Fault remote diagnosis maintenance method and device, equipment and medium

Abstract

This application provides a method, apparatus, equipment, and medium for remote fault diagnosis and maintenance, belonging to the field of fault diagnosis technology. The method includes: obtaining equipment operation characteristic data based on vending machine operation data; processing the equipment operation characteristic data using a lightweight terminal detection model corresponding to the deployment scenario to obtain anomaly judgment results; selectively extracting and encapsulating the equipment operation characteristic data corresponding to the anomaly judgment results to generate anomaly characteristic data packets, uploading them to the cloud based on the terminal network connectivity status; after the cloud parses and verifies the anomaly characteristic data, obtaining a cloud-based set of anomaly characteristics to be diagnosed and equipment scenario identification data; determining the root cause location result of the fault by combining the set of anomaly characteristics to be diagnosed with the equipment cluster operation baseline data using a full-scale detection model corresponding to the equipment scenario identification; and determining the fault handling level and maintaining the vending machine based on the root cause location result. This application can improve the accuracy of remote operation and maintenance of vending machines.

Description

Technical Field

[0001] This application belongs to the field of fault diagnosis technology, and more specifically, relates to a method, device, equipment and medium for remote fault diagnosis and maintenance. Background Technology

[0002] As core self-service terminals in new retail scenarios, vending machines have been widely deployed in diverse settings such as supermarkets, communities, and transportation hubs. With the large-scale popularization of these machines, traditional manual on-site inspections and fault repair methods are no longer adequate for the efficient operation and maintenance needs of a massive number of terminals. Remote fault diagnosis and maintenance technology has become a core direction for the industry's intelligent upgrade. Currently, mainstream remote fault diagnosis solutions in the industry mostly employ centralized cloud analysis or a single detection model to perform the entire process of fault detection and location. These solutions suffer from mismatches between the detection model and the computing power of the operating environment, and low efficiency in the initial judgment of terminal-side anomalies. These shortcomings directly lead to unclear fault maintenance hierarchy divisions and unreasonable handling strategies, hindering the efficient, accurate, and automated remote operation and maintenance of large-scale vending machine terminals and restricting the intelligent development of the self-service vending equipment operation and maintenance system. Summary of the Invention

[0003] The purpose of this application is to provide a method, apparatus, device, and medium for remote fault diagnosis and maintenance, so as to improve the accuracy of remote operation and maintenance of vending machines. To achieve the above objective, the technical solution provided by this application is as follows: Firstly, a remote fault diagnosis and maintenance method is provided, including: Obtain the device operation data and deployment scenario data reported by the target vending machine terminal, standardize and extract features from the device operation data to obtain device operation feature data; Determine the lightweight detection model deployed on the target vending machine terminal corresponding to the deployment scenario data; obtain anomaly judgment results through the lightweight detection model based on the device operation characteristic data; The device operation characteristic data corresponding to the anomaly judgment result is selectively intercepted and encapsulated to obtain an anomaly characteristic data packet; the anomaly characteristic data packet is uploaded to the cloud based on the terminal network connectivity status data; the cloud is used to parse and verify the anomaly characteristic data to obtain the cloud-based anomaly feature set to be diagnosed and the device scene identifier; Determine the full-scale detection model and equipment cluster operation baseline data corresponding to the equipment scene identifier; based on the cloud-based anomaly feature set to be diagnosed and the equipment cluster operation baseline data, determine the root cause location results of the fault through the full-scale detection model; Based on the root cause location results, the fault handling level is determined, and the target vending machine is maintained based on the fault handling level. The fault handling level can be a local self-healing fault level, a cloud-based remote repair fault level, or an on-site handling fault level.

[0004] Secondly, a remote fault diagnosis and maintenance device is provided, comprising: The device data acquisition module is used to acquire device operation data and deployment scenario data reported by the target vending machine terminal, standardize and extract features from the device operation data to obtain device operation feature data. The lightweight detection module is used to determine the lightweight detection model deployed on the target vending machine terminal corresponding to the deployment scenario data; based on the device operation characteristic data, the lightweight detection model is used to obtain the anomaly judgment result; The cloud module is used to selectively extract and encapsulate the device operation characteristic data corresponding to the anomaly judgment result to obtain an anomaly characteristic data packet; based on the terminal network connectivity status data, the anomaly characteristic data packet is uploaded to the cloud; the cloud is used to parse and verify the anomaly characteristic data to obtain the cloud-based anomaly feature set to be diagnosed and the device scene identifier; The full-scale detection module is used to determine the full-scale detection model and equipment cluster operation baseline data corresponding to the equipment scene identifier; based on the cloud-based abnormal feature set to be diagnosed and the equipment cluster operation baseline data, the full-scale detection model is used to determine the root cause location of the fault. The equipment maintenance module is used to determine the fault handling level based on the fault root cause location results, and to maintain the target vending machine based on the fault handling level; the fault handling level can be a local self-healing fault level, a cloud-based remote repair fault level, or an on-site handling fault level.

[0005] Thirdly, embodiments of this application also provide an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the remote fault diagnosis and maintenance method provided by any possible implementation of the first aspect.

[0006] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the remote fault diagnosis and maintenance method provided by any possible implementation of the first aspect.

[0007] The beneficial effects of the technical solution provided in this application are as follows: The remote fault diagnosis and maintenance method, apparatus, equipment, and medium provided in this application embodiment, compared with related technologies: This application adopts a layered architecture of a lightweight terminal detection model and a full-scale cloud detection model. The lightweight detection model is adapted to the hardware computing power of the vending machine terminal and completes the real-time anomaly judgment of the device operation data locally on the terminal. This fundamentally solves the problems of mismatch between the detection model and the computing power of the terminal operating environment and low efficiency of the initial anomaly judgment on the terminal side in the existing technology, and ensures the real-time nature of anomaly identification.

[0008] Meanwhile, the cloud-based system leverages a comprehensive detection model combined with baseline data from the equipment cluster's operation to accurately pinpoint the root cause of faults. Based on the location results, it clearly categorizes fault handling into three levels: local self-healing, remote cloud repair, and on-site intervention. This significantly improves upon the shortcomings of existing solutions, such as ambiguous fault maintenance levels and unreasonable handling strategies. Furthermore, by adaptively uploading abnormal data based on terminal network status, it ensures reliable transmission of abnormal information and optimizes data interaction efficiency.

[0009] The embodiments of this application can realize remote fault diagnosis and hierarchical handling throughout the entire process, reduce manual operation and maintenance costs, shorten fault response and processing time, adapt to the operation and maintenance needs of large-scale vending machine terminals in multiple scenarios such as supermarkets and communities, effectively improve equipment operation stability, and promote the upgrade of the self-service vending terminal operation and maintenance system towards high efficiency and automation. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below.

