Image terminal fault prediction method and device, equipment and storage medium

Abstract

The invention discloses an image terminal fault prediction method and device, equipment and a storage medium. The method comprises the steps that equipment basic data, operation state data, detection index data and working environment data of a plurality of image terminals are determined, the equipment basic data comprise equipment models and equipment working time, and the operation state data comprise an equipment CPU occupancy rate, an equipment memory utilization rate, equipment temperature and equipment signal strength; the detection index data comprises an equipment image frame rate, equipment identification accuracy and equipment alarm frequency, and the working environment data comprises environment temperature and vibration intensity; and determining image terminal features based on the equipment basic data, the operation state data, the detection index data and the working environment data, and predicting potential faults of each image terminal in a preset time window based on the image terminal features. According to the scheme, the accuracy and the real-time performance of equipment health assessment are improved by constructing an image terminal state modeling mechanism fusing the multi-source features.

Description

Technical Field

[0001] The present application relates to the field of operation and maintenance management technology, and in particular to an image terminal fault prediction method, device, equipment and storage medium. Background Art

[0002] With the rapid development of applications such as image recognition, intelligent security, and industrial visual inspection, a large number of image terminal devices have been deployed in various edge scenarios. Over the long term, the performance of image terminals is affected by a variety of factors, including hardware aging, load fluctuations, and environmental interference. These factors can easily lead to decreased recognition accuracy, increased processing latency, and even image acquisition anomalies. These failures not only impact business stability but also increase maintenance costs.

[0003] Existing image terminal operation and maintenance methods typically rely on regular manual inspections or alarm mechanisms based on static threshold rules. These methods suffer from delayed response, inaccurate fault location, and high false alarm and missed alarm rates, making them difficult to adapt to complex and changing edge deployment environments. Furthermore, the operation of image terminals involves multi-source heterogeneous data, including device-level status, network-level communication quality, and environmental-level disturbances. Traditional single-indicator monitoring makes it difficult to effectively capture potential failure risks. Therefore, there is an urgent need for an image terminal fault prediction method that integrates multi-dimensional device operation data and possesses predictive and fault trend analysis capabilities to enhance the intelligent operation and maintenance of image terminals and achieve early warning and rapid response to faults. Summary of the Invention

[0004] The present application provides an image terminal fault prediction method, apparatus, device, and storage medium. This method can construct a complete image terminal status profile by acquiring basic device data, operating status data, detection index data, and working environment data of several image terminals and extracting various types of feature information. The basic device data includes the device model and operating time, which are used to characterize the hardware attributes and usage cycle of the device itself; the operating status data includes the device CPU occupancy rate, memory usage rate, device temperature, and signal strength, which are used to reflect the current load and operating stability of the device; the detection index data includes image frame rate, recognition accuracy, and alarm frequency, which are used to measure image processing performance and algorithm reliability; and the working environment data includes ambient temperature and vibration intensity, which reflect the potential impact of the external operating environment on the device status. Based on multi-dimensional device data, a feature extraction model is used to obtain basic device features, operating status features, detection index features, and working environment features, and then a full feature set of the image terminal is constructed based on this fusion. Subsequently, an intelligent prediction model is introduced to determine the possibility of each image terminal failing within a preset time window based on the image terminal features, thereby achieving early perception of potential device anomalies. This approach, driven by multi-source feature modeling, state fusion analysis, and prediction algorithms, overcomes the lag and single-point false alarm issues inherent in traditional threshold-triggered alarm mechanisms, significantly improving the granularity of perception and proactive prediction of the operational health of image terminals. Suitable for scenarios requiring continuous and stable device operation, such as video surveillance, industrial vision, and edge computing node management, it boasts high accuracy, real-time performance, and system compatibility, providing an efficient and scalable technical solution for intelligent operations, smart diagnosis, and edge device management.

[0005] In a first aspect, the present application provides an image terminal fault prediction method, comprising:

[0006] Determine basic device data, operating status data, detection index data, and working environment data of several image terminals. The basic device data includes the device model and device operating hours. The operating status data includes the device CPU occupancy rate, device memory usage rate, device temperature, and device signal strength. The detection index data includes the device image frame rate, device recognition accuracy, and device alarm frequency. The working environment data includes the ambient temperature and vibration intensity.

[0007] Extracting basic device features, operating status features, detection index features, and working environment features based on the device basic data, the operating status data, the detection index data, and the working environment data;

[0008] Image terminal characteristics are determined based on the basic characteristics of the equipment, the operating status characteristics, the detection index characteristics, and the working environment characteristics, and potential failures of each image terminal within a preset time window are predicted based on the image terminal characteristics.

[0009] In a second aspect, the present application provides an image terminal fault prediction device, comprising:

[0010] A data acquisition module is used to determine basic device data, operating status data, detection index data, and working environment data of several image terminals. The basic device data includes the device model and device operating hours. The operating status data includes the device CPU occupancy rate, device memory usage rate, device temperature, and device signal strength. The detection index data includes the device image frame rate, device recognition accuracy, and device alarm frequency. The working environment data includes the ambient temperature and vibration intensity.

[0011] A feature extraction module is used to extract basic device features, operating status features, detection index features and working environment features based on the basic device data, the operating status data, the detection index data and the working environment data;

[0012] A fault prediction module is used to determine image terminal characteristics based on the basic characteristics of the equipment, the operating status characteristics, the detection index characteristics and the working environment characteristics, and predict potential faults of each image terminal within a preset time window based on the image terminal characteristics.

[0013] In a third aspect, the present application provides an image terminal fault prediction device, comprising:

[0014] one or more processors;

[0015] The memory stores one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the image terminal fault prediction method as described in the first aspect.

[0016] In a fourth aspect, the present application provides a storage medium comprising computer-executable instructions, which, when executed by a computer processor, are used to execute the image terminal fault prediction method as described in the first aspect.

[0017] In this application, a forward-looking and practical integrated strategy for image terminal fault prediction is constructed by integrating multi-source data feature modeling with intelligent prediction algorithms. First, multi-dimensional operational information of the image terminal is obtained, including basic device data, operational status data, detection index data, and operating environment data. Basic device data characterizes the terminal's hardware configuration and service life, operational status data reflects load and stability, detection index data reflects image processing performance and the quality of its response to external tasks, and operating environment data reveals potential interference factors in the terminal's environment. Using this data, basic device features, operational status features, detection index features, and operating environment features are extracted and integrated to construct a comprehensive operational profile of the image terminal. Based on this, a trained prediction model is introduced to determine the potential failure probability of each terminal within a preset future time window based on the comprehensive features, thereby achieving proactive perception of the image terminal's operational status and risk prevention. This method, through the synergistic mechanism of data perception, feature analysis, and predictive modeling, accurately identifies operational anomalies that may be caused by device aging, abnormal load, reduced recognition accuracy, or environmental interference, significantly improving the operational visibility and predictive capabilities of the image terminal system. Compared with traditional mechanisms that rely on static thresholds or single-indicator abnormality triggers, this solution has stronger adaptability and proactiveness. It is particularly suitable for the operating environment of image terminals under edge deployment, large-scale networking and high continuity requirements. It provides stable, efficient and scalable technical support for building an intelligent and refined visual terminal health management system. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flowchart of the image terminal fault prediction method provided by an embodiment of the present application;

