Heterogeneous Elevator Fault Detection Method and System Based on Deep Reinforcement Learning Algorithm

Abstract

This application relates to a heterogeneous elevator fault detection method and system based on a deep reinforcement learning algorithm. The method includes obtaining elevator operation data and extracting corresponding operation feature variables, combining the elevator operation time, converting the data format of the operation feature variables to obtain operation feature data in a unified data format, analyzing the variable association relationship between relevant operation feature data at the same elevator operation time, and constructing a fault prediction model for heterogeneous elevators, obtaining fault feature variables and corresponding fault operation states when the elevator is abnormal, inputting the fault feature variables and corresponding fault operation states into the fault prediction model to obtain an elevator fault prediction result, performing fault determination processing on the elevator fault prediction result, and performing fault traceability processing on the elevator according to the fault determination result to obtain fault detection data for heterogeneous elevators. This application has the effect of improving the accuracy of fault detection for heterogeneous elevators.

Description

Technical Field

[0001] The present invention relates to the technical field of elevator management, and in particular, to a heterogeneous elevator fault detection method and system based on a deep reinforcement learning algorithm. Background Art

[0002] Currently, with the wide application of the Internet of Things technology in daily life, data collection of operating equipment is realized by combining the Internet of Things technology with the equipment operation terminal, so as to detect the operating state of the operating equipment.

[0003] The existing fault detection method for heterogeneous elevators usually collects the operation data of different monitoring terminals through Internet of Things devices. The data obtained by different monitoring terminals are heterogeneous data with inconsistent formats. Combining the elevator inspection data of elevator maintenance personnel, the elevator maintenance personnel conduct a manual joint evaluation between the heterogeneous data and the elevator inspection data, so as to obtain the fault detection result of the heterogeneous elevator. However, there is a data barrier between the heterogeneous data, and there are large subjective factors in the fault evaluation only through manual work. The fault detection accuracy is affected by human experience, and there is room for further optimization of the fault detection method for heterogeneous elevators. Summary of the Invention

[0004] In order to improve the fault detection accuracy of heterogeneous elevators, the present application provides a heterogeneous elevator fault detection method and system based on a deep reinforcement learning algorithm.

[0005] In the first aspect, the above-mentioned invention object of the present application is achieved through the following technical solutions:

[0006] A heterogeneous elevator fault detection method based on a deep reinforcement learning algorithm, comprising:

[0007] Obtain elevator operation data and extract corresponding operation feature variables, and combine the elevator operation time to convert the data format of the operation feature variables to obtain operation feature data with a unified data format;

[0008] Analyze the variable association relationship between relevant operation feature data at the same elevator operation time, and construct a fault prediction model for heterogeneous elevators;

[0009] Obtain the fault feature variables and corresponding fault operation states when the elevator is abnormal, and input the fault feature variables and corresponding fault operation states into the fault prediction model to obtain an elevator fault prediction result;

[0010] Perform fault determination processing on the elevator fault prediction result, and perform fault traceability processing on the elevator according to the fault determination result to obtain fault detection data of the heterogeneous elevator.

[0011] In a preferred example, the present application can be further configured as follows: analyzing the variable association relationship between relevant operation characteristic data under the same elevator operation time, and constructing a fault prediction model for heterogeneous elevators, specifically including:

[0012] Under the same elevator operation time, obtaining the elevator stop position and the target floor height in the corresponding moving direction, and analyzing the operation speed between the elevator stop position and the next target floor;

[0013] The variable association relationship between elevator-related operation characteristic data is represented by formula (1), and formula (1) is as follows:

[0014]

[0015] Among them, E(v, h) represents the elevator variable association function, a i represents the acceleration bias of the i-th movement of the elevator, v i represents the operation speed of the i-th movement of the elevator, n represents the number of movements between the elevator's current position and the target stop floor, h j represents the operation height of the j-th operation, b j represents the gravity acceleration bias corresponding to the elevator operation height h j The number of times of change in the elevator operation height in the same operation direction, ω ij represents the association weight between the j-th target floor height and the operation speed of the i-th operation;

[0016] According to the variable association relationship, calculate the activation probability of the j-th target floor height h j by formula (2), and formula (2) is as follows:

[0017]

[0018] Calculate the activation probability of the operation speed v i of the i-th operation by formula (3), and formula (3) is as follows:

[0019]

[0020] Construct a fault prediction model for heterogeneous elevators according to the variable association relationship, the activation probability of the operation speed of the corresponding variable, and the activation probability of the target floor height.