[0011] Figure 1 A flowchart illustrating the remote fault diagnosis and maintenance method provided in this application embodiment; Figure 2 This is a structural block diagram of the remote fault diagnosis and maintenance device provided in the embodiments of this application; Figure 3 A schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0012] The embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the embodiments described below with reference to the accompanying drawings are exemplary descriptions for explaining the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions of the embodiments of this application.

[0013] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the terms “comprising” and “including” as used in embodiments of this application mean that the corresponding feature can be implemented as the presented feature, information, data, step, operation, element, and / or component, but do not exclude implementation as other features, information, data, step, operation, element, component, and / or combinations thereof supported by the art. It should be understood that when we say that an element is “connected” or “coupled” to another element, the one element can be directly connected or coupled to the other element, or it can mean that the one element and the other element establish a connection relationship through an intermediate element. Furthermore, “connected” or “coupled” as used herein can include wireless connection or wireless coupling. The term “and / or” as used herein indicates at least one of the items defined by the term; for example, “A and / or B” can be implemented as “A,” or as “B,” or as “A and B.” When describing multiple (two or more) items, if the relationship between the multiple items is not explicitly defined, the multiple items can refer to one, several or all of the multiple items. For example, the description of "parameter A includes A1, A2, A3" can be implemented as parameter A includes A1 or A2 or A3, or it can be implemented as parameter A includes at least two of the three items A1, A2 and A3.

[0014] It is understood that in the embodiments of this application, data such as user information are involved. When the embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with relevant laws, regulations and standards.

[0015] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0016] This application provides a method for remote fault diagnosis and maintenance, which can be executed by a remote fault diagnosis and maintenance platform that can be deployed on cloud resources. Figure 1 As shown, the method may include: S101: Obtain the device operation data and deployment scenario data reported by the target vending machine terminal, standardize the device operation data and extract features to obtain device operation feature data.

[0017] In this embodiment, the target vending machine terminal refers to a self-service vending hardware device with data collection and reporting capabilities, such as a beverage vending machine or a snack vending machine; the device operation data refers to the operating status data of various components of the terminal, such as motor speed, temperature and humidity, and the status of the delivery channel; the deployment scenario data refers to the actual installation environment data of the terminal, such as supermarkets, communities, and transportation hubs; standardization refers to the unified format processing of the device operation data; feature extraction refers to the process of extracting effective features from the device operation data; and the device operation feature data refers to the analysis data obtained after standardization and feature extraction, which is used to characterize the terminal's operating status.

[0018] For example, in this embodiment, the core component operation data, environmental perception data and equipment condition data of the terminal can be collected in real time through the sensors, controllers and other acquisition modules built into the target vending machine terminal to form the original equipment operation data. At the same time, the target vending machine terminal can obtain its own deployment scenario data through pre-configuration input or distribution from the back-end management system. Both types of data are reported to the back-end processing terminal through the terminal's communication module at a preset cycle.

[0019] This embodiment can standardize the reported equipment operation data. Specifically, this embodiment can first perform preprocessing on the data, such as filling in missing values ​​and removing outliers, and then normalize the operation data of different dimensions and units to unify the data format and numerical range, eliminate the differences in units between data, and ensure the comparability of data.

[0020] This embodiment allows for feature extraction after standardization. Specifically, based on the operating characteristics of the vending machine, this embodiment can extract statistical features such as mean, variance, and peak value from the time domain dimension, and extract operating trend features and start / stop status features of core components from the operating condition dimension, thereby filtering out feature indicators that are strongly correlated with the operating status of the equipment.

[0021] This embodiment can integrate and structure the extracted effective features according to preset feature dimension rules to form structured equipment operation feature data, complete the data preprocessing process, and provide a standardized and highly effective analytical data foundation for subsequent anomaly detection.

[0022] This embodiment eliminates format differences and invalid information in the original data by standardizing and extracting features from the data reported by the target vending machine terminal, thereby improving the effectiveness and comparability of the data. The resulting equipment operation feature data can accurately characterize the actual operating status of the equipment, providing reliable data support for subsequent anomaly detection and ensuring the accuracy and rationality of subsequent diagnostic processes.

[0023] S102: Determine the lightweight detection model deployed on the target vending machine terminal corresponding to the deployment scenario data; based on the device operation characteristic data, obtain the anomaly judgment result through the lightweight detection model.

[0024] In this embodiment, the lightweight detection model is obtained by training an isolated forest model on a historical equipment operation feature dataset corresponding to multiple high-frequency fault types; high-frequency faults are fault types whose fault frequency is higher than a preset fault frequency threshold; based on the equipment operation feature data, the lightweight detection model obtains anomaly judgment results, including: performing anomaly inference calculation on the equipment operation feature data through the lightweight detection model to obtain real-time anomaly score data; comparing and judging the real-time anomaly score data with the scenario-based anomaly threshold data to obtain anomaly judgment result data; the scenario-based anomaly threshold data is determined based on the deployment scenario data.

[0025] In this embodiment, the lightweight detection model refers to a fault detection model adapted to the computing power of the vending machine terminal, used for initial local anomaly judgment, such as a simplified isolated forest model; high-frequency fault types refer to fault categories that occur frequently during vending machine operation, such as jammed product aisles or unresponsive barcode scanning modules; historical equipment operation feature dataset refers to a set of feature data collected and processed during the historical operation of the vending machine, such as the operation feature data of terminals in the same scenario over the past year. The isolated forest model refers to an unsupervised anomaly detection model built based on isolated trees, used to identify abnormal features in the data; fault frequency refers to the number of times a certain type of fault occurs within a preset period; the preset fault frequency threshold refers to a fault frequency judgment benchmark set based on maintenance data, such as 5 occurrences per month as the threshold. Anomaly inference calculation refers to the model's anomaly identification calculation process on the input data; real-time anomaly score data refers to the numerical value representing the degree of data anomaly output by the model after calculation, used to determine whether the device is abnormal; scenario-based anomaly threshold data refers to anomaly judgment numerical benchmarks adapted to different deployment scenarios, such as different thresholds for supermarket scenarios and community scenarios; anomaly judgment result data refers to the conclusion of the device's operating status obtained after comparing the thresholds, such as normal or abnormal.

[0026] For example, this embodiment can first train and build a lightweight detection model. Specifically, this embodiment can first obtain a historical equipment operation feature dataset of vending machines in multiple scenarios. Simultaneously, this embodiment can count the number of occurrences of various faults within a preset statistical period to obtain the fault frequency of each type of fault. This embodiment can compare the fault frequency with a preset fault frequency threshold to filter out high-frequency fault types with a fault frequency higher than the threshold. This embodiment can extract the historical equipment operation feature dataset corresponding to the high-frequency fault types as training samples to train the isolated forest model. During training, the model structure is simplified and optimized, redundant computational branches and feature dimensions are removed, and the core structure strongly correlated with high-frequency fault detection is retained. After completing the training of the lightweight detection model, this embodiment can associate and store data from different deployment scenarios with the corresponding lightweight detection models, establishing a matching mapping library between scenarios and models. Simultaneously, based on the historical operation data of vending machines in each scenario, it can calibrate scenario-specific anomaly threshold data suitable for that scenario and associate it with the corresponding model to store it on the terminal or backend side.