[0019] Figure 2 This is a flow chart of image terminal data acquisition provided by an embodiment of the present application;

[0020] Figure 3 This is a flow chart of load balancing between the master station and edge nodes provided by an embodiment of the present application;

[0021] Figure 4 This is a flow chart of cloud computing power allocation calculation provided by an embodiment of the present application;

[0022] Figure 5 This is a flow chart of image terminal feature determination provided by an embodiment of the present application;

[0023] Figure 6 This is a flowchart of the visualization of the image terminal status provided by an embodiment of the present application;

[0024] Figure 7 This is a structural diagram of an image terminal fault prediction device provided in an embodiment of the present application;

[0025] Figure 8 This is a structural diagram of the image terminal fault prediction device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solutions and advantages of the present application clearer, the specific embodiments of the present application are further described in detail below in conjunction with the accompanying drawings. It is understood that the specific embodiments described herein are merely used to explain the present application and are not intended to limit the present application. It should also be noted that, for ease of description, only portions related to the present application, not all of the contents, are shown in the accompanying drawings. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the various operations (or steps) as being processed sequentially, many of the operations therein can be performed in parallel, concurrently or simultaneously. In addition, the order of the various operations can be rearranged. The process can be terminated when its operations are completed, but may also have additional steps not included in the accompanying drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0027] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or precedence. It should be understood that the data used in this manner are interchangeable where appropriate, so that the embodiments of this application can be implemented in an order other than those illustrated or described herein, and that the objects distinguished by "first," "second," and the like are generally of the same type, and do not limit the number of objects. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects connected before and after are in an "or" relationship.

[0028] As the power system continues to become more intelligent and digitalized, an increasing number of intelligent image terminals, such as video surveillance cameras, infrared imaging devices, and intelligent inspection robots, are being widely deployed in key locations like substations, transmission lines, and distribution rooms. They are used for a variety of tasks, including environmental perception, operational monitoring, hazard identification, and fault warning. These image terminals have played a vital role in improving grid operational safety and enhancing remote operation and maintenance capabilities, becoming core sensing nodes in the power Internet of Things (IoT) system.

[0029] However, in actual operation and maintenance, the current intelligent image terminal operation supervision system still has several technical shortcomings, hindering its large-scale application and further development of intensive management capabilities. On the one hand, the lack of unified communication protocols and interface specifications between devices from different manufacturers and models leads to complex device access processes and inconsistent data formats, making it difficult to achieve unified access and centralized management of multiple terminals. On the other hand, most image terminals lack the ability to upload real-time status, making it impossible to remotely monitor the health of the equipment. This leads to delayed operation and maintenance responses and the risk of missed anomalies and false alarms.

[0030] Furthermore, existing methods for analyzing terminal-collected images and operational data remain relatively primitive, lacking comprehensive feature extraction, trend analysis, and predictive modeling. This results in a significant amount of data being collected but not used, or used inaccurately, impacting decision-making for back-end intelligent scheduling and fault diagnosis. Furthermore, the lack of standardized service records and response mechanisms throughout the equipment lifecycle leads to unclear service responsibilities and inconsistent service quality, further reducing overall system operational efficiency and service traceability.

[0031] Therefore, there is an urgent need to build an intelligent image terminal operation supervision system with unified access specifications, real-time status perception, intelligent data analysis and service process traceability capabilities, to achieve digital management of the entire life cycle of image terminals, improve the operating stability, maintenance response efficiency and system collaboration capabilities of image equipment in the power industry, and provide key support for the construction of smart grids and digital operation and maintenance systems.

[0032] To address the aforementioned issues, this embodiment provides an image terminal fault prediction method. By building a multi-source data perception, feature fusion modeling, and intelligent fault prediction linkage mechanism, this method forms an integrated intelligent operation and maintenance process with high-dimensional data support and state prediction capabilities. This method first determines the operational data inputs for several image terminals, including basic device data, operational status data, detection indicator data, and operating environment data. Basic device data includes the device model and operating hours, characterizing the device hardware specifications and service life. Operational status data includes CPU utilization, memory usage, device temperature, and signal strength, reflecting the terminal's current load and thermal stability. Detection indicator data includes image frame rate, recognition accuracy, and alarm frequency, quantifying the terminal's task processing capability and fault sensitivity. Operating environment data includes ambient temperature and vibration intensity, assisting in identifying the impact of external interference on device stability. Based on this data, basic device features, operational status features, detection indicator features, and operating environment features are extracted, and a unified image terminal feature vector is constructed. Driven by these fused features, a fault prediction model is deployed to model and analyze the operational risk of each terminal within a preset future time window, outputting a fault probability or risk level, thereby providing early warning of potential image terminal anomalies.

[0033] This method constructs a closed-loop "perception-modeling-prediction" pathway, integrating multi-dimensional, heterogeneous data into a unified feature space and introducing risk assessment logic within a time series window, significantly improving the accuracy and predictiveness of fault identification. Compared to traditional alarm methods that rely on static thresholds or single-metric monitoring, this method offers advantages such as strong adaptability, fine-grained prediction, and robust data interpretability. It can be further expanded with modules such as cloud platform integration, image terminal health scoring, and fault location visualization to achieve a comprehensive, automated, and traceable intelligent maintenance system for image terminals. This solution boasts excellent deployment adaptability and platform compatibility, making it particularly suitable for scenarios requiring distributed management of large numbers of image terminal devices, such as power systems, industrial monitoring, and smart campuses. It offers significant engineering value in improving terminal operational stability, reducing manual operation and maintenance costs, and shortening abnormal response times. By integrating multi-dimensional capabilities such as device status perception, indicator modeling, and intelligent prediction, this method provides a replicable and feasible technical path for high-reliability image terminal operation and the construction of intelligent perception infrastructure.

[0034] The image terminal fault prediction method provided in this embodiment can be executed by an image terminal fault prediction device. The image terminal fault prediction device can be implemented through software and / or hardware. The image terminal fault prediction device can be composed of two or more physical entities, or a single physical entity. For example, the image terminal fault prediction device can be an operation and maintenance server used to maintain normal business operations.

[0035] The image terminal fault prediction device is installed with at least one operating system, including but not limited to Android, Linux, and Windows. The image terminal fault prediction device can install at least one application based on the operating system. The application can be native to the operating system or downloaded from a third-party device or server. In this embodiment, the image terminal fault prediction device includes at least one application capable of executing the image terminal fault prediction method.

[0036] For ease of understanding, this embodiment is described by taking an operation and maintenance server as an example of the main body for executing the image terminal fault prediction method.

[0037] Figure 1 A flowchart of a method for predicting image terminal failures provided by an embodiment of the present application is given. Figure 1 , the image terminal fault prediction method specifically includes:

[0038] S110. Determine basic device data, operating status data, detection index data, and working environment data of several image terminals, where the basic device data includes the device model and device operating hours, the operating status data includes the device CPU occupancy rate, device memory usage rate, device temperature, and device signal strength, the detection index data includes the device image frame rate, device recognition accuracy, and device alarm frequency, and the working environment data includes the ambient temperature and vibration intensity.