[0021] In a preferred example, the present application can be further configured as follows: obtaining the fault characteristic variables and the corresponding fault operation states when the elevator is abnormal, and inputting the fault characteristic variables and the corresponding fault operation states into the fault prediction model to obtain the elevator fault prediction result, specifically including:

[0022] Obtain the number of operating floors when the elevator is abnormal and analyze the change of the operating variable between adjacent operating floors of the elevator to obtain the fault characteristic variables of the elevator abnormality and the corresponding fault operating states;

[0023] Input the fault operating state and the fault characteristic variables into the fault prediction model, analyze the fault occurrence probability corresponding to the fault characteristic variables, and obtain the elevator fault prediction result.

[0024] In a preferred example, the present application can be further configured as follows: the inputting the fault operating state and the fault characteristic variables into the fault prediction model, analyzing the fault occurrence probability corresponding to the fault characteristic variables, and obtaining the elevator fault prediction result specifically includes:

[0025] Calculate the fault occurrence probability corresponding to the fault characteristic variables through formula (4), and formula (4) is as follows:

[0026]

[0027] where, P G represents the fault occurrence probability corresponding to the fault characteristic variables, Δh represents the operating mileage difference between the abnormal operating height of the elevator and the normal operating height h i and, Δv represents the operating speed difference between the abnormal operating speed of the elevator and the normal operating speed v i and, ΔD represents the variable difference between the elevator fault characteristic variables other than the operating height and the operating speed and the normal operating variable D i and, P(v i =1|h i ), P(h j =1|v i ) respectively represent the operating height activation probability and the operating speed activation probability, and E(v i , h i ) represents the variable correlation function.

[0028] In a preferred example, the present application can be further configured as follows: performing a fault determination process on the elevator fault prediction result, and performing a fault tracing process on the elevator according to the fault determination result to obtain the fault detection data of the heterogeneous elevator, specifically including:

[0029] Perform a fault determination on the elevator fault prediction result through formula (5), and formula (5) is as follows:

[0030]

[0031] where, ω i , ω j respectively represent the elevator operating speed weight and the elevator operating height weight, and H represents the maximum operating height of the elevator;

[0032] When the fault determination condition is not met, an elevator stacking structure is constructed in combination with the number of operating floors, and the fault feature variables are input into the elevator stacking structure to trace the fault features, so as to obtain the fault detection data of heterogeneous elevators.

[0033] By adopting the above technical solutions, during the operation of the elevator, various heterogeneous data such as the operating current, operating speed, operating mileage of the elevator, and the loading situation of personnel and goods in the elevator car are obtained, and the data format is converted into a unified data format, which is convenient for unified processing of the heterogeneous elevator data. A fault prediction model is constructed by combining the variable correlation relationship between the operating feature data, which helps to more quickly and conveniently perform fault prediction management on heterogeneous elevators. Combining the fault feature variables and the corresponding fault operating states when an operation anomaly occurs, the elevator fault prediction result is obtained and the fault determination process is carried out, which helps to verify the prediction accuracy of the elevator fault prediction result. An elevator stacking structure is constructed in combination with the number of elevator operating floors, and the fault feature variables are traced for fault source, so as to obtain the fault location of the heterogeneous elevator, and thus the fault detection data of the heterogeneous elevator is obtained, which helps to improve the fault detection accuracy of the heterogeneous elevator.

[0034] In a second aspect, the above-mentioned invention object of the present application is achieved through the following technical solutions:

[0035] A heterogeneous elevator fault detection system based on a deep reinforcement learning algorithm, comprising:

[0036] A data acquisition module, configured to acquire elevator operation data and extract corresponding operation feature variables, and combine the elevator operation time to convert the operation feature variables into operation feature data in a unified data format;

[0037] A model construction module, configured to analyze the variable correlation relationship between relevant operation feature data at the same elevator operation time, and construct a fault prediction model for heterogeneous elevators;

[0038] A fault prediction module, configured to acquire the fault feature variables and the corresponding fault operation states when the elevator is abnormal, and input the fault feature variables and the corresponding fault operation states into the fault prediction model to obtain the elevator fault prediction result;

[0039] A fault source tracing module, configured to perform fault determination processing on the elevator fault prediction result, and perform fault source tracing processing on the elevator according to the fault determination result to obtain the fault detection data of the heterogeneous elevator.