[0027] This embodiment can first perform data acquisition and preprocessing. Specifically, this embodiment can acquire equipment operation data through various acquisition modules of the target vending machine terminal. This embodiment can acquire deployment scenario data through pre-configuration or background distribution. This embodiment can sequentially perform standardization and feature extraction processing on the equipment operation data. During the standardization process, missing value imputation, outlier removal, and unit unification are completed. During the feature extraction process, effective features related to equipment fault detection are selected, and finally, structured equipment operation feature data is obtained, ensuring that the data matches the input dimensions of the lightweight detection model.

[0028] This embodiment can determine the matching between the deployment scenario and the lightweight detection model. Specifically, after obtaining the deployment scenario data of the target vending machine terminal, this embodiment can retrieve the established scenario-model matching mapping library, search the mapping library according to the deployment scenario data, and determine the lightweight detection model that should be deployed on the target vending machine terminal. If the terminal is deploying for the first time, the matched lightweight detection model is distributed to the target vending machine terminal via the network to complete the local deployment of the model. If the terminal has already deployed the model, the matching between the model and the scenario is verified. If they do not match, the model is updated and replaced.

[0029] This embodiment utilizes a lightweight detection model to perform anomaly inference calculations. Specifically, preprocessed device operation feature data can be input into a lightweight detection model deployed locally on the target vending machine terminal. Based on its core detection structure, the lightweight detection model performs dimension-by-dimensional anomaly identification calculations on the input device operation feature data. By calculating the degree of isolation of the data in the feature space, it obtains real-time anomaly scores characterizing the degree of anomaly in the device operation feature data. The calculation process is completed locally on the terminal, without relying on cloud computing power, ensuring real-time performance. Furthermore, the streamlined structure of the lightweight detection model is adapted to the terminal's hardware computing power, avoiding computational lag caused by insufficient computing power.

[0030] This embodiment can determine and judge scenario-based anomaly thresholds. Specifically, based on the deployment scenario data of the target vending machine terminal, this embodiment can retrieve the corresponding scenario-based anomaly threshold data from the associated stored threshold library. This embodiment can compare the real-time anomaly score data output by the lightweight detection model with the scenario-based anomaly threshold data. If the real-time anomaly score data is higher than the scenario-based anomaly threshold data, the device operating status is determined to be abnormal; if the real-time anomaly score data is lower than or equal to the scenario-based anomaly threshold data, the device operating status is determined to be normal. This embodiment can organize the final judgment conclusion into structured anomaly judgment result data, store it locally on the terminal, and report it to the cloud as needed, providing a basic judgment basis for subsequent fault handling.

[0031] This embodiment enables dynamic adaptation and optimization of the model and thresholds. Specifically, when the deployment scenario of the target vending machine terminal changes, the terminal reports new deployment scenario data. This embodiment can re-match the corresponding lightweight detection model and scenario-based anomaly threshold data, and send the updated model and thresholds to the terminal. Simultaneously, when the vending machine's operating conditions change, this embodiment can continuously collect new historical equipment operating feature datasets, incrementally train the lightweight detection model, and recalibrate the scenario-based anomaly threshold data to ensure that the model and thresholds always adapt to the actual operating state of the equipment.

[0032] This embodiment achieves precise scenario-based deployment of the model by matching deployment scenario data with a lightweight detection model. The lightweight detection model is trained and optimized based on a dataset of high-frequency fault types, adapting to the computing power of the vending machine terminal and improving the efficiency of local anomaly inference. This embodiment combines scenario-based anomaly threshold data with real-time anomaly score data for comparison and judgment, improving the scenario adaptability and accuracy of anomaly judgment results, providing a reliable initial judgment basis for subsequent fault handling. Simultaneously, the local computing mode reduces network transmission dependence, further ensuring the real-time performance of anomaly detection.

[0033] S103: Targeted interception and encapsulation of device operation characteristic data corresponding to the anomaly judgment result to obtain an anomaly characteristic data packet; upload the anomaly characteristic data packet to the cloud based on the terminal network connectivity status data; the cloud is used to parse and verify the anomaly characteristic data to obtain the cloud-based anomaly feature set to be diagnosed and the device scene identifier.

[0034] In this embodiment, the device operation characteristic data corresponding to the anomaly determination result is selectively extracted and encapsulated to obtain an anomaly characteristic data packet, including: Based on the anomaly detection results, determine the time node where the anomaly occurred and the preset feature extraction period; Based on the time node of the anomaly occurrence and the preset feature extraction period, the equipment operation feature data is selectively extracted to obtain the feature subset data of the anomaly period; The abnormal period feature subset data is standardized and encapsulated to obtain an abnormal feature data package.

[0035] In this embodiment, uploading abnormal feature data packets to the cloud based on terminal network connectivity status data includes: The terminal network connectivity status determination result is determined based on the terminal network connectivity status data. If the terminal network connectivity status is determined to be connected, then abnormal characteristic data packets are uploaded to the cloud in real time. If the terminal's network connectivity status is determined to be disconnected, the abnormal feature data packets are locally persistently cached to obtain locally cached abnormal feature data packets; network detection data is obtained, and the upload trigger data after network recovery is determined based on the network detection data and the locally cached abnormal feature data packets; based on the upload trigger data, the locally cached abnormal feature data packets are re-uploaded to the cloud after network recovery.

[0036] In this embodiment, targeted interception refers to the operation of extracting device operation feature data based on a specific time dimension, such as intercepting feature data for a preset period before and after an anomaly occurs; encapsulation refers to the process of formatting and integrating target data. Anomaly feature data packets refer to a structured set of anomaly data formed after targeted interception and encapsulation, used for uploading to the cloud for analysis; terminal network connectivity status data refers to relevant data characterizing the connection status between the terminal and the network, such as whether the network connection is normal and network latency. The cloud refers to the hardware resources / infrastructure layer that provides computing power, storage, and network transmission for the remote fault diagnosis and maintenance platform; parsing refers to the operation of disassembling and extracting the content of data packets; verification refers to the process of verifying the integrity and compliance of the data. The cloud-based anomaly feature set to be diagnosed refers to the anomaly feature data obtained after cloud parsing and verification that can be directly used for fault diagnosis; device scenario identifier refers to the unique identifier information characterizing the vending machine deployment scenario, used to match the cloud-based diagnostic model. The anomaly occurrence time node refers to the specific time at which the device anomaly is determined. Preset feature extraction period refers to a fixed data extraction time range based on fault characteristics, such as 5 minutes before to 10 minutes after an anomaly occurs. Anomaly period feature subset data refers to fault-related feature data obtained after targeted extraction; standardized encapsulation refers to an encapsulation method that integrates data according to a unified format and adds basic identification information. Terminal network connectivity status determination result refers to the conclusion of whether the terminal network is connected or not; real-time cloud upload processing refers to the operation of directly sending data packets to the cloud when the network is normal. Local persistent caching processing refers to the operation of storing data packets to the terminal's local storage medium; local cached anomaly feature data packets refer to anomaly feature data packets stored locally on the terminal. Network detection data refers to network status data collected periodically by the terminal; upload triggered data refers to network recovery determination data that triggers the retransmission of locally cached data packets; cloud retransmission processing after network recovery refers to the operation of sending locally cached data packets to the cloud when the network recovers.