[0039] In some embodiments, the basic device data, operating status data, detection index data and working environment data of several image terminals are first determined, where the image terminal refers to an edge intelligent device or perception node used for image acquisition, processing or recognition.

[0040] Basic device data refers to information about the image terminal's static configuration and usage cycle, including the device model and device operating hours. The device model identifies the terminal's hardware specifications, and the device operating hours refer to the cumulative usage time since the terminal was put into operation. Operational status data refers to real-time parameters reflecting the terminal's current operating load and hardware status, including CPU utilization, memory usage, temperature, and signal strength. CPU utilization and memory usage represent computing resource load, temperature reflects hardware cooling status, and signal strength describes network connection quality. Detection indicator data assesses the image terminal's performance in image acquisition and analysis tasks, including image frame rate, recognition accuracy, and alarm frequency. Image frame rate refers to the number of images captured per unit time, recognition accuracy indicates the reliability of image recognition results, and alarm frequency monitors the frequency of anomaly detection triggers. Operating environment data refers to indicators of the physical conditions of the terminal's operating environment, including ambient temperature and vibration intensity. Ambient temperature is used to assess the impact of external temperature on device operation, while vibration intensity reflects the impact of potential mechanical vibration on device stability.

[0041] In one embodiment, the method of collecting the above-mentioned various types of data can be: obtaining and aggregating relevant information in real time through local sensors of the device, built-in operating system interfaces and external environmental monitoring modules, providing basic support for subsequent status assessment and fault warning.

[0042] Optionally, Figure 2 The flowchart of image terminal data acquisition provided by the embodiment of the present application is given. Figure 2 , the image terminal data acquisition method specifically includes:

[0043] S1101. Receive terminal original data sent by several image terminals, where the terminal original data includes basic device original data, operating status original data, detection index original data, and working environment original data.

[0044] Exemplarily, the terminal raw data sent by several image terminals is first received, where the image terminal refers to the edge computing device deployed in the image acquisition, recognition or monitoring scenario, and the terminal raw data refers to the original monitoring data that has not been processed, including basic information on the device's operating status and environmental perception.

[0045] The terminal raw data includes: basic equipment raw data, which refers to basic identity information such as equipment model, factory serial number, and operation start time; operation status raw data, which refers to real-time operation load data such as CPU occupancy, memory usage, equipment temperature, and signal strength in the current cycle of the terminal; detection index raw data, which refers to image processing performance data such as image frame rate, recognition accuracy, and alarm frequency; working environment raw data, which refers to external physical factors that affect equipment operation, such as ambient temperature, humidity, and vibration intensity of the terminal's environment.

[0046] In one embodiment, the method of receiving terminal raw data can be: each image terminal reports the collected raw data to the central processing system periodically or in an event-triggered manner through the edge network, and the central system receives and caches this data through a message queue, data bus or API interface for subsequent feature extraction and fault analysis.

[0047] S1102: Perform semantic analysis on the original terminal data to determine the communication protocol corresponding to each original terminal data.

[0048] Exemplarily, semantic parsing is performed on raw terminal data to determine the communication protocol corresponding to each terminal's raw data. Raw terminal data refers to the unstructured data packets uploaded by the image terminal, and semantic parsing refers to the process of identifying and mapping the field meanings, encoding structure, data type, and other contents of the raw data. Communication protocols refer to the protocol standards or format specifications followed by different image terminals during data transmission, which guide data decoding, verification, and field interpretation.

[0049] In one embodiment, the semantic parsing method can be: parsing the header fields, identifiers, and content structures in the terminal uploaded data through a protocol identification engine, automatically matching the preset protocol template, and identifying the protocol type to which the data belongs, such as MQTT, Modbus, HTTP, custom binary format, etc.

[0050] In one embodiment, the communication protocol can be determined by building a communication protocol feature library, matching the original data with the communication protocol one by one based on key field rule matching, protocol identifier mapping, or data frame feature analysis, thereby laying the foundation for subsequent data decoding and standardization processing.

[0051] S1103: Process the original terminal data based on the communication protocol to obtain basic device data, operating status data, detection index data, and working environment data corresponding to each of the image terminals.

[0052] Exemplarily, the raw terminal data is processed based on the communication protocol to obtain basic device data, operating status data, detection index data, and working environment data corresponding to each image terminal. The communication protocol specifies how to extract field content, parse data structure, and recover semantic information from the raw data, and serves as the foundation for data standardization. Basic device data refers to parsed static attribute information such as the device model, device number, and accumulated operating hours; operating status data includes real-time operating parameters such as CPU occupancy, memory usage, device temperature, and signal strength; detection index data includes indicators reflecting terminal task performance, such as image frame rate, image recognition accuracy, and device alarm frequency; and working environment data includes measurement data such as ambient temperature, humidity, and vibration intensity that describe the external operating environment of the terminal.

[0053] In one embodiment, the terminal original data may be processed by performing field decomposition, unit conversion, format conversion, and semantic mapping on the original message according to the field parsing rules, offset, and encoding format defined in the communication protocol, and finally generating a structured standard data object.

[0054] In one embodiment, the protocol-driven parsing process can support the parallel access of multiple heterogeneous terminals, ensuring that the original data of each type of image terminal can be accurately restored to standard indicator data for subsequent feature extraction and status analysis.

[0055] S120 , extracting basic device features, operating status features, detection index features, and working environment features based on the device basic data, the operating status data, the detection index data, and the working environment data.

[0056] In some embodiments, based on the basic data of the equipment, operating status data, detection index data and working environment data, the basic characteristics of the equipment, operating status characteristics, detection index characteristics and working environment characteristics are extracted respectively, wherein each type of characteristics is used to construct a multi-dimensional status portrait of the image terminal as an input basis for subsequent modeling analysis and fault prediction.

[0057] Basic device characteristics refer to information representing the static device identity and aging degree extracted from the device model and device working time, such as device type code, usage cycle level, etc.; operating status characteristics refer to dynamic load and stability characteristics extracted by analyzing CPU occupancy, memory usage, device temperature and signal strength, such as load score, temperature stability coefficient, network connection quality grade, etc.; detection index characteristics refer to performance index characteristics extracted based on image frame rate, recognition accuracy and alarm frequency, which are used to reflect the performance of the terminal in image processing tasks, such as real-time evaluation value, recognition accuracy scalar, alarm density factor, etc.; working environment characteristics refer to external interference representation extracted from ambient temperature and vibration intensity, which are used to evaluate the impact of the environment in which the device is located on the operating status, such as thermal environment pressure level, vibration disturbance frequency, etc.

[0058] In one embodiment, the feature extraction method may be: performing pre-processing operations such as normalization, binning, sliding window statistics, and trend fitting on various types of raw data, and generating structured feature vectors in combination with expert rules or training models.

[0059] Optionally, Figure 3 A flow chart of load balancing between the master station and edge nodes provided in the embodiment of the present application is given. Figure 3 , the master station and edge node load balancing method specifically includes:

[0060] S1201: Obtain the edge available computing power of the edge node where the image terminal is located and the cloud available computing power of the master station to which it belongs.

