Fault detection and early warning method and system for navigation beacon data acquisition system and medium

Abstract

The application discloses a beacon data acquisition system fault detection early warning method and system and a medium, and relates to the technical field of fault detection. The method comprises the following steps: key module splitting is performed on a beacon data acquisition system to obtain N modules, historical data sets are constructed based on fault data mining, features are extracted and quantified, cascade analysis is performed to generate an evaluation tree, and a fault evaluation forest is packaged and built; N module working data streams are detected, matching evaluation is performed based on the evaluation forest, and a fault feature set is obtained; fault detection results are generated through fusion analysis, and a hierarchical early warning mechanism is triggered for matching early warning, so that the problems of early warning lag and high false alarm rate caused by the difficulty in tracing the fault of the beacon data acquisition system under complex marine environment due to the coupling fault of multiple sensors are solved, the modular fault evaluation forest and the integrated fusion analysis mechanism are constructed, the collaborative diagnosis of multi-dimensional fault features is realized, and the fault detection sensitivity and accuracy of the beacon data acquisition system under complex marine environment are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fault detection, in particular to a fault detection and early warning method and system for a navigation mark data acquisition system and a medium. BACKGROUND

[0002] As an important infrastructure for marine monitoring and navigation mark management, the navigation mark data acquisition system is widely used in real-time acquisition and transmission of multiple parameters such as wave, current, weather, water quality and visibility. Its running state is directly related to the safety of sea navigation and the accuracy and continuity of marine environment monitoring. Since the system is usually deployed in harsh marine environments far from land, it is exposed to extreme conditions such as high humidity, high salt fog, strong wind and wave, and temperature difference changes for a long time, and it also needs to work with multiple types of sensors, communication interfaces and energy supply modules. Therefore, it is easily affected by factors such as device aging, sensor misalignment, communication abnormalities, power failure, etc. Once the system fails, it may cause monitoring data interruption, accuracy decline or even loss, which has a serious impact on navigation guidance, maritime traffic safety and marine environment warning. However, the existing technology generally has problems such as single monitoring method, difficult fault location, delayed warning, etc. in the fault detection and early warning of the navigation mark data acquisition system. It lacks the ability to model and analyze multiple modules and multiple data sources comprehensively, and cannot realize accurate diagnosis and hierarchical warning of faults, resulting in high maintenance cost, long response time and low system reliability. SUMMARY

[0003] The present application provides a fault detection and early warning method, system and medium for a navigation mark data acquisition system, which solves the problems of delayed warning and high false alarm rate caused by the difficulty in tracing the fault of multiple sensors in the navigation mark data acquisition system under complex marine environment.

[0004] In view of the above problems, the present application provides a fault detection and early warning method, system and medium for a navigation mark data acquisition system.

[0005] In a first aspect, a fault detection and early warning method for a beacon data acquisition system is provided. The method comprises: splitting a target beacon data acquisition system into key modules to obtain N data acquisition key modules; performing fault data mining based on the N data acquisition key modules to construct N data acquisition module fault historical data sets; performing fault feature extraction and quantitative cascade analysis on the N data acquisition module fault historical data sets respectively to generate N data acquisition module fault evaluation trees, and encapsulating and identifying the N data acquisition module fault evaluation trees to build a beacon data acquisition system fault evaluation forest; detecting N data acquisition module working data streams of the N data acquisition key modules, performing matching mapping and fault evaluation on the N data acquisition module working data streams based on the beacon data acquisition system fault evaluation forest to obtain N data acquisition module fault feature sets; performing integrated fusion analysis on the N data acquisition module fault feature sets to generate a beacon data acquisition system fault detection result, and triggering a hierarchical early warning mechanism to perform matching hierarchical early warning on the beacon data acquisition system fault detection result.

[0006] In a second aspect, a fault detection and early warning system for a beacon data acquisition system is provided. The fault detection and early warning system comprises: a fault data mining unit configured to split a target beacon data acquisition system into key modules to obtain N data acquisition key modules, and perform fault data mining based on the N data acquisition key modules to construct N data acquisition module fault historical data sets; an encapsulation and identification unit configured to perform fault feature extraction and quantitative cascade analysis on the N data acquisition module fault historical data sets respectively to generate N data acquisition module fault evaluation trees, and encapsulate and identify the N data acquisition module fault evaluation trees to build a beacon data acquisition system fault evaluation forest; a fault evaluation unit configured to detect N data acquisition module working data streams of the N data acquisition key modules, and perform matching mapping and fault evaluation on the N data acquisition module working data streams based on the beacon data acquisition system fault evaluation forest to obtain N data acquisition module fault feature sets; and a matching hierarchical early warning unit configured to perform integrated fusion analysis on the N data acquisition module fault feature sets to generate a beacon data acquisition system fault detection result, and trigger a hierarchical early warning mechanism to perform matching hierarchical early warning on the beacon data acquisition system fault detection result.