[0037] For example, this embodiment can first perform a targeted extraction operation of abnormal feature data. Specifically, after obtaining the anomaly determination result, this embodiment can extract the specific time information of the device anomaly from the result to obtain the anomaly occurrence time node. This time node is synchronized with the generation time of the anomaly determination result, and a pre-configured preset feature extraction period is retrieved. This period is based on the characteristic performance cycle of various high-frequency faults of vending machines. This embodiment can combine historical operation and maintenance data with fault diagnosis requirements to define and store it locally on the terminal. Subsequently, this embodiment can use the anomaly occurrence time node as a benchmark and the preset feature extraction period as a range to perform targeted extraction of the device operation feature data stored locally in the time dimension, remove redundant data unrelated to the anomaly, and retain only the key feature data before and after the fault occurrence to obtain the feature subset data of the anomaly period, ensuring a strong correlation between the data and fault diagnosis.

[0038] This embodiment performs standardized encapsulation processing on a subset of feature data from abnormal time periods. Specifically, this embodiment uses the subset of feature data from abnormal time periods as the core content, according to a preset unified data encapsulation format. This embodiment integrates basic information such as the unique device identifier of the target vending machine, the time node of the anomaly occurrence, and the device scene identifier, and performs structured arrangement and format regularization of various types of data to generate an abnormal feature data package after standardized encapsulation. This data package adopts a common data transmission format to ensure that the cloud can directly recognize and parse it. Simple format verification is performed on the data during the encapsulation process to avoid data format disorder problems.

[0039] This embodiment can determine the network connectivity status of a terminal. Specifically, the terminal uses its built-in network communication module to collect data such as network connection status, network latency, and packet loss rate in real time, forming terminal network connectivity status data. This embodiment can perform real-time analysis on the terminal network connectivity status data. If the network connection status is "connected" and both latency and packet loss rate are within a preset normal range, the terminal network connectivity status is determined to be connected; if the network connection status is "disconnected," or latency and packet loss rate exceed the preset normal range, the terminal network connectivity status is determined to be "disconnected."

[0040] This embodiment can perform the uploading process of abnormal feature data packets based on network connectivity status. Specifically, if the determination result is that the network is connected, the terminal directly sends the abnormal feature data packets to the designated data receiving port in the cloud through the established network communication link, performs real-time cloud upload processing, and provides real-time feedback on the upload status until it receives a reception confirmation message from the cloud; if the determination result is that the network is not connected, the terminal immediately starts local persistent caching processing, stores the abnormal feature data packets in the terminal's local solid-state storage medium in chronological order, forming a local cached abnormal feature data packet, and adds a unique cache identifier to the data packet to avoid data overwriting or loss.

[0041] This embodiment can perform data packet retransmission processing after network recovery. Specifically, when the network is not connected, this embodiment can acquire the terminal's network status data at a fixed detection period to form network detection data. This fixed detection period can be preset to 1 minute to 5 minutes based on maintenance requirements. When the network detection data meets the preset network recovery judgment conditions, i.e., the network connection is restored to normal and the communication status is stable, upload trigger data is generated. This embodiment can immediately trigger the retransmission process based on the upload trigger data, retrieving local cached abnormal characteristic data packets with cache identifiers from the local storage medium. This embodiment can send them to the cloud sequentially according to the cache time, perform cloud retransmission processing after network recovery, and after the retransmission is completed, the terminal deletes the corresponding data packets from local storage and records the retransmission log.

[0042] This embodiment enables cloud-based parsing and verification of abnormal feature data packets. Specifically, after obtaining the abnormal feature data packet, the cloud first performs a data verification operation. By verifying the integrity check code, data field length, and format compliance of the data packet, it verifies whether the data packet has been lost during transmission, has an incorrect format, or other issues. If the verification fails, the cloud sends a retransmission command to the terminal; if the verification passes, the cloud performs a structured decomposition of the data packet, extracting core information such as the abnormal time period feature subset data, device scene identifier, and device unique identifier. The cloud can then organize and categorize the abnormal time period feature subset data according to feature dimensions to obtain the cloud-based abnormal feature set to be diagnosed. Simultaneously, it extracts and retains the device scene identifier, providing a basis for subsequent matching with the cloud-based full-scale detection model.

[0043] This embodiment selectively extracts abnormal device operational characteristic data, eliminating redundant data, reducing data transmission volume, and improving network transmission efficiency. Standardized encapsulation ensures data structure and universality, facilitating cloud-based parsing and processing. This embodiment executes differentiated upload strategies based on the terminal's network connectivity status: uploading immediately when the network is normal, and persistently caching locally and automatically re-uploading upon recovery when the network is abnormal. This fundamentally avoids data loss due to network interruptions, ensuring the integrity of fault diagnosis data. Cloud-based data packet parsing and verification effectively filters invalid and erroneous data, ensuring the accuracy of the cloud-based abnormal feature set and device scenario identification. This provides reliable and effective data support for subsequent cloud-based fault root cause localization, adapting to the network environment characteristics of vending machines in various scenarios and improving the stability and reliability of fault data transmission.

[0044] S104: Determine the full-scale detection model and equipment cluster operation baseline data corresponding to the equipment scene identifier; based on the cloud-based abnormal feature set to be diagnosed and the equipment cluster operation baseline data, determine the root cause location result of the fault through the full-scale detection model.

[0045] In this embodiment, the full-scale detection model is obtained by training the isolated forest model based on the historical cloud-based anomaly feature set to be diagnosed and the baseline operating data of all device clusters; based on the cloud-based anomaly feature set to be diagnosed and the baseline operating data of the device clusters, the full-scale detection model determines the root cause location results of the fault, including: The full-scale detection model is used to align the features of the cloud-based anomaly feature set to be diagnosed and the baseline data of the device cluster operation. The difference between the feature-aligned cloud-based anomaly feature set to be diagnosed and the baseline data of the device cluster operation is calculated to obtain the difference comparison data. By using a full-scale detection model to perform fault feature matching and inference on the difference comparison data, a set of candidate fault root causes data is obtained. The confidence score of each candidate root cause is determined based on the candidate root cause set data. Based on the confidence scores of each candidate root cause and the scenario-based root cause confidence threshold, the root cause localization results are determined; the scenario-based root cause confidence threshold is determined based on the deployment scenario data.