[0040] By adopting the above technical solutions, during the operation of the elevator, various heterogeneous data such as elevator operating current, operating speed, operating mileage, and the loading situation of personnel and goods in the elevator car are obtained, and through data format conversion into a unified data format, it is convenient to uniformly process the heterogeneous elevator data. By combining the variable correlation relationships among the operating characteristic data to construct a fault prediction model, it helps to more quickly and conveniently conduct fault prediction management for heterogeneous elevators. By combining the fault characteristic variables and the corresponding fault operating states when an operating anomaly occurs, the elevator fault prediction result is obtained and the fault determination process is carried out, which helps to verify the prediction accuracy of the elevator fault prediction result. And by constructing an elevator stacking structure in combination with the elevator operating floors, fault tracing processing is carried out on the fault characteristic variables to obtain the fault location of the heterogeneous elevator, thereby obtaining the fault detection data of the heterogeneous elevator, which helps to improve the fault detection accuracy of the heterogeneous elevator.

[0041] In a third aspect, the above object of the present application is achieved by the following technical solutions:

[0042] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above heterogeneous elevator fault detection method based on the deep reinforcement learning algorithm are implemented.

[0043] In a fourth aspect, the above object of the present application is achieved by the following technical solutions:

[0044] A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the steps of the above heterogeneous elevator fault detection method based on the deep reinforcement learning algorithm are implemented.

[0045] In summary, the present application includes at least one of the following beneficial technical effects:

[0046] During the operation of the elevator, various heterogeneous data such as elevator operating current, operating speed, operating mileage, and the loading situation of personnel and goods in the elevator car are obtained, and through data format conversion into a unified data format, it is convenient to uniformly process the heterogeneous elevator data. By combining the variable correlation relationships among the operating characteristic data to construct a fault prediction model, it helps to more quickly and conveniently conduct fault prediction management for heterogeneous elevators. By combining the fault characteristic variables and the corresponding fault operating states when an operating anomaly occurs, the elevator fault prediction result is obtained and the fault determination process is carried out, which helps to verify the prediction accuracy of the elevator fault prediction result. And by constructing an elevator stacking structure in combination with the elevator operating floors, fault tracing processing is carried out on the fault characteristic variables to obtain the fault location of the heterogeneous elevator, thereby obtaining the fault detection data of the heterogeneous elevator, which helps to improve the fault detection accuracy of the heterogeneous elevator. Description of the Drawings

[0047] Figure 1 It is the implementation flowchart of a heterogeneous elevator fault detection method based on the deep reinforcement learning algorithm in this embodiment.

[0048] Figure 2 It is the structural block diagram of a heterogeneous elevator fault detection system based on the deep reinforcement learning algorithm in this embodiment.

[0049] Figure 3 It is the internal structural schematic diagram of a computer device for implementing the heterogeneous elevator fault detection method. Specific implementation manner

[0050] The following further details this application in conjunction with the accompanying drawings.

[0051] In one embodiment, as Figure 1 shown, this application discloses a heterogeneous elevator fault detection method based on the deep reinforcement learning algorithm, which specifically includes the following steps:

[0052] S10: Obtain elevator operation data and extract corresponding operation feature variables, and combine the elevator operation time to convert the data format of the operation feature variables to obtain operation feature data with a unified data format.

[0053] Specifically, according to the preset detection device, collect the image information of the passengers and goods carried during the elevator operation, the operation information such as the current, speed, and mileage of the elevator operation, and the height information such as the elevator operation height and the height of the stopped floor, and extract the corresponding operation feature variables. For example, perform image processing on the image of the passengers and goods carried to obtain the actual load variable of the elevator, perform feature extraction on the operation information such as the current, speed, and mileage of the elevator operation to obtain the elevator operation variable, and combine the elevator operation time to convert all the operation feature variables at the same operation time, such as in the data format with time as the prefix + feature variable as the suffix, so as to obtain operation feature data with a unified data format.

[0054] S20: Analyze the variable correlation relationship between relevant operation feature data at the same elevator operation time, and construct a fault prediction model for the heterogeneous elevator.