[0061] Exemplarily, the first step is to obtain the edge available computing power of the edge node where the image terminal is located and the cloud available computing power of the main station to which it belongs. The edge node refers to the local computing resource node deployed near the image terminal, and the edge available computing power refers to the computing resource capacity currently available for task execution at the node, such as the number of CPU cores, GPU processing power, or memory bandwidth. The cloud available computing power of the main station refers to the computing resources that can be dispatched by the central cloud platform within the current time period to support large-scale model reasoning, task aggregation, or historical data analysis.

[0062] In one embodiment, the method for obtaining the available computing power at the edge can be: through the resource status information regularly reported by the edge computing node, extract the CPU usage, GPU idle time, available memory and container load, and construct a quantitative representation of the current available computing power.

[0063] In one embodiment, the method for obtaining available computing power in the cloud can be: querying the resource allocation status of the current master station through the cloud resource scheduling platform or resource management interface, including indicators such as idle computing nodes, remaining capacity of the AI ​​inference resource pool, and distributed storage bandwidth, as the resource basis for task decision-making.

[0064] In one embodiment, the coordinated acquisition of available computing power at the edge and in the cloud can provide basic data support for subsequent task scheduling strategies, enabling dynamic allocation and optimized execution of image processing tasks at the edge or in the cloud.

[0065] S1202. Calculate the cloud-allocated computing power of the master station based on the available cloud computing power and the number of image terminals managed by the master station, and calculate the edge-allocated computing power of the edge node based on the available edge computing power and the number of image terminals managed by the edge node.

[0066] For example, first, the cloud-allocated computing power of the master station is calculated based on the available computing power of the cloud and the number of image terminals managed by the master station; at the same time, the edge-allocated computing power of the edge node is calculated based on the available computing power of the edge and the number of image terminals managed by the edge node.

[0067] Among them, cloud-allocated computing power refers to the average amount of computing resources allocated on demand to each image terminal from the main station cloud platform, which is used for remote task processing, model reasoning or historical data analysis, etc.; edge-allocated computing power refers to the unit computing resources allocated locally by the edge node to the connected image terminal, which is used for front-end preprocessing, real-time recognition and lightweight control decision-making, etc.

[0068] In one embodiment, the cloud computing power allocation method can be: divide the total cloud computing power available at the master station by the number of image terminals currently managed by the master station to obtain the theoretical cloud resource value that can be allocated to each terminal.

[0069] In one embodiment, the edge allocation computing power may be calculated by dividing the current available computing power of the edge node by the number of image terminals managed by the node to obtain the average available resources of each terminal on the edge side.

[0070] In one embodiment, the above computing power allocation results can be used for task offloading decisions, model running path selection, or computing power scheduling priority setting to achieve optimal configuration of edge and cloud collaborative resources.

[0071] Optionally, Figure 4 A flowchart of cloud computing power allocation calculation provided by the embodiment of this application is given. Figure 4 , the cloud computing power allocation calculation method specifically includes:

[0072] S12021. Determine the cloud unit computing power of the master station based on the available cloud computing power and the number of image terminals managed by the master station.

[0073] Exemplarily, the cloud unit computing power of the master station is determined based on the available cloud computing power and the number of image terminals managed by the master station. The cloud unit computing power refers to the single-terminal cloud computing power that the master station can theoretically allocate to each managed image terminal under the current resource status, which is used to guide cloud task scheduling and resource load balancing.

[0074] In one embodiment, the cloud unit computing power may be determined by dividing the total cloud computing power currently schedulable by the master station by the number of image terminals managed by the master station to obtain the cloud unit computing power value. The formula is as follows:

[0075] Cloud computing power per unit = cloud computing power available ÷ number of image terminals

[0076] In one embodiment, the cloud unit computing power can be used to: guide whether image terminal tasks can be migrated to the cloud for execution, or evenly distribute reasoning and analysis task resources among multiple terminals to achieve dynamic cloud load balancing and resource optimization configuration.

[0077] S12022. Determine a transmission coefficient based on the network status of the master station and the edge node, and determine the cloud-side allocated computing power based on the transmission coefficient and the cloud-side unit computing power.

[0078] For example, the transmission coefficient is determined based on the network status of the master station and edge nodes, and the cloud-allocated computing power is then determined based on this transmission coefficient and the cloud-based unit computing power. The network status refers to the quality of the communication link between the master station and the edge nodes, including parameters such as network bandwidth, transmission delay, and packet loss rate. The transmission coefficient is used to quantify the impact of the network status on the efficiency of remote computing power calls and is a regulatory factor in the cloud-based computing power allocation strategy. The cloud-allocated computing power refers to the total amount of cloud computing resources that can actually be allocated to the image terminal under current network conditions.

[0079] In one embodiment, the heat transfer coefficient may be determined by calculating the transmission coefficient value using a preset weighted formula based on the network bandwidth, average latency, and packet loss rate between the master station and the edge node, for example:

[0080] Transmission coefficient = α × bandwidth utilization - β × average delay - γ × packet loss rate

[0081] Where α, β, and γ are weighted coefficients of bandwidth utilization, average delay, and packet loss rate set based on experience.

[0082] In one embodiment, the cloud computing power allocation method can be: multiplying the cloud unit computing power by the transmission coefficient to obtain the actual cloud computing power that can be allocated, reflecting the impact of network transmission on task offloading capability. The calculation formula is:

[0083] Cloud computing power distribution = cloud computing power per unit × transmission coefficient

[0084] In one embodiment, the introduction of the transmission coefficient can effectively avoid blindly moving tasks to the cloud when the network status is poor, thereby improving the overall computing resource utilization and task processing stability.

[0085] S1203. When the edge-allocated computing power is greater than or equal to the cloud-allocated computing power, the basic characteristics of the device, the operating status characteristics, the detection index characteristics and the working environment characteristics are respectively extracted on the edge node based on the basic data of the device, the operating status data, the detection index data and the working environment data.

[0086] For example, when the edge-allocated computing power is greater than or equal to the cloud-allocated computing power, the edge node extracts basic device features, operating status features, detection indicator features, and working environment features based on the device's basic data, operating status data, detection indicator data, and working environment data. The edge-allocated computing power refers to the total amount of local computing resources that the edge node can provide for each image terminal. When this power is greater than or equal to the cloud-allocated computing power, feature extraction tasks are preferentially performed on the edge side to reduce transmission latency and alleviate cloud load.

[0087] The basic characteristics of the device refer to the terminal identity and usage status characteristics derived from static information such as the device model and operating time; the operating status characteristics refer to the real-time operating load and performance stability characteristics calculated through CPU occupancy, memory usage, temperature and signal strength; the detection index characteristics refer to the task performance capabilities extracted based on key indicators such as image frame rate, recognition accuracy, and alarm frequency; the working environment characteristics refer to the operating environment influencing factors inferred from external conditions such as ambient temperature and vibration intensity.

[0088] In one embodiment, the feature extraction method on the edge node can be: using a locally deployed lightweight feature extraction engine to normalize, perform statistical analysis and structural modeling on the original data to form a standardized feature vector for subsequent fault prediction model call.

[0089] In one embodiment, the edge-side feature extraction process can effectively improve response speed, reduce bandwidth overhead, and enhance the real-time and autonomy of image terminal status assessment.