[0046] In this embodiment, the full-scale detection model refers to a complete fault detection model adapted to the high computing power of the cloud, used for accurate fault root cause localization, such as an isolated forest model that retains the full structure; the equipment cluster operation baseline data refers to the benchmark data of normal operation characteristics of vending machine clusters in the same scenario, such as the average normal operation characteristics of core components of vending machines in supermarket scenarios. Feature alignment refers to the operation of unifying the feature dimensions and formats of different data; difference calculation refers to the process of quantitatively analyzing the feature differences between data; difference comparison data refers to the dataset representing the difference between abnormal features and baseline features. Fault feature matching inference refers to the operation process of the model matching the difference data with the fault feature library; candidate fault root cause set data refers to the set of potential fault root causes matched by the model, such as channel jamming, unstable power supply, etc. Confidence score refers to the numerical value representing the matching degree of each candidate fault root cause; scenario-based fault root cause confidence threshold refers to the numerical benchmark for root cause determination adapted to different scenarios; fault root cause localization result data refers to the final fault root cause conclusion determined after threshold filtering.

[0047] For example, this embodiment can first complete the training and construction of a full-scale detection model. Specifically, this embodiment can first collect historical cloud-based abnormal feature sets to be diagnosed from vending machines in multiple scenarios, and simultaneously collect the baseline operating data of the device cluster of all vending machines in the corresponding scenarios, integrating the two types of data into a full-scale training dataset. This embodiment can perform full-process training of the isolated forest model based on the full-scale training dataset, preserving the model's complete detection structure and full-dimensional fault feature matching capability. After training, this embodiment can associate and store different device scenario identifiers with the corresponding full-scale detection models, establishing a cloud-based scenario-model matching library. At the same time, this embodiment can calibrate the scenario-specific fault root cause confidence threshold adapted to the scenario based on the historical fault diagnosis data of each scenario, and associate and store it with the model.

[0048] This embodiment can acquire and match the operational baseline data of a device cluster. Specifically, based on the received device scene identifier, the cloud retrieves the operational baseline data of the corresponding scene and the same batch of vending machines from the preset cluster baseline database. This operational baseline data is obtained by the cloud continuously collecting operational characteristic data of vending machines operating normally in the same scene, and performing statistical calculations such as mean and variance to ensure that the data can represent the normal operating status of the devices within the scene.

[0049] This embodiment can perform feature alignment and difference calculation. Specifically, this embodiment can input the cloud-based set of abnormal features to be diagnosed and the baseline data of the device cluster operation into the full-scale detection model. The full-scale detection model first unifies the feature dimensions, data format, and units of the two types of data to complete feature alignment. Then, it performs dimension-by-dimensional feature difference quantification calculation on the aligned data, extracts the deviation data of abnormal features relative to the baseline features, and forms difference comparison data to provide a quantitative basis for fault root cause matching.

[0050] This embodiment utilizes a full-scale detection model for fault feature matching and inference. Specifically, the full-scale detection model incorporates a full-scenario fault feature mapping library. It performs one-by-one matching operations between the difference comparison data and various fault features in the fault feature mapping library, filters out potential fault root causes that highly match the difference data, integrates all matching results into a candidate fault root cause set, and generates a corresponding quantitative value for each candidate root cause based on the degree of matching, i.e., the confidence score of each candidate fault root cause.

[0051] This embodiment can determine the root cause of a fault. Specifically, the cloud retrieves the corresponding scenario-based root cause confidence threshold based on the device scenario identifier, compares the confidence score of each candidate root cause with the threshold, and filters out candidate root causes with confidence scores higher than the threshold. If only one root cause meets the condition, it is directly used as the root cause location result data; if multiple root causes exist, the one with the highest confidence score is selected as the final root cause location result data, thus completing the accurate root cause location in the cloud.

[0052] In this embodiment, the full-scale detection model is trained based on historical anomaly feature sets and full cluster baseline data, retaining complete fault detection capabilities. Combined with device scenario identifiers, it achieves accurate matching between the model and baseline data, adapting to the operating characteristics of equipment in different scenarios. This embodiment completes feature alignment and difference calculation through the full-scale detection model, realizing a quantitative comparison between abnormal features and normal baselines, improving the accuracy of fault feature matching. This embodiment uses confidence scoring combined with scenario-based root cause thresholds to screen the final fault root cause, effectively eliminating potential root causes with low matching degrees, achieving accurate fault root cause location. This provides a reliable and accurate technical basis for subsequent fault handling level division and maintenance strategy formulation, improving the professionalism and accuracy of cloud-based fault diagnosis, and avoiding unreasonable operation and maintenance due to misjudgment.

[0053] S105: Determine the fault handling level based on the root cause location results, and maintain the target vending machine based on the fault handling level; the fault handling level is the local self-healing fault level, the cloud-based remote repair fault level, or the on-site handling fault level.

[0054] In this embodiment, determining the fault handling level based on the root cause localization result includes: Based on the root cause location results, determine the fault type data, fault component data, and fault repairability attribute data; Based on preset fault level determination rules, the fault type data, fault component data, and fault repairability attribute data are determined to obtain the fault handling level corresponding to the target vending machine.

[0055] In this embodiment, maintenance of the target vending machine based on fault handling levels includes: If the fault handling level is a local self-healing fault level, then the local self-healing repair instruction is determined based on the fault root cause location result data, and the local self-healing repair instruction is sent to the target vending machine terminal; the self-healing handling result data and handling log data fed back by the target vending machine terminal are received. If the fault handling level is a cloud-based remote repair level, then based on the fault root cause location result data, a standardized remote repair control command is determined and sent to the target vending machine terminal; the remote repair handling result data and handling log data fed back from the target vending machine terminal are received. If the fault handling level is the on-site fault handling level, then obtain the target vending machine deployment location data and regional operation and maintenance resource data, determine the operation and maintenance work order based on the fault root cause location result data, target vending machine deployment location data and regional operation and maintenance resource data, push the operation and maintenance personnel to the operation and maintenance work order, and receive manual operation and maintenance handling data and handling log data.

[0056] In this embodiment, the fault handling level refers to the equipment maintenance handling level defined based on the root cause of the fault, used to match the corresponding maintenance strategy; the local self-healing fault level refers to the fault level that the terminal can repair autonomously, such as minor false alarms from sensors; the cloud-repairable fault level refers to the fault level that requires remote repair via cloud commands, such as abnormal drive parameters in the freight channel; the on-site handling fault level refers to the fault level that requires on-site operation by maintenance personnel, such as hardware damage faults. Fault type data refers to information representing the category to which the fault belongs, such as electrical faults, mechanical faults, etc.; fault component data refers to information representing the specific component where the fault occurred, such as the freight channel motor, barcode scanning module, etc.; fault repairability attribute data refers to attribute information representing whether the fault can be repaired remotely or locally.