[0055] Specifically, taking the elevator operation time as the correlation point, perform data association on all the operation feature data at the same elevator operation time, and perform sequential association with the elevator operation time, so as to obtain the variable correlation relationship between the operation feature data. Among them, the construction of the fault prediction model specifically includes:

[0056] S201: At the same elevator operation time, obtain the target floor height at the elevator stop position and the corresponding moving direction, and analyze the operation speed between the elevator stop position and the next target floor.

[0057] Specifically, within the same elevator operation time, obtain the elevator stop position. If the elevator stops running at a certain location for more than a preset time, it is determined as a stop, and the corresponding moving direction, such as the target floor height when moving downward or upward. The target floor height is calculated based on the distance between the floor reached by the passengers inside the elevator and the current stop position. Combining the current load of the elevator and the preset acceleration, calculate the running speed between the elevator stop position and the next target floor.

[0058] S202: Represent the variable correlation relationship between the operation characteristic data related to the elevator through formula (1), and formula (1) is as follows:

[0059]

[0060] Among them, E(v, h) represents the elevator variable correlation function, a i represents the acceleration offset of the i-th movement of the elevator, v i represents the running speed of the i-th movement of the elevator, n represents the number of movements between the elevator's current position and the target stop floor, h j represents the running height of the j-th movement, b j represents the gravitational acceleration offset corresponding to the elevator running height h j m represents the number of changes in the elevator running height in the same running direction, ω ij represents the correlation weight between the j-th target floor height and the running speed of the i-th movement.

[0061] S203: According to the variable correlation relationship, calculate the activation probability of the j-th target floor height h j through formula (2), and formula (2) is as follows:

[0062]

[0063] Calculate the activation probability of the running speed v of the i-th movement through formula (3) i and formula (3) is as follows:

[0064]

[0065] S204: Based on the variable correlation relationship and the activation probabilities of the running speeds and target floor heights of the corresponding variables, construct a fault prediction model for heterogeneous elevators.

[0066] Specifically, correlate the elevator correlation relationship with the activation probabilities of the running speeds and target floor heights of the corresponding variables, and construct a fault prediction model for heterogeneous elevators based on the correlated data structure.

[0067] S30: Obtain the fault feature variables and corresponding fault operation states when the elevator is abnormal, input the fault feature variables and corresponding fault operation states into the fault prediction model, and obtain the elevator fault prediction result.

[0068] Specifically, step S30 includes:

[0069] S301: Obtain the number of running floors of the elevator when it is abnormal and analyze the change of operation variables between adjacent running floors of the elevator to obtain the fault feature variables and corresponding fault operation states of the elevator abnormality.

[0070] Specifically, when the elevator is operating abnormally, in combination with the number of running floors at the time of the abnormality, analyze the change of operation variables between the previous running floor and the next running floor of the abnormal running floor, such as the change of elevator operation variables, the change of carrying weight, the change of running speed, etc., so as to obtain the fault feature variables and corresponding fault operation states of the elevator abnormality. The fault operation state is the parameter change outside the normal operation parameter change range.

[0071] S302: Input the fault operation state and fault feature variables into the fault prediction model, analyze the fault occurrence probability corresponding to the fault feature variables, and obtain the elevator fault prediction result.

[0072] Specifically, step S302 specifically includes:

[0073] Calculate the fault occurrence probability corresponding to the fault feature variables through formula (4), and formula (4) is as follows:

[0074]

[0075] Among them, P G represents the fault occurrence probability corresponding to the fault feature variables, Δh represents the running mileage difference between the abnormal running height of the elevator and the normal running height h i and, Δv represents the running speed difference between the abnormal running speed of the elevator and the normal running speed v i and, ΔD represents the variable difference between the elevator fault feature variables other than the running height and running speed and the normal running variable D i and, Ρ(v i =1|h i ), Ρ(h j =1|v i ) respectively represent the running height activation probability and the running speed activation probability, E(v i ,h i ) represents the variable correlation function.

[0076] S40: Perform fault determination processing on the elevator fault prediction result, perform fault tracing processing on the elevator according to the fault determination result, and obtain the fault detection data of the heterogeneous elevator.

[0077] Specifically, step S40 specifically includes:

[0078] S401: Perform fault determination on the elevator fault prediction result through formula (5), and formula (5) is as follows:

[0079]

[0080] Among them, ω i and ω j respectively represent the elevator running speed weight and the elevator running height weight, and H represents the maximum running height of the elevator.

