract the time-frequency domain features of the sound data, vibration data, and current data included in the action data of each group respectively; Compare the time-frequency domain features of the action data of each group with the time-frequency domain features of the action data in the normal state of the single-ladder respectively to obtain differential action data; Establish and use the time-frequency domain features of the differential action data to train an autoencoder model to obtain the feature expressions of the sound data, vibration data, and current data of the differential action data; After associating the feature expressions, braking distance, and load of each group with the fault, obtain the fault sample data of this group; Establish and use the fault sample data of each group to train a machine learning model respectively, and obtain the fault diagnosis model of each group according to the machine learning model.

3. The method for non-destructive online diagnosis of elevator brake faults and fatigue degree prediction according to claim 2, characterized in that The method for obtaining differential action data by comparing the time-frequency domain features of the said action data with those of the action data in the normal state of a single elevator includes: Calculate the similarity of the time-frequency domain features according to the frequency composition and its amplitude, and calculate the similarity of the action data of the associated fault and the action data in the normal state of a single elevator; When the similarity is lower than a preset threshold, obtain the differential action data according to the action data corresponding to the time-frequency domain features.

4. The method for non-destructive online diagnosis and fatigue prediction of elevator brake faults according to claim 3, characterized in that, The method for obtaining a fault diagnosis result by comparing the emergency stop action data with a pre-established fault diagnosis model includes: Input the time-frequency domain features of the sound data, vibration data and current data included in the emergency stop action data into the autoencoder model to obtain the feature expressions of the sound data, vibration data and current data; Input the feature expressions of the sound data, vibration data and current data, after associating the corresponding braking distance and load, into the fault diagnosis model to obtain the fault diagnosis result.

5. The method for non-destructive online diagnosis and fatigue prediction of elevator brake faults according to any one of claims 1 to 4, characterized in that, The method for presetting the brake fatigue baseline includes: Collect the action data and braking times of the brakes of multiple single elevators operating normally; Extract the action data in the starting stage and braking stage; The time-frequency domain features of the sound data, vibration data and current data included in the action data in the starting stage and braking stage; Establish a brake fatigue baseline according to the time-frequency domain features of the sound data, vibration data and current data, braking distance, load and braking times.

6. The method for non-destructive online diagnosis and fatigue prediction of elevator brake faults according to claim 5, characterized in that, The method for obtaining the fatigue prediction result of a single elevator brake by comparing the action data with a preset brake fatigue baseline includes: Extract the time-frequency domain features of the sound data, vibration data and current data of the action data, and read the braking distance and load of the action data; Obtain the fatigue prediction result of a single elevator brake according to the comparison of the time-frequency domain features of the sound data, vibration data and current data, braking distance and load with the brake fatigue baseline.

7. The method for non-destructive online diagnosis and fatigue prediction of elevator brake faults according to claim 6, characterized in that, The method for establishing a brake fatigue baseline according to the time-frequency domain features of the sound data, vibration data and current data, braking distance, load and braking times includes: Divide the load into several load intervals according to a preset interval, and group the time-frequency domain features and braking distances of the sound data, vibration data and current data according to the load intervals; Use the braking times as labels, and associate the time-frequency domain features and braking distances of the sound data, vibration data and current data in each group to obtain calibrated data; Establish a brake fatigue baseline for each group according to the calibrated data.

8. The method for non-destructive on-line diagnosis of elevator brake faults and fatigue prediction according to claim 7, characterized in that The method for obtaining the fatigue prediction result of the single elevator brake by comparing the time-frequency domain characteristics, braking distance, load of the sound data, vibration data and current data with the brake fatigue baseline includes: Extracting the time-frequency domain characteristics of the sound data, vibration data and current data included in the action data; Obtaining the brake fatigue baseline of the corresponding load range according to the load of the action data; Comparing the time-frequency domain characteristics of the sound data, vibration data, current data and braking distance included in the action data with the brake fatigue baseline respectively to obtain the number of braking times that match; The average value of all the matching braking times is used as the fatigue prediction result of the single elevator brake.

9. The method for non-destructive on-line diagnosis of elevator brake faults and fatigue prediction according to any one of claims 1 to 4, characterized in that The method further includes the steps: Obtaining the action data of the single elevator brake operating normally at a preset period and extracting the time-frequency domain characteristics; Obtaining the predicted braking times of the single elevator brake according to the action data and the brake fatigue baseline; Obtaining the fatigue prediction result of the single elevator brake according to the predicted braking times of the single elevator brake.

10. The method for non-destructive on-line diagnosis of elevator brake faults and fatigue prediction according to claim 9, characterized in that The method for presetting the equivalent braking times of emergency stops includes: Under laboratory conditions, measuring the brake shoe wear amount caused by N brakings and the brake shoe wear amount caused by N emergency stops at a preset load; Obtaining the equivalent braking times of emergency stops according to the ratio of the brake shoe wear amount caused by N brakings to the brake shoe wear amount caused by N emergency stops.

11. An elevator brake fault non-destructive online diagnosis and fatigue prediction system, characterized in that, Includes: An on-line acquisition module that on-line acquires the action data and the number of braking times when the single elevator brake operates, and the action data includes sound data, vibration data, current data, braking distance and load; A judgment module that immediately extracts the action data in the emergency stop stage and records it as the emergency stop action data when it judges that an emergency stop state occurs; A fault diagnosis module that obtains a fault diagnosis result according to the comparison of the emergency stop action data with a pre-established fault diagnosis model; A fatigue prediction module that, when the fault diagnosis result is no fault, compares the action data with a preset brake fatigue baseline to obtain the fatigue prediction result of the single elevator brake; The action data is associated with a label of the braking action generated by the brake, and the label includes one or more of a stage label and a direction label. The stage label includes start, brake and emergency stop, and the direction label includes up and down; The fatigue prediction module also executes the steps: Receiving single elevator data, where the single elevator data includes the usage year, the number of brakings, the number of emergency stops, and the maintenance status; Incorporating the usage year, the number of emergency stops, and the maintenance status of the single elevator data into the corresponding action data; Associating the number of brakings of the single elevator data as a label to the corresponding action data to obtain sample data; Establishing a general fatigue prediction model for the corresponding type of brake according to the sample data of the same type of brake When an emergency stop occurs, the preset equivalent braking times of the emergency stop are superimposed on the predicted braking times of the single-ladder brake, and based on the predicted braking times of the single-ladder brake, a prediction result of the fatigue degree of the single-ladder brake is obtained; The fatigue prediction module further performs the steps of: Sort the action data of the single-ladder brake according to the time axis, obtain the action data of two single-ladder brakes with an interval of M, and extract time-frequency domain features; Based on the action data of the two single-ladder brakes and the brake fatigue baseline, respectively obtain the predicted braking times of the two single-ladder brakes; Calculate the difference between the predicted braking times of the two single-ladder brakes as the prediction difference, and calculate the ratio of the prediction difference to M as the proportionality coefficient; Obtain the equivalent braking times according to the proportionality coefficient.

12. Electronic device, characterized in that, Comprising a processor and a memory; The processor is connected to the memory; The memory is used to store executable program codes; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the method according to any one of claims 1-10.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the method according to any one of claims 1-10 is implemented.

14. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, the method according to any one of claims 1-10 is implemented.

Images