[0057] Preset fault level judgment rules refer to the pre-defined level division criteria based on fault characteristics, calibrated by operation and maintenance experience and equipment characteristics; local self-healing repair instructions refer to the operation instructions for terminal self-repair; self-healing handling result data refers to the repair result information after the terminal performs self-healing; handling log data refers to the information recording the entire fault handling process. Standardized remote repair control instructions refer to unified format remote repair instructions issued from the cloud; remote repair handling result data refers to the repair result information after the terminal executes remote instructions; deployment location data refers to the actual installation location information of the vending machine; regional operation and maintenance resource data refers to the resource information such as operation and maintenance personnel and tools in each region; operation and maintenance work orders refer to standardized documents recording faults and maintenance requirements; operation and maintenance personnel targeted push processing refers to the operation of pushing operation and maintenance work orders to the corresponding regional operation and maintenance personnel; manual operation and maintenance handling data refers to the handling result information after on-site operation and maintenance.

[0058] For example, this embodiment can first determine the fault handling level. Specifically, after obtaining the root cause location result, this embodiment extracts core feature information from the result to determine the fault type data, fault component data, and fault repairability attribute data, respectively. This process is completed through feature mapping. This embodiment pre-stores an association mapping library between the root cause of the fault and the three types of data, which can be directly matched to obtain the data. This embodiment can call a preset fault level judgment rule. This rule is formulated and stored in the cloud based on the device hardware characteristics, operation and maintenance technical specifications, and historical fault handling experience, and includes the level division standards corresponding to various fault types, components, and repairability attributes. This embodiment can substitute the three types of data into the rule for matching and judgment one by one, and finally obtain the fault handling level corresponding to the target vending machine.

[0059] This embodiment can perform maintenance operations at the local self-healing fault level. Specifically, if the fault is determined to be at this level, this embodiment can match the corresponding local self-healing repair instruction from a preset self-healing instruction library based on the fault root cause location result. This instruction is a standardized operation instruction adapted for local execution on the terminal and is sent to the target vending machine terminal via the network. After receiving the instruction, the terminal automatically executes the repair operation in the instruction. After the operation is completed, the self-healing handling result data and the handling log data recording the operation process are uploaded to this embodiment in real time. This embodiment can verify the results and retain the log.

[0060] This embodiment can perform maintenance operations at the cloud-based, remotely repairable fault level. Specifically, if the fault is determined to be at this level, this embodiment can generate standardized remote repair control commands based on the fault root cause location results. These commands are formatted according to the device communication protocol and sent to the terminal via an encrypted communication link. After receiving and parsing the commands, the terminal executes the corresponding remote repair operation. Upon completion of the repair, the remote repair result data and handling log data are uploaded to this embodiment. This embodiment can verify the repair effect; if the repair is not successful, the commands are regenerated and reissued.

[0061] This embodiment can perform maintenance operations at the on-site fault handling level. Specifically, if the fault is determined to be at this level, this embodiment can retrieve the deployment location data of the target vending machine from the equipment information database, and simultaneously obtain the regional operation and maintenance resource data of the area to which the location belongs from the operation and maintenance resource management database, including information such as operation and maintenance personnel schedules, affiliated sites, and operation and maintenance tools. This embodiment can generate an operation and maintenance work order based on the fault root cause location results. The operation and maintenance work order includes fault information, equipment location, and maintenance requirements. This embodiment can also target and push the work order to operation and maintenance personnel in the corresponding area based on the regional operation and maintenance resource data and the operation and maintenance work order. After the operation and maintenance personnel complete the on-site maintenance, they enter the manual operation and maintenance handling data and handling log data into the system, and the system synchronously uploads them to the cloud, completing the data retention and fault handling closed loop in the cloud.

[0062] This embodiment extracts multi-dimensional data based on the root cause of the fault and divides the fault handling levels according to preset rules, achieving standardization and accuracy in level division and avoiding inappropriate handling strategies. This embodiment matches differentiated maintenance methods to different levels; local self-healing and cloud-based remote repair eliminate the need for on-site human intervention, shortening fault handling time and reducing manual maintenance costs. On-site handling levels are precisely pushed to work orders based on equipment location and regional maintenance resources, improving the efficiency of maintenance personnel scheduling and the targeted nature of on-site handling. The entire process retains handling logs and result data, enabling traceability of fault handling and ensuring the efficiency, accuracy, and standardization of vending machine fault maintenance, thereby improving the overall operational stability of the equipment.

[0063] Based on the same principle as the remote fault diagnosis and maintenance method provided in the embodiments of this application, the embodiments of this application also provide a remote fault diagnosis and maintenance device, such as... Figure 2 As shown, the remote fault diagnosis and maintenance device 20 may specifically include: a device data acquisition module 21, a lightweight detection module 22, a cloud module 23, a full detection module 24, and a device maintenance module 25. The device data acquisition module 21 is used to acquire the device operation data and deployment scenario data reported by the target vending machine terminal, standardize and extract features from the device operation data to obtain device operation feature data. The lightweight detection module 22 is used to determine the lightweight detection model deployed on the target vending machine terminal corresponding to the deployment scenario data; based on the device operation characteristic data, the lightweight detection model is used to obtain the anomaly judgment result; The cloud module 23 is used to selectively extract and encapsulate the device operation feature data corresponding to the anomaly judgment result to obtain an anomaly feature data packet; based on the terminal network connectivity status data, the anomaly feature data packet is uploaded to the cloud; the cloud is used to parse and verify the anomaly feature data to obtain the cloud-based anomaly feature set to be diagnosed and the device scene identifier; The full-scale detection module 24 is used to determine the full-scale detection model and equipment cluster operation baseline data corresponding to the equipment scene identifier; based on the cloud-based abnormal feature set to be diagnosed and the equipment cluster operation baseline data, the full-scale detection model is used to determine the fault root cause location result. The equipment maintenance module 25 is used to determine the fault handling level based on the fault root cause location result, and to maintain the target vending machine based on the fault handling level; the fault handling level is the local self-healing fault level, the cloud-based remote repair fault level, or the on-site handling fault level.

[0064] In one embodiment of this application, the lightweight detection model is obtained by training an isolated forest model on a historical equipment operation feature dataset corresponding to multiple high-frequency fault types; high-frequency faults are fault types whose fault frequency is higher than a preset fault frequency threshold; the lightweight detection module 22 is specifically used to: perform anomaly inference calculation on the equipment operation feature data through the lightweight detection model to obtain real-time anomaly score data; compare and judge the real-time anomaly score data with the scenario-based anomaly threshold data to obtain anomaly judgment result data; the scenario-based anomaly threshold data is determined based on the deployment scenario data.