[0081] S402: When the fault determination condition is not met, construct an elevator stacking structure in combination with the number of running floors, input the fault feature variables into the elevator stacking structure to trace the fault features, and obtain the fault detection data of heterogeneous elevators.

[0082] Specifically, when the elevator prediction result does not meet the determination condition of formula (5), that is, the fault determination condition is not met, construct an elevator stacking structure in combination with the number of running floors and the stopping probability of each stopping floor, input the fault feature variables into the elevator stacking structure, and perform reverse tracking of the fault feature variables until the variable difference of the fault feature variables is the same as the normal running variable and then end. Take the position corresponding to the last feature variable difference as the fault source of the fault feature, so as to obtain the fault detection data of heterogeneous elevators.

[0083] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0084] In one embodiment, a heterogeneous elevator fault detection system based on a deep reinforcement learning algorithm is provided. The heterogeneous elevator fault detection system based on the deep reinforcement learning algorithm corresponds one-to-one with the heterogeneous elevator fault detection method based on the deep reinforcement learning algorithm in the above embodiment. As Figure 2 shown, the heterogeneous elevator fault detection system based on the deep reinforcement learning algorithm includes a data acquisition module, a model construction module, a fault prediction module, and a fault traceability module. The detailed description of each functional module is as follows:

[0085] The data acquisition module is used to acquire elevator operation data, extract the corresponding operation feature variables, and combine the elevator operation time to convert the data format of the operation feature variables to obtain operation feature data in a unified data format.

[0086] The model construction module is used to analyze the variable correlation relationship between relevant operation feature data under the same elevator operation time and construct a fault prediction model for heterogeneous elevators.

[0087] A fault prediction module, configured to obtain fault feature variables and corresponding fault operating states when the elevator is abnormal, and input the fault feature variables and corresponding fault operating states into a fault prediction model to obtain an elevator fault prediction result.

[0088] A fault tracing module, configured to perform fault determination processing on the elevator fault prediction result, and perform fault tracing processing on the elevator according to the fault determination result to obtain fault detection data of heterogeneous elevators.

[0089] Preferably, the model construction module specifically includes:

[0090] A variable acquisition sub-module, configured to obtain the elevator stop position and the target floor height in the corresponding moving direction at the same elevator operation time, and analyze the running speed between the elevator stop position and the next target floor.

[0091] An association analysis sub-module, configured to represent the variable association relationship between the running feature data related to the elevator through formula (1), and formula (1) is as follows:

[0092]

[0093] Among them, E(v,h) represents the elevator variable association function, a i represents the acceleration offset of the i-th movement of the elevator, v i represents the running speed of the i-th movement of the elevator, n represents the number of movements between the elevator's location and the target stop floor, h j represents the running height of the j-th movement, b j represents the gravity acceleration offset corresponding to the elevator running height h j m represents the number of changes in the elevator running height in the same running direction, ω ij represents the association weight between the j-th target floor height and the running speed of the i-th movement.

[0094] A height probability calculation sub-module, configured to calculate the activation probability of the j-th target floor height h according to the variable association relationship through formula (2), and formula (2) is as follows: j The activation probability of the running speed v of the i-th movement is calculated through formula (3), and formula (3) is as follows:

[0095]

[0096] A speed probability calculation sub-module, configured to calculate the activation probability of the running speed v of the i-th movement through formula (3), and formula (3) is as follows: i The activation probability of the running speed v of the i-th movement is calculated through formula (3), and formula (3) is as follows:

[0097]

[0098] A model construction sub-module, configured to construct a fault prediction model for a heterogeneous elevator according to the variable association relationship, the activation probability of the corresponding variable's running rate, and the activation probability of the target floor height.

[0099] Preferably, the fault prediction module specifically includes:

[0100] An abnormal variable acquisition sub-module, configured to acquire the number of running floors of the elevator during an anomaly and analyze the change in the running variables between adjacent running floors of the elevator, so as to obtain the fault characteristic variables of the elevator anomaly and the corresponding fault running states.

[0101] A fault prediction sub-module, configured to input the fault running state and the fault characteristic variables into the fault prediction model, analyze the fault occurrence probability corresponding to the fault characteristic variables, and obtain the elevator fault prediction result.