[0090] S1204. When the edge allocated computing power is less than the actual allocated computing power, the basic characteristics of the device, the operating status characteristics, the detection index characteristics and the working environment characteristics are respectively extracted on the master station based on the basic data of the device, the operating status data, the detection index data and the working environment data.

[0091] For example, when the edge-allocated computing power is less than the actual allocated computing power, the master station extracts basic device features, operating status features, detection indicator features, and working environment features based on the device's basic data, operating status data, detection indicator data, and working environment data. The actual allocated computing power refers to the minimum computing resource threshold required for the current task of each image terminal after dynamic adjustment based on the business load. When the edge-allocated computing power is insufficient to support the feature extraction task, the system will automatically move the processing task to the master station's cloud platform for processing.

[0092] Basic device characteristics refer to structured information used to describe the device's identity and service life, including device model code, working time range, etc.; operating status characteristics refer to a combination of indicators that reflect the device's current operating efficiency and stability, such as CPU load index, memory pressure score, signal strength fluctuation range, etc.; detection indicator characteristics refer to core performance indicators that describe the execution effect of image terminal tasks, such as image processing frame rate, recognition accuracy, alarm trigger rate, etc.; working environment characteristics refer to variables that characterize the physical environment in which the device is located and may affect its operating status, such as ambient temperature level, vibration amplitude level, etc.

[0093] In one embodiment, the master station extracts the above features by receiving and aggregating raw data uploaded from edge nodes, calling a feature extraction algorithm on a high-performance server for batch processing or stream processing, and generating a unified feature vector set.

[0094] In one embodiment, moving the feature extraction task to the master station can alleviate edge computing pressure and ensure the stability and processing accuracy of feature analysis. It is particularly suitable for scenarios with high edge node load or severe network congestion.

[0095] S130: Determine image terminal characteristics based on the basic characteristics of the device, the operating status characteristics, the detection index characteristics, and the working environment characteristics, and predict potential failures of each image terminal within a preset time window based on the image terminal characteristics.

[0096] In some embodiments, image terminal characteristics are first determined based on basic device characteristics, operating status characteristics, detection indicator characteristics, and operating environment characteristics. Image terminal characteristics are a unified representation formed by integrating multiple dimensions of status information, which comprehensively characterizes the current operating health and environmental adaptability of the image terminal. These characteristics may include a composite representation of dimensions such as device aging, operating stability, detection performance, and environmental adaptability.

[0097] Subsequently, based on the image terminal characteristics, the potential failures of each image terminal within the preset time window are predicted. Potential failures refer to abnormal software and hardware conditions that have the risk of occurring within a specific time range in the future but have not yet manifested. The preset time window refers to the short-term or medium-term prediction period set by the system, such as the next 24 hours or the next 7 days.

[0098] In one embodiment, the image terminal features may be determined by combining basic device features, operating status features, detection index features, and working environment features to generate a unified representation using methods such as feature splicing, weighted fusion, or principal component analysis.

[0099] In one embodiment, a method for predicting potential failures may be: inputting image terminal features into a trained time series prediction model, such as LSTM (long short-term memory network), GRU (gated recurrent unit) or random forest regression, etc., combining historical failure data to evaluate the probability and type of future failures, and outputting warning results.

[0100] Optionally, Figure 5 A flow chart of determining the characteristics of an image terminal provided by an embodiment of the present application is given. Figure 5 , the image terminal feature determination method specifically includes:

[0101] S1301: Obtain historical root causes of faults of the image terminal, and determine historical fault types of the image terminal based on the historical root causes of faults.

[0102] Exemplarily, the historical root causes of the image terminal's failures are first obtained, and the historical failure types of the image terminal are determined based on the historical root causes of the failures. The historical root causes of the failures refer to the key causes of the terminal failures determined through log tracing, expert annotation or system analysis in the past operation cycles, and the historical failure types refer to the standardized failure category labels formed after classifying and summarizing the historical root causes of the failures, which are used to support similar problem matching, predictive modeling and risk warning.

[0103] In one embodiment, the root cause of historical faults may be obtained by extracting the fault triggering event, impact scope, and processing records from the device fault log, and combining it with an expert system or causal analysis model to identify the direct and indirect causes of the fault, such as "image blur leading to recognition failure" and "CPU overheating triggering system restart."

[0104] In one embodiment, the method for determining the historical fault type can be: based on a defined fault classification system, such as hardware class, communication class, identification class, environment class, etc., the identified root cause of the fault is mapped to the corresponding fault type label, such as "image acquisition abnormality", "communication interruption", "reduction in recognition accuracy", etc., to achieve unified coding management and subsequent statistical analysis.

[0105] S1302: Based on the historical fault types of the image terminal, feature weighting is performed on the basic features of the device, the operating status features, the detection index features, and the working environment features to obtain image terminal features.

[0106] For example, based on the historical fault types of image terminals, the basic characteristics of the equipment, operating status characteristics, detection index characteristics, and working environment characteristics are weighted to obtain image terminal features. Image terminal features are comprehensive state representations that are obtained by integrating multi-dimensional operating status information and adjusting weights based on historical fault experience. These features serve as input for subsequent fault prediction or health assessment models.

[0107] In one embodiment, the feature weighting method can be: assign different weight coefficients to each type of feature according to the fault sensitivity factor corresponding to the historical fault type. For example, when the historical fault type is "overheating", the temperature index in the operating status feature will obtain a higher weight; when the fault type is "image recognition abnormality", the recognition accuracy and image frame rate in the detection index feature will be emphasized.

[0108] In one embodiment, feature weighting can be achieved by using a preset fault type and feature weight mapping table, or dynamically adjusting the contribution of feature dimensions based on statistical learning methods, and ultimately generating an image terminal feature vector that integrates contextual fault information.

[0109] In one embodiment, the weighting mechanism improves the ability of image terminal features to express the current operating status and potential failure risks, providing a more discriminative input data basis for accurate prediction.

[0110] Optionally, after predicting the potential failure of each of the image terminals within a preset time window based on the image terminal characteristics, the method further includes:

[0111] Based on the potential fault, the basic characteristics of the equipment, the operating status characteristics, the detection index characteristics and the working environment characteristics, a causal link of the potential fault is determined in a preset causal relationship map, and the root cause of the potential fault is determined according to the causal link.

[0112] For example, a causal relationship graph refers to a structured knowledge graph model built between historical fault data and multi-dimensional features. Nodes represent fault events and influencing factors, and edges represent the direction and strength of causal relationships. A causal link refers to a sequence of paths connecting potential faults with various triggering features, reflecting the causal mechanism that may cause potential faults. The root cause of the fault is the key factor node that is triggered earliest and has a significant impact in the causal link, representing the root problem that may cause the potential fault.

[0113] In one embodiment, the causal link can be determined by inputting the potential fault label of the current image terminal and the various extracted feature vectors into a causal graph search engine, and identifying the causal path related to the fault event based on algorithms such as path matching, edge weight calculation, or probability propagation.

[0114] In one embodiment, the root cause of the fault may be determined by performing a traceability analysis on the causal link, combining historical weights or path importance scores, and extracting the first node or the intermediate node with the largest causal contribution as the basis for inferring the root cause of the fault, such as "high temperature causes a decrease in recognition accuracy," where "high temperature" is the root cause.