[0065] In one embodiment of this application, the full-scale detection model is obtained by training the isolated forest model based on the historical cloud-based set of abnormal features to be diagnosed and the baseline data of all device cluster operations; the full-scale detection module 24 is specifically used for: The full-scale detection model is used to align the features of the cloud-based anomaly feature set to be diagnosed and the baseline data of the device cluster operation. The difference between the feature-aligned cloud-based anomaly feature set to be diagnosed and the baseline data of the device cluster operation is calculated to obtain the difference comparison data. By using a full-scale detection model to perform fault feature matching and inference on the difference comparison data, a set of candidate fault root causes data is obtained. The confidence score of each candidate root cause is determined based on the candidate root cause set data. Based on the confidence scores of each candidate root cause and the scenario-based root cause confidence threshold, the root cause localization results are determined; the scenario-based root cause confidence threshold is determined based on the deployment scenario data.

[0066] In one embodiment of this application, the equipment maintenance module 25 is specifically used to: determine fault type data, fault component data, and fault repairability attribute data based on the fault root cause location results; and determine the fault type data, fault component data, and fault repairability attribute data based on preset fault level determination rules to obtain the fault handling level corresponding to the target vending machine.

[0067] In one embodiment of this application, the equipment maintenance module 25 is further configured to: If the fault handling level is a local self-healing fault level, then the local self-healing repair instruction is determined based on the fault root cause location result data, and the local self-healing repair instruction is sent to the target vending machine terminal; the self-healing handling result data and handling log data fed back by the target vending machine terminal are received. If the fault handling level is a cloud-based remote repair level, then based on the fault root cause location result data, a standardized remote repair control command is determined and sent to the target vending machine terminal; the remote repair handling result data and handling log data fed back from the target vending machine terminal are received. If the fault handling level is the on-site fault handling level, then obtain the target vending machine deployment location data and regional operation and maintenance resource data, determine the operation and maintenance work order based on the fault root cause location result data, target vending machine deployment location data and regional operation and maintenance resource data, push the operation and maintenance personnel to the operation and maintenance work order, and receive manual operation and maintenance handling data and handling log data.

[0068] In one embodiment of this application, the cloud module 23 is specifically used for: determining the terminal network connectivity status based on terminal network connectivity status data; if the terminal network connectivity status determination result is network connectivity, then uploading the abnormal feature data packet to the cloud in real time; if the terminal network connectivity status determination result is network disconnection, then persistently caching the abnormal feature data packet locally to obtain a locally cached abnormal feature data packet; acquiring network detection data, and determining the upload trigger data after network recovery based on the network detection data and the locally cached abnormal feature data packet; and re-uploading the locally cached abnormal feature data packet to the cloud after network recovery based on the upload trigger data.

[0069] In one embodiment of this application, the cloud module 23 is further configured to: determine the time node of the anomaly occurrence and the preset feature extraction period based on the anomaly determination result; perform targeted extraction of the device operation feature data according to the time node of the anomaly occurrence and the preset feature extraction period to obtain a subset of abnormal time period feature data; and standardize and encapsulate the subset of abnormal time period feature data to obtain an anomaly feature data package.

[0070] The apparatus in this application embodiment can execute the method provided in this application embodiment, and the implementation principle is similar. The actions performed by each module in the apparatus of each embodiment of this application correspond to the steps in the method of each embodiment of this application. For detailed functional descriptions of each module of the apparatus, please refer to the descriptions in the corresponding methods shown above, which will not be repeated here.

[0071] Figure 3 A schematic diagram of the structure of an electronic device to which this application embodiment applies is shown, such as... Figure 3 As shown, the electronic device can be used to implement the methods provided in any embodiment of this application.

[0072] like Figure 3 As shown, the electronic device 300 may primarily include at least one processor 301. Figure 3 The diagram shows components such as a memory 302, a communication module 303, and an input / output interface 304. Optionally, these components can be connected and communicate with each other via a bus 305. It should be noted that... Figure 3 The structure of the electronic device 300 shown is merely illustrative and does not constitute a limitation on the electronic devices to which the methods provided in the embodiments of this application are applicable.

[0073] The memory 302 can be used to store operating systems and applications, etc. The applications can include computer programs that implement the methods shown in the embodiments of this application when invoked by the processor 301, and can also include programs for implementing other functions or services. The memory 302 can be ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices that can store information and computer programs, or it can be EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited thereto.

[0074] Processor 301 is connected to memory 302 via bus 305 and implements corresponding functions by calling the application programs stored in memory 302. Processor 301 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 301 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0075] Electronic device 300 can connect to a network via communication module 303 (which may include, but is not limited to, components such as a network interface) to communicate with other devices (such as user terminals or servers) through the network and achieve data interaction, such as sending data to or receiving data from other devices. Communication module 303 may include wired network interfaces and / or wireless network interfaces, meaning the communication module may include at least one of wired or wireless communication modules.

[0076] The electronic device 300 can connect to necessary input / output devices, such as a keyboard and display device, via the input / output interface 304. The electronic device 300 itself may have a display device, and other display devices can also be connected externally via the interface 304. Optionally, a storage device, such as a hard drive, can also be connected via the interface 304 to store data from the electronic device 300, retrieve data from the storage device, or store data from the storage device in the memory 302. It is understood that the input / output interface 304 can be a wired interface or a wireless interface. Depending on the actual application scenario, the device connected to the input / output interface 304 can be a component of the electronic device 300 or an external device connected to the electronic device 300 when needed.

[0077] The bus 305 used to connect the components may include a path for transmitting information between the components. The bus 305 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Depending on its function, the bus 305 may be divided into an address bus, a data bus, a control bus, etc.

[0078] Optionally, for the solution provided in the embodiments of this application, the memory 302 can be used to store a computer program that executes the solution of this application, and the processor 301 runs the computer program. When the processor 301 runs the computer program, it implements the operation of the method or apparatus provided in the embodiments of this application.

[0079] Based on the same principle as the method provided in the embodiments of this application, the embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the corresponding content of the aforementioned method embodiments.

[0080] It should be noted that the terms "first," "second," "third," "fourth," "1," "2," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in a sequence other than that shown in the figures or text.

[0081] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0082] It should be understood that although arrows indicate various operation steps in the flowcharts of this application's embodiments, the order in which these steps are implemented is not limited to the order indicated by the arrows. Unless explicitly stated herein, in some implementation scenarios of this application's embodiments, the implementation steps in each flowchart can be executed in other orders as required. Furthermore, some or all steps in each flowchart, based on the actual implementation scenario, may include multiple sub-steps or multiple stages. Some or all of these sub-steps or stages can be executed at the same time, and each sub-step or stage can also be executed at different times. In scenarios where execution times differ, the execution order of these sub-steps or stages can be flexibly configured according to requirements, and this application's embodiments do not limit this.

[0083] The above description is only an optional implementation method for some implementation scenarios of this application. It should be noted that for those skilled in the art, other similar implementation methods based on the technical concept of this application without departing from the technical concept of this application also fall within the protection scope of the embodiments of this application.