[0102] Preferably, the fault prediction sub-module specifically includes:

[0103] Calculate the fault occurrence probability corresponding to the fault characteristic variables through formula (4), and formula (4) is as follows:

[0104]

[0105] Wherein, P G represents the fault occurrence probability corresponding to the fault characteristic variables, Δh represents the running mileage difference between the abnormal running height of the elevator and the normal running height h i between, Δv represents the running speed difference between the abnormal running speed of the elevator and the normal running speed v i between, ΔD represents the variable difference between the elevator fault characteristic variables other than the running height and the running speed and the normal running variable D i between, P(v i =1|h i ), P(h j =1|v i ) respectively represent the activation probability of the running height and the activation probability of the running speed, and E(v i , h i ) represents the variable association function.

[0106] Preferably, the fault tracing module specifically includes:

[0107] A fault determination sub-module, configured to perform fault determination on the elevator fault prediction result through formula (5), and formula (5) is as follows:

[0108]

[0109] Wherein, ω i , ω j respectively represent the elevator running speed weight and the elevator running height weight, and H represents the maximum running height of the elevator.

[0110] A fault tracing sub-module, which is used to construct an elevator stacking structure in combination with the number of operating floors when the fault determination condition is not met, input fault feature variables into the elevator stacking structure to perform fault feature tracing processing, and obtain fault detection data of heterogeneous elevators.

[0111] For the specific limitations of the heterogeneous elevator fault detection system based on the deep reinforcement learning algorithm, reference can be made to the limitations of the heterogeneous elevator fault detection method based on the deep reinforcement learning algorithm in the above text, which will not be elaborated here. Each module in the above heterogeneous elevator fault detection system based on the deep reinforcement learning algorithm can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules.

[0112] In one embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 3 shown. The computer device includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data during the fault detection process of heterogeneous elevators. The network interface of the computer device is used to communicate with external terminals through a network connection. When the computer program is executed by the processor, it implements a heterogeneous elevator fault detection method based on the deep reinforcement learning algorithm.

[0113] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of a heterogeneous elevator fault detection method based on the deep reinforcement learning algorithm.

[0114] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0115] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

[0116] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A heterogeneous elevator fault detection method based on a deep reinforcement learning algorithm, characterized in that, including: Obtain elevator operation data and extract corresponding operation characteristic variables. Combine with the elevator operation time, and convert the data format of the operation characteristic variables to obtain operation characteristic data in a unified data format; Analyze the variable correlation relationship between relevant operation characteristic data at the same elevator operation time, and construct a fault prediction model for heterogeneous elevators; Obtain the fault characteristic variables and corresponding fault operation states when the elevator is abnormal, input the fault characteristic variables and corresponding fault operation states into the fault prediction model, and obtain the elevator fault prediction result; Perform fault determination processing on the elevator fault prediction result, and perform fault traceability processing on the elevator according to the fault determination result to obtain fault detection data for heterogeneous elevators; Among them, the step of analyzing the variable correlation relationship between relevant operation characteristic data at the same elevator operation time and constructing a fault prediction model for heterogeneous elevators specifically includes: At the same elevator operation time, obtain the elevator stop position and the target floor height in the corresponding moving direction, and analyze the operation speed between the elevator stop position and the next target floor; Represent the variable correlation relationship between elevator-related operation characteristic data by formula (1), and formula (1) is as follows: （1）; Among them, represents the elevator variable correlation function, represents the acceleration offset of the nth movement of the elevator, represents the running speed of the nth movement of the elevator, represents the number of movements between the position of the elevator and the target stop floor, represents the running height of the nth operation, represents the gravitational acceleration offset corresponding to the elevator running height represents the number of changes in the elevator running height in the same running direction, represents the correlation weight between the height of the nth target floor and the running speed of the nth operation; According to the variable association relationship, calculate the activation probability of the height of the th target floor through formula (2) as follows: （2）; Calculate the running rate of the th run and the activation probability of using Equation (3) as follows: （3）; Construct a fault prediction model for heterogeneous elevators according to the variable correlation relationship and the activation probabilities of the operation speeds of the corresponding variables and the activation probability of the target floor height.