[0115] In one embodiment, the fault tracing process can be used to assist in operation and maintenance decision-making, intervene in fault occurrence in advance, and optimize image terminal configuration and environment deployment strategies.

[0116] Optionally, Figure 6 A flowchart of the image terminal status visualization provided by an embodiment of the present application is given.

[0117] refer to Figure 6 , the image terminal state visualization method specifically includes:

[0118] S140: Generate a three-dimensional geographical distribution map based on the deployment location of the image terminal, and highlight the image terminal on the three-dimensional geographical distribution map if the image terminal has a potential failure within a preset time window.

[0119] For example, a three-dimensional geographic distribution map is generated based on the deployment locations of image terminals. If an image terminal has a potential fault within a preset time window, the relevant image terminal is highlighted on the three-dimensional geographic distribution map. The three-dimensional geographic distribution map is a visualization of the actual deployment locations of image terminals in the form of three-dimensional spatial coordinates, used to demonstrate the spatial distribution of equipment, operating status, and fault density.

[0120] Potential faults refer to abnormal operating states of image terminals that may occur within a specified time period in the future, as identified by predictive models; highlighting refers to distinguishing potentially faulty equipment from normal equipment through color enhancement, mark enlargement, flashing highlighting, etc., to improve the efficiency of operation and maintenance personnel in perceiving key risk areas.

[0121] In one embodiment, a three-dimensional geographic distribution map may be generated by obtaining the latitude and longitude information and deployment height data of each image terminal, constructing a three-dimensional map coordinate system based on a GIS system or a three-dimensional visualization engine, and mapping the position of each terminal to a scene to generate a corresponding node.

[0122] In one embodiment, the method of highlighting potential faulty image terminals can be: applying visual enhancement styles to image terminal nodes that are judged as "high risk" or "medium risk" in the prediction results, such as red flashing logos, raised icons, animated markers, etc., and can also display their device identification, failure probability, key characteristic indicators and other information to assist operation and maintenance personnel in quickly locating and responding.

[0123] In one embodiment, the three-dimensional distribution map not only improves the visualization effect of fault warning, but also supports regional operation and maintenance resource scheduling and risk distribution situation awareness.

[0124] S150 . In response to a selection operation of an image terminal in the three-dimensional geographical distribution map, basic device data, operating status data, detection index data, and working environment data of the selected image terminal are displayed on the three-dimensional geographical distribution map.

[0125] For example, in response to a selection operation on an image terminal in a three-dimensional geographic distribution map, the basic device data, operating status data, detection indicator data, and working environment data of the selected image terminal are displayed on the three-dimensional geographic distribution map. The selection operation refers to the user selecting a specific image terminal node in the three-dimensional map interface through interactive methods such as clicking, hovering, or selecting a box; the display operation refers to the presentation of key operating information of the image terminal in the form of a pop-up window, floating layer, or sidebar on the interface.

[0126] Basic device data includes device identity and life cycle parameters such as device model, installation time, and operating cycle; operating status data includes real-time performance indicators such as current CPU occupancy, memory usage, device temperature, and signal strength; detection indicator data includes image acquisition frame rate, recognition accuracy, alarm trigger frequency, and other data items that reflect task execution efficiency; working environment data includes external perception parameters related to the physical operating environment of the device, such as ambient temperature, humidity, and vibration intensity.

[0127] In one embodiment, the above information can be displayed in the following manner: after the user selects an image terminal node, the latest status data of the terminal is automatically retrieved from the background database, and a structured information card pops up on the map interface, grouped and displayed by data category, while supporting auxiliary functions such as chart rendering, threshold red marking, and historical trajectory backtracking.

[0128] In one embodiment, the responsive visual display mechanism can help operation and maintenance personnel quickly obtain fault context information, perform problem diagnosis, status comparison and remote intervention operations.

[0129] Optionally, in one embodiment, the overall system architecture includes a core control platform, a communication infrastructure layer, and an intelligent technology supervision module. The core control platform is the production and operation support system master station / IoT platform. As the main control center of the system, the platform is responsible for unified access management of equipment, data flow integration, task process scheduling, rule execution, and result display. At the same time, it provides standardized data interface protocols and supports multiple communication mechanisms, such as MQTT, HTTP, CoAP, etc., to achieve compatible access of equipment from different manufacturers; it also has mechanisms such as device registration and authentication, and access permission classification to ensure the security of data links and platform stability. The communication infrastructure layer builds reliable low-latency data channels through various communication methods, such as 5G public networks, dedicated optical fiber networks, edge gateways, etc.; by introducing network disconnection retransmission mechanisms, cache upload strategies, and link health assessment modules, it ensures communication continuity and complete data transmission.

[0130] The intelligent technical supervision module, deployed at the edge or on the main platform, continuously monitors the operational status and quality of equipment. It primarily consists of an operational status perception subsystem, an intelligent anomaly identification and warning subsystem, a quality assessment engine, a service closed-loop control module, and a data storage and analysis unit. The operational status perception subsystem collects key operational parameters. The intelligent anomaly identification and warning subsystem utilizes a deep learning model based on multimodal data, integrating CNN, LSTM, and graph neural network architectures to proactively predict operational anomalies. It also combines environmental factors and equipment operational trend analysis to provide fault warnings up to 72 hours in advance and identifies anomaly transmission paths based on causal inference models. The quality assessment engine uses multiple scoring mechanisms to construct a quantitative operational quality model for quality assessment. Quality assessment metrics include equipment availability, stability, response time, and recognition accuracy. Visual reports are also provided by manufacturer, region, and equipment type. The service closed-loop control module records the entire equipment maintenance process, including repair request, response execution, issue resolution, and final acceptance, generating quantifiable service quality data. All process data is centrally stored and serves as the core basis for evaluating manufacturer services. The data storage and analysis unit centrally stores operation records, evaluation data and service process information; supports historical data tracking, operation trend analysis and multi-dimensional intelligent report generation.

[0131] Optionally, in one embodiment, a protocol conversion method based on semantic modeling is adopted in the process of determining image device data, and a communication semantic model is constructed using a device interface description language; mainstream or customized protocols are automatically parsed to achieve "plug and play" compatible access capabilities. At the same time, in the process of data processing, combined with a dynamic load scheduling algorithm, the resource allocation of the main platform and edge nodes is adjusted in real time according to the number of device accesses and data traffic, thereby improving the system throughput and concurrent processing capabilities. A quantitative model is constructed based on operating data and service feedback to achieve multi-dimensional quality assessment, supporting rating ranking, classification analysis and trend insights. In terms of data processing, a "three-stage" processing chain is adopted: the edge node completes the primary screening, such as basic noise filtering, feature extraction, etc.; only high-value feature vectors are uploaded to alleviate bandwidth pressure; the main station integrates multi-node data based on federated learning, trains and optimizes the global model, realizes cross-regional recognition and analysis, and controls the response time within 50 milliseconds.

[0132] In processing image terminal maintenance data, blockchain technology is introduced to maintain an immutable record of the entire maintenance process. Smart contracts are used to automatically handle events such as service timeouts. Combined with zero-knowledge proofs, this system ensures a fair and equitable service evaluation system while protecting manufacturer privacy. Regarding visualization, digital twin models of intelligent equipment can be constructed, mapping the equipment's operating status to a three-dimensional spatial structure in real time. This supports augmented reality interaction and spatial location traceability, while also providing intuitive display methods such as distribution views, operating dashboards, and quality heat maps to facilitate rapid fault location and support management decisions.