Claims

1. A method for remote fault diagnosis and maintenance, characterized in that, include: Obtain device operation data and deployment scenario data reported by the target vending machine terminal, standardize and extract features from the device operation data to obtain device operation feature data; A lightweight detection model is deployed for the target vending machine terminal corresponding to the deployment scenario data; based on the device operation characteristic data, an anomaly determination result is obtained through the lightweight detection model; The device operation characteristic data corresponding to the anomaly determination result is selectively intercepted and encapsulated to obtain an anomaly characteristic data package; The abnormal feature data packet is uploaded to the cloud based on the terminal network connectivity status data; the cloud is used to parse and verify the abnormal feature data to obtain the cloud-based abnormal feature set to be diagnosed and the device scene identifier; Determine the full-scale detection model and equipment cluster operation baseline data corresponding to the equipment scene identifier; Based on the cloud-based set of abnormal features to be diagnosed and the baseline data of the device cluster operation, the root cause location of the fault is determined by the full-scale detection model. Based on the fault root cause location results, the fault handling level is determined, and the target vending machine is maintained based on the fault handling level. The fault handling levels are: local self-healing fault level, cloud-based remote repair fault level, or on-site fault handling level.

2. The remote fault diagnosis and maintenance method as described in claim 1, characterized in that, The lightweight detection model is obtained by training an isolated forest model based on a historical equipment operation feature dataset corresponding to multiple high-frequency fault types; the high-frequency faults are fault types with a fault frequency higher than a preset fault frequency threshold. The process of obtaining anomaly determination results based on the device's operational characteristic data and the lightweight detection model includes: The lightweight detection model is used to perform anomaly inference calculations on the device operation feature data to obtain real-time anomaly score data. Anomaly determination results are obtained by comparing the real-time anomaly score data with the scenario-based anomaly threshold data; the scenario-based anomaly threshold data is determined based on the deployment scenario data.

3. The remote fault diagnosis and maintenance method as described in claim 1, characterized in that, The full-scale detection model is obtained by training the isolated forest model based on the historical cloud-based set of abnormal features to be diagnosed and the baseline data of all device cluster operations; The determination of the root cause location result of the fault based on the cloud-based abnormal feature set to be diagnosed and the baseline operating data of the device cluster, through the full-scale detection model, includes: The full-scale detection model is used to align the cloud-based abnormal feature set to be diagnosed with the baseline data of the device cluster operation, and the difference between the feature-aligned cloud-based abnormal feature set to be diagnosed with the baseline data of the device cluster operation is calculated to obtain difference comparison data. The full-scale detection model is used to perform fault feature matching and inference on the difference comparison data to obtain a set of candidate fault root causes. The confidence score of each candidate root cause is determined based on the candidate root cause set data. Based on the confidence scores of each candidate root cause and the scenario-based root cause confidence threshold, the root cause localization result data is determined; the scenario-based root cause confidence threshold is determined based on the deployment scenario data.

4. The remote fault diagnosis and maintenance method as described in claim 1, characterized in that, The process of determining the fault handling level based on the fault root cause localization results includes: Based on the fault root cause localization results, fault type data, fault component data, and fault repairability attribute data are determined. Based on preset fault level determination rules, the fault type data, the fault component data, and the fault repairability attribute data are determined to obtain the fault handling level corresponding to the target vending machine.

5. The remote fault diagnosis and maintenance method as described in claim 1, characterized in that, The maintenance of the target vending machine based on the fault handling level includes: If the fault handling level is a local self-healing fault level, then a local self-healing repair instruction is determined based on the fault root cause location result data, and the local self-healing repair instruction is sent to the target vending machine terminal; the self-healing handling result data and handling log data fed back by the target vending machine terminal are received. If the fault handling level is a cloud-based remotely repairable fault level, then a standardized remote repair control command is determined based on the fault root cause location result data, and the standardized remote repair control command is sent to the target vending machine terminal; the remote repair handling result data and handling log data fed back by the target vending machine terminal are received; If the fault handling level is the on-site fault handling level, then the target vending machine deployment location data and regional operation and maintenance resource data are obtained; an operation and maintenance work order is determined based on the fault root cause location result data, the target vending machine deployment location data, and the regional operation and maintenance resource data; the operation and maintenance work order is used to push the work order to operation and maintenance personnel; and manual operation and maintenance handling data and handling log data are received.

6. The remote fault diagnosis and maintenance method as described in claim 1, characterized in that, The process of uploading the abnormal feature data packet to the cloud based on the terminal network connectivity status data includes: The terminal network connectivity status determination result is determined based on the terminal network connectivity status data. If the terminal network connectivity status determination result is that the network is connected, then the abnormal feature data packet is uploaded to the cloud in real time; If the terminal network connectivity status determination result is that the network is not connected, the abnormal feature data packet is locally persistently cached to obtain a locally cached abnormal feature data packet; network detection data is obtained, and upload trigger data after network recovery is determined based on the network detection data and the locally cached abnormal feature data packet; the locally cached abnormal feature data packet is re-uploaded to the cloud after network recovery based on the upload trigger data.

7. The remote fault diagnosis and maintenance method as described in claim 1, characterized in that, The process of selectively extracting and encapsulating the device operation characteristic data corresponding to the anomaly determination result to obtain an anomaly characteristic data packet includes: Based on the anomaly determination results, the time node of the anomaly occurrence and the preset feature extraction period are determined; The device operation feature data is selectively extracted according to the time node of the anomaly occurrence and the preset feature extraction period to obtain the feature subset data of the anomaly period. The abnormal time period feature subset data is standardized and encapsulated to obtain the abnormal feature data package.

8. A remote fault diagnosis and maintenance device, characterized in that, include: The device data acquisition module is used to acquire device operation data and deployment scenario data reported by the target vending machine terminal, and to standardize and extract features from the device operation data to obtain device operation feature data. A lightweight detection module is used to determine the lightweight detection model deployed on the target vending machine terminal corresponding to the deployment scenario data; based on the device operation characteristic data, an anomaly judgment result is obtained through the lightweight detection model; The cloud module is used to selectively extract and encapsulate the device operation characteristic data corresponding to the anomaly determination result to obtain an anomaly characteristic data package; The abnormal feature data packet is uploaded to the cloud based on the terminal network connectivity status data; the cloud is used to parse and verify the abnormal feature data to obtain the cloud-based abnormal feature set to be diagnosed and the device scene identifier; The full-scale detection module is used to determine the full-scale detection model and the baseline data of the device cluster operation corresponding to the device scene identifier. Based on the cloud-based set of abnormal features to be diagnosed and the baseline data of the device cluster operation, the root cause location of the fault is determined by the full-scale detection model. The equipment maintenance module is used to determine the fault handling level based on the fault root cause location result, and to maintain the target vending machine based on the fault handling level; The fault handling levels are: local self-healing fault level, cloud-based remote repair fault level, or on-site fault handling level.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the remote fault diagnosis and maintenance method according to any one of claims 1 to 7 when running the computer program.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the remote fault diagnosis and maintenance method according to any one of claims 1 to 7.

Images