[0007] In a third aspect, a computer readable medium storing a computer program is provided. When the program is executed by a processor, the fault detection and early warning method for a beacon data acquisition system is implemented.

[0008] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0009] Firstly, the target beacon data acquisition system is split into N key modules, fault data mining is performed based on the N key modules, and N data acquisition module fault historical data sets are constructed; subsequently, fault feature extraction and quantitative cascade analysis are performed on the N data acquisition module fault historical data sets, N data acquisition module fault evaluation trees are generated, the N data acquisition module fault evaluation trees are packaged and identified, and a beacon data acquisition system fault evaluation forest is built; further, N data acquisition module working data streams of the N data acquisition key modules are detected, matching mapping and fault evaluation of the N data acquisition module working data streams are performed based on the beacon data acquisition system fault evaluation forest, and N data acquisition module fault feature sets are obtained; finally, the N data acquisition module fault feature sets are integrated and fused, a beacon data acquisition system fault detection result is generated, and a hierarchical early warning mechanism is triggered to match and grade the beacon data acquisition system fault detection result. The problems of early warning lag and high false alarm rate caused by the difficulty in tracing the fault of the beacon data acquisition system in the complex marine environment due to the coupling fault of multiple sensors are solved, and the effects of realizing multi-dimensional fault feature collaborative diagnosis by constructing a modular fault evaluation forest and an integrated fusion analysis mechanism, and improving the fault detection sensitivity and accuracy of the beacon data acquisition system in the complex marine environment are achieved. BRIEF DESCRIPTION OF DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0011] Figure 1 The fault detection and early warning method flowchart of the beacon data acquisition system provided by the embodiments of the present application.

[0012] Figure 2 The fault detection and early warning system structure diagram of the beacon data acquisition system provided by the embodiments of the present application.

[0013] Marked with: fault data mining unit 11, packaging identification unit 12, fault evaluation unit 13, matching and hierarchical early warning unit 14. DETAILED DESCRIPTION

[0014] The present application provides a fault detection and early warning method, system and medium for a beacon data acquisition system, which solves the problems of early warning lag and high false alarm rate caused by the difficulty in tracing the fault of the beacon data acquisition system in the complex marine environment due to the coupling fault of multiple sensors.

[0015] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0016] It should be noted that the terms “comprising” and “having” are intended to cover the inclusion of not exclusive, for example, a process, method, system, product or server containing a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices.

[0017] Embodiment one, as shown in the present application provides a fault detection and early warning method of a beacon data acquisition system, wherein the method comprises: Figure 1

[0018] The target beacon data acquisition system is split into key modules to obtain N data acquisition key modules, and fault data mining is performed based on the N data acquisition key modules to construct N data acquisition module fault historical data sets.

[0019] In one embodiment, the overall structure and function of the target beacon data acquisition system are first comprehensively analyzed to identify various types of hardware, software and communication units contained therein, such as wave sensors, current meters, multi-parameter water quality meters, weather stations, visibility meters, data acquisition and processing boards, communication modules (RS232 / RS485 / 4G / Beidou), storage units, power management units, etc. Subsequently, in combination with application function descriptions, interface definition documents, etc., the target beacon data acquisition system is divided into several independent functional modules according to the functional boundaries, and according to the working environment of the modules, it is finely split into N data acquisition key modules. Then, for each data acquisition key module, collect various types of fault-related data in its historical running process, including but not limited to sensor range abnormalities, data loss, communication interruption, voltage and current abnormalities, storage read / write errors, etc. information, and combine the environmental parameters, running state and time series data at the time of fault occurrence to organize, form N structured fault historical data sets corresponding to the N data acquisition modules, and provide a high-quality data basis for subsequent fault feature extraction, modeling analysis and evaluation.