[0133] Based on the above embodiments, Figure 7 This is a schematic diagram of the structure of the image terminal fault prediction device provided in the embodiment of the present application. Figure 7 The image terminal fault prediction device provided in this embodiment specifically includes: a data acquisition module 21, a feature extraction module 22, and a fault prediction module 23.

[0134] Among them, the data acquisition module 21 is configured to determine the basic device data, operating status data, detection index data and working environment data of several image terminals, the basic device data includes the device model and the device working time, the operating status data includes the device CPU occupancy rate, the device memory usage rate, the device temperature and the device signal strength, the detection index data includes the device image frame rate, the device recognition accuracy and the device alarm frequency, and the working environment data includes the ambient temperature and vibration intensity; the feature extraction module 22 is configured to extract the basic device features, operating status features, detection index features and working environment features based on the basic device data, the operating status data, the detection index data and the working environment data; the fault prediction module 23 is configured to determine the image terminal features based on the basic device features, the operating status features, the detection index features and the working environment features, and predict the potential faults of each of the image terminals within a preset time window based on the image terminal features.

[0135] Based on the above embodiment, the data acquisition module 21 includes: a raw data unit, configured to receive terminal raw data sent by several image terminals, the terminal raw data including basic equipment raw data, operating status raw data, detection index raw data and working environment raw data; a communication protocol unit, configured to perform semantic analysis on the terminal raw data and determine the communication protocol corresponding to each of the terminal raw data; a data acquisition unit, configured to process the terminal raw data based on the communication protocol to obtain the basic equipment data, operating status data, detection index data and working environment data corresponding to each of the image terminals.

[0136] Based on the above embodiment, the feature extraction module 22 includes: an available computing power unit, configured to obtain the edge available computing power of the edge node where the image terminal is located and the cloud available computing power of the master station to which it belongs; an allocation computing power unit, configured to calculate the cloud allocated computing power of the master station based on the cloud available computing power and the number of image terminals managed by the master station, and calculate the edge allocated computing power of the edge node based on the edge available computing power and the number of image terminals managed by the edge node; an edge processing unit, configured to extract device basic features, operating status features, detection index features and working environment features based on the device basic data, the operating status data, the detection index data and the working environment data on the edge node when the edge allocated computing power is greater than or equal to the cloud allocated computing power; and a cloud processing unit, configured to extract device basic features, operating status features, detection index features and working environment features based on the device basic data, the operating status data, the detection index data and the working environment data on the master station when the edge allocated computing power is less than the actual allocated computing power.

[0137] Based on the above embodiment, the computing power allocation unit includes: a unit computing power sub-unit, which is configured to determine the cloud unit computing power of the master station based on the available computing power of the cloud and the number of image terminals managed by the master station; an computing power allocation sub-unit, which is configured to determine the transmission coefficient based on the network status of the master station and the edge node, and determine the cloud allocation computing power based on the transmission coefficient and the cloud unit computing power.

[0138] Based on the above embodiment, the image terminal fault prediction device also includes: a fault root cause module, which is configured to determine the causal link of the potential fault in a preset causal relationship map based on the potential fault, the basic characteristics of the equipment, the operating status characteristics, the detection index characteristics and the working environment characteristics, and determine the root cause of the potential fault according to the causal link.

[0139] Based on the above embodiment, the fault prediction module 23 includes: a fault type unit configured to obtain a historical fault root cause of the image terminal and determine a historical fault type of the image terminal based on the historical fault root cause;

[0140] The feature weighting unit is configured to perform feature weighting on the basic features of the device, the operating status features, the detection index features and the working environment features based on the historical fault types of the image terminal to obtain image terminal features.

[0141] Based on the above embodiment, the image terminal fault prediction device also includes: a three-dimensional map module, which is configured to generate a three-dimensional geographic distribution map based on the deployment location of the image terminal, and highlight the image terminal on the three-dimensional geographic distribution map when there is a potential fault in the image terminal within a preset time window; an operation response module, which is configured to respond to the selection operation of the image terminal in the three-dimensional geographic distribution map, and display the basic equipment data, operating status data, detection index data and working environment data of the selected image terminal on the three-dimensional geographic distribution map.

[0142] The image terminal fault prediction device provided in the embodiments of this application integrates key modules, including a data acquisition module, a feature extraction module, and a fault prediction module, to create an intelligent fault prediction architecture centered on multidimensional data fusion, feature analysis, and early warning judgment. This device, comprised of multiple functional units, forms a closed-loop automated process from device operation data collection, multi-angle feature extraction, to potential fault prediction, significantly improving the accuracy and response speed of image terminal fault prediction.

[0143] The image terminal fault prediction device provided in the embodiment of the present application can be used to execute the image terminal fault prediction method provided in the above embodiment, and has corresponding functions and beneficial effects.

[0144] Figure 8 This is a schematic diagram of the structure of an image terminal fault prediction device provided by an embodiment of the present application, with reference to Figure 8 The image terminal fault prediction device includes: a processor 31, a memory 32, a communication device 33, an input device 34, and an output device 35. The number of processors 31 in the image terminal fault prediction device can be one or more, and the number of memories 32 in the image terminal fault prediction device can be one or more. The processor 31, memory 32, communication device 33, input device 34, and output device 35 of the image terminal fault prediction device can be connected via a bus or other means.

[0145] The memory 32, as a computer-readable storage medium, can be used to store software programs, computer executable programs, and modules, such as the program instructions / modules corresponding to the image terminal fault prediction method of any embodiment of the present application (for example, the data acquisition module 21, the feature extraction module 22, and the fault prediction module 23 in the image terminal fault prediction device). The memory 32 may mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system and at least one application required for a function; the data storage area can store data created based on the use of the device, etc. In addition, the memory 32 may include a high-speed random access memory and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory may further include a memory remotely located relative to the processor, and these remote memories may be connected to the device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0146] The communication device 33 is used for data transmission.

[0147] The processor 31 executes various functional applications and data processing of the device by running the software programs, instructions and modules stored in the memory 32, that is, realizes the above-mentioned image terminal fault prediction method.

[0148] The input device 34 may be used to receive input digital or character information and generate key signal input related to user settings and function control of the device. The output device 35 may include a display device such as a display screen.

[0149] The image terminal fault prediction device provided above can be used to execute the image terminal fault prediction method provided in the above embodiment, and has corresponding functions and beneficial effects.

[0150] An embodiment of the present application also provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to execute an image terminal fault prediction method, the image terminal fault prediction method comprising: determining basic device data, operating status data, detection index data and working environment data of a number of image terminals, the basic device data comprising device model and device working time, the operating status data comprising device CPU occupancy, device memory usage, device temperature and device signal strength, the detection index data comprising device image frame rate, device recognition accuracy and device alarm frequency, and the working environment data comprising ambient temperature and vibration intensity; extracting basic device features, operating status features, detection index features and working environment features based on the basic device data, the operating status data, the detection index data and the working environment data; determining image terminal features based on the basic device features, the operating status features, the detection index features and the working environment features, and predicting potential faults of each of the image terminals within a preset time window based on the image terminal features.