2. The heterogeneous elevator fault detection method based on the deep reinforcement learning algorithm according to claim 1, characterized in that, The step of obtaining the fault characteristic variables and corresponding fault operation states when the elevator is abnormal, inputting the fault characteristic variables and corresponding fault operation states into the fault prediction model, and obtaining the elevator fault prediction result specifically includes: Obtain the number of operating floors of the elevator when it is abnormal and analyze the change of operation variables between adjacent operating floors of the elevator to obtain the fault characteristic variables and corresponding fault operation states of the elevator abnormality; Input the fault operation state and the fault characteristic variables into the fault prediction model, analyze the fault occurrence probability corresponding to the fault characteristic variables, and obtain the elevator fault prediction result.

3. The heterogeneous elevator fault detection method based on the deep reinforcement learning algorithm according to claim 2, characterized in that, The step of inputting the fault operation state and the fault characteristic variables into the fault prediction model, analyzing the fault occurrence probability corresponding to the fault characteristic variables, and obtaining the elevator fault prediction result specifically includes: Calculate the fault occurrence probability corresponding to the fault characteristic variables by formula (4), and formula (4) is as follows: （4）; Among them, represents the probability of a fault occurring corresponding to the fault feature variable, represents the mileage difference between the abnormal running height and the normal running height of the elevator among them, represents the running speed difference between the abnormal running speed and the normal running speed of the elevator among them, represents the variable difference between the elevator fault feature variable other than the running height and the running speed and the normal running variable among them, and respectively represent the activation probability of the running height and the activation probability of the running speed, represents the variable correlation function.

4. The heterogeneous elevator fault detection method based on the deep reinforcement learning algorithm according to claim 1, wherein The step of performing fault determination processing on the elevator fault prediction result, performing fault traceability processing on the elevator according to the fault determination result, and obtaining fault detection data for heterogeneous elevators specifically includes: Perform fault determination on the elevator fault prediction result by formula (5), and formula (5) is as follows: （5）; Among them, and represent the elevator running speed weight and the elevator running height weight respectively, represents the maximum running height of the elevator; When the fault determination condition is not met, construct an elevator stacking structure in combination with the number of operating floors, input the fault characteristic variables into the elevator stacking structure to perform fault traceability processing on the fault characteristics, and obtain the fault detection data for heterogeneous elevators.

5. A heterogeneous elevator fault detection system based on a deep reinforcement learning algorithm, characterized in that, including: A data acquisition module for obtaining elevator operation data and extracting corresponding operation characteristic variables. Combine with the elevator operation time, and convert the data format of the operation characteristic variables to obtain operation characteristic data in a unified data format; A model construction module, configured to analyze the variable correlation relationship between relevant operation feature data at the same elevator operation time, and construct a fault prediction model for heterogeneous elevators; A fault prediction module, configured to obtain fault feature variables and corresponding fault operation states when the elevator is abnormal, and input the fault feature variables and corresponding fault operation states into the fault prediction model to obtain an elevator fault prediction result; A fault tracing module, configured to perform fault determination processing on the elevator fault prediction result, and perform fault tracing processing on the elevator according to the fault determination result to obtain fault detection data of heterogeneous elevators; Wherein, the model construction module specifically includes: A variable acquisition sub-module, configured to obtain the elevator stop position and the target floor height in the corresponding moving direction at the same elevator operation time, and analyze the operation speed between the elevator stop position and the next target floor; An association analysis sub-module, configured to represent the variable correlation relationship between the operation feature data related to the elevator by formula (1), and formula (1) is as follows: （1）; Among them, represents the elevator variable correlation function, represents the acceleration offset of the th movement of the elevator, represents the running speed of the th movement of the elevator, represents the number of movements between the position where the elevator is located and the target stopping floor, represents the th running height of the operation, represents the elevator running height corresponding gravitational acceleration offset, represents the number of changes in the elevator running height in the same running direction, represents the th target floor height and the correlation weight between the running speed of the th operation; A high-probability calculation sub-module, which is used to calculate the activation probability of the height of the th target floor according to the variable association relationship through formula (2). The formula (2) is as follows: The activation probability of the height of the （2）; A rate probability calculation sub-module, which is used to calculate the running rate of the th run through formula (3), and the activation probability of is as shown in formula (3): （3）; Construct a fault prediction model for heterogeneous elevators according to the variable correlation relationship, the activation probability of the operation speed of the corresponding variable, and the activation probability of the target floor height.

6. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, the steps of the heterogeneous elevator fault detection method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, the steps of the heterogeneous elevator fault detection method according to any one of claims 1 to 4 are implemented.

Images