[0020] Further, the N data acquisition key modules are obtained, including:

[0021] ​The M number of independent function modules are obtained by dividing the module boundary based on the application function information of the target navigation beacon data acquisition system; the environmental simulation parameter table is designed according to the working environment of the target navigation beacon data acquisition system; the M number of independent function modules are analyzed and simulated according to the environmental simulation parameter table, and M number of function module performance parameters are obtained; and the M number of independent function modules are disassembled and subdivided based on the M number of function module performance parameters, and the N number of data acquisition key modules are obtained, wherein N≥M.

[0022] Preferably, the requirements, functions and mutual relationships between the parts of the target navigation beacon data acquisition system are comprehensively analyzed first, including the work tasks, data interaction and control processes undertaken by each module, for example, the application functions of the target navigation beacon data acquisition system can include the measurement tasks of various sensors, data acquisition frequency, communication mode, power supply requirements and signal flow path within the system, and each function can be completed by multiple sub-modules. Subsequently, the boundaries of each function module are determined according to the system functions and the interaction between the modules, combined with the hardware and software design of the target navigation beacon data acquisition system. These modules should have high independence and closure, and can perform specific tasks independently, such as wave sensors responsible for wave monitoring, current meters responsible for current measurement, and multi-parameter water quality meters responsible for water quality analysis. Based on the clear module boundaries, the modules in the target navigation beacon data acquisition system are disassembled into M number of independent function modules, each of which has clear input and output and is responsible for specific work. Then, an environmental simulation parameter table is designed in combination with the typical operating environment of the navigation beacon data acquisition system. This environmental simulation parameter table should cover external conditions that can affect operation, such as temperature, humidity, salt mist concentration, wind speed, wave height, electromagnetic interference level, power supply fluctuation range, and clearly specify the simulation range and change rate of each parameter. Then, the above M number of independent function modules are placed on a laboratory or simulation platform, and are analyzed and simulated one by one according to different working condition combinations in the environmental simulation parameter table. The running performance of each module under different environmental conditions is recorded, and a set of M function module performance parameters including maximum working temperature, response time, data accuracy, communication stability, energy consumption level, fault tolerance capability, etc. are obtained. Finally, based on these function module performance parameters, the performance bottlenecks of each module are analyzed, for example, if a module shows a large error or instability in a high temperature environment, the module may need to be further refined to ensure its reliability in a specific environment. According to the analysis results, the M number of independent function modules are disassembled, for example, the sub-modules with large performance differences are separated out as new modules, or the modules with high functional complexity are functionally decomposed, thereby obtaining N number of data acquisition key modules (N≥M), each of which has independent working components and can be maintained, upgraded or replaced independently, laying a precise modular foundation for subsequent fault detection and early warning analysis.

[0023] Table 1: Environmental simulation parameter example table

[0024]

[0025] As the above Table 1 is an environmental simulation parameter example table, the table shows a variety of external working conditions that the beacon data acquisition system may encounter in the real marine environment, including environmental parameters, simulation range, change rate, and test purpose description, for evaluating the adaptability and stability of each key module in complex marine environments.

[0026] The N data acquisition module fault history data sets are respectively subjected to fault feature extraction and quantitative cascade analysis to generate N data acquisition module fault evaluation trees, and the N data acquisition module fault evaluation trees are packaged and identified to build a beacon data acquisition system fault evaluation forest.

[0027] In one embodiment, after obtaining N data acquisition module fault history data sets, the fault history data of each data acquisition key module is subjected to data cleaning to remove missing values, duplicate values, and invalid records with obvious abnormalities, and is subjected to unified formatting and time alignment processing to ensure the comparability of data from different sources and at different times. Subsequently, the cleaned data acquisition module fault history data sets are identified using a data acquisition module fault label system, and the identified data is subjected to fault feature extraction and quantitative cascade analysis, and a fault evaluation tree is constructed for each data acquisition key module using the name of the data acquisition key module as a root node, each node of the evaluation tree corresponding to a specific fault type, degree, and possible cause, and the node weight representing the possibility and impact degree of the fault. Subsequently, the N module fault evaluation trees constructed are subjected to fault propagation analysis and evaluation using a beacon data acquisition system fault impact data set, and the N module fault evaluation trees are packaged and identified according to the results of the analysis and evaluation to construct a final beacon data acquisition system fault evaluation forest. In this beacon data acquisition system fault evaluation forest, each evaluation tree can independently analyze the fault of the module, and can also realize fault reasoning and joint diagnosis through cross-module coupling relationship, thereby providing a unified decision model basis for subsequent real-time data matching evaluation and hierarchical early warning.

[0028] Further, the generating N data acquisition module fault evaluation trees comprises:

[0029