[0151] Storage medium - any of various types of memory devices or storage devices. The term "storage medium" is intended to include: installation media, such as CD-ROMs, floppy disks, or tape drives; computer system memory or random access memory, such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; non-volatile memory, such as flash memory, magnetic media (such as hard disks or optical storage); registers or other similar types of memory elements, etc. Storage media may also include other types of memory or combinations thereof. In addition, the storage medium may be located in the first computer system in which the program is executed, or may be located in a different second computer system that is connected to the first computer system via a network (such as the Internet). The second computer system can provide program instructions to the first computer for execution. The term "storage medium" may include two or more storage media residing in different locations (e.g., in different computer systems connected via a network). The storage medium may store program instructions (e.g., embodied as a computer program) that can be executed by one or more processors.

[0152] Of course, the storage medium containing computer-executable instructions provided in the embodiment of the present application is not limited to the above-mentioned image terminal fault prediction method, and can also execute related operations in the image terminal fault prediction method provided in any embodiment of the present application.

[0153] The image terminal fault prediction device, storage medium and image terminal fault prediction equipment provided in the above embodiments can execute the image terminal fault prediction method provided in any embodiment of the present application. For technical details not described in detail in the above embodiments, please refer to the image terminal fault prediction method provided in any embodiment of the present application.

[0154] The above are only preferred embodiments of the present application and the technical principles employed. The present application is not limited to the specific embodiments described herein, and any obvious changes, readjustments, and substitutions that are apparent to those skilled in the art will not depart from the scope of protection of the present application. Therefore, although the present application has been described in detail through the above embodiments, the present application is not limited to the above embodiments and may include many other equivalent embodiments without departing from the scope of the present application. The scope of the present application is determined by the scope of the claims.

Claims

1. A method for predicting image terminal failure, characterized in that: include: Determine basic device data, operating status data, detection index data, and working environment data of several image terminals. The basic device data includes the device model and device operating hours. The operating status data includes the device CPU occupancy rate, device memory usage rate, device temperature, and device signal strength. The detection index data includes the device image frame rate, device recognition accuracy, and device alarm frequency. The working environment data includes the ambient temperature and vibration intensity. Extracting basic device features, operating status features, detection index features, and working environment features based on the device basic data, the operating status data, the detection index data, and the working environment data; Image terminal characteristics are determined based on the basic characteristics of the equipment, the operating status characteristics, the detection index characteristics, and the working environment characteristics, and potential failures of each image terminal within a preset time window are predicted based on the image terminal characteristics.

2. The image terminal fault prediction method according to claim 1, characterized in that: The determination of basic device data, operating status data, detection index data, and working environment data of a plurality of image terminals includes: Receive terminal raw data sent by several image terminals, wherein the terminal raw data includes basic equipment raw data, operating status raw data, detection index raw data and working environment raw data; Performing semantic analysis on the original terminal data to determine the communication protocol corresponding to each original terminal data; The original terminal data is processed based on the communication protocol to obtain basic device data, operating status data, detection index data and working environment data corresponding to each of the image terminals.

3. The image terminal fault prediction method according to claim 1, characterized in that: Extracting basic device features, operating status features, detection index features, and working environment features based on the device basic data, the operating status data, the detection index data, and the working environment data, respectively, includes: Obtain the edge available computing power of the edge node where the image terminal is located and the cloud available computing power of the master station to which it belongs; Calculating the cloud-allocated computing power of the master station based on the available cloud computing power and the number of image terminals managed by the master station, and calculating the edge-allocated computing power of the edge node based on the available edge computing power and the number of image terminals managed by the edge node; When the edge-allocated computing power is greater than or equal to the cloud-allocated computing power, extracting device basic features, operating status features, detection index features, and working environment features based on the device basic data, the operating status data, the detection index data, and the working environment data, respectively, on the edge node; When the edge allocated computing power is less than the actual allocated computing power, the basic characteristics of the device, the operating status characteristics, the detection index characteristics and the working environment characteristics are respectively extracted on the main station side based on the basic data of the device, the operating status data, the detection index data and the working environment data.

4. The image terminal fault prediction method according to claim 3, characterized in that: Calculating the cloud-allocated computing power of the master station based on the available computing power of the cloud and the number of image terminals managed by the master station includes: Determining the cloud unit computing power of the master station based on the available cloud computing power and the number of image terminals managed by the master station; A transmission coefficient is determined based on the network status of the master station and the edge node, and a cloud-side allocated computing power is determined based on the transmission coefficient and the cloud-side unit computing power.

5. The image terminal fault prediction method according to claim 1, characterized in that: After predicting the potential failure of each of the image terminals within a preset time window based on the image terminal characteristics, the method further includes: Based on the potential fault, the basic characteristics of the equipment, the operating status characteristics, the detection index characteristics and the working environment characteristics, a causal link of the potential fault is determined in a preset causal relationship map, and the root cause of the potential fault is determined according to the causal link.

6. The image terminal fault prediction method according to claim 1, characterized in that: Determining image terminal characteristics based on the basic characteristics of the device, the operating status characteristics, the detection index characteristics, and the working environment characteristics includes: Acquire a historical fault root cause of the image terminal, and determine a historical fault type of the image terminal based on the historical fault root cause; The image terminal features are obtained by weighting the basic features of the equipment, the operating status features, the detection index features and the working environment features based on the historical fault types of the image terminal.

7. The image terminal fault prediction method according to claim 1, characterized in that: After predicting the potential failure of each of the image terminals within a preset time window based on the image terminal characteristics, the method further includes: generating a three-dimensional geographical distribution map based on the deployment location of the image terminal, and highlighting the image terminal on the three-dimensional geographical distribution map if the image terminal has a potential failure within a preset time window; In response to a selection operation of an image terminal in the three-dimensional geographical distribution map, basic device data, operating status data, detection index data and working environment data of the selected image terminal are displayed on the three-dimensional geographical distribution map.

8. An image terminal fault prediction device, characterized in that: include: A data acquisition module is used to determine basic device data, operating status data, detection index data, and working environment data of several image terminals. The basic device data includes the device model and device operating hours. The operating status data includes the device CPU occupancy rate, device memory usage rate, device temperature, and device signal strength. The detection index data includes the device image frame rate, device recognition accuracy, and device alarm frequency. The working environment data includes the ambient temperature and vibration intensity. A feature extraction module is used to extract basic device features, operating status features, detection index features and working environment features based on the basic device data, the operating status data, the detection index data and the working environment data; A fault prediction module is used to determine image terminal characteristics based on the basic characteristics of the equipment, the operating status characteristics, the detection index characteristics and the working environment characteristics, and predict potential faults of each image terminal within a preset time window based on the image terminal characteristics.

9. An image terminal fault prediction device, characterized in that: include: one or more processors; A memory stores one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the image terminal fault prediction method as described in any one of claims 1-7.

10. A storage medium containing computer-executable instructions, characterized in that: The computer executable instructions, when executed by a computer processor, are used to execute the image terminal fault prediction method according to any one of claims 1 to 7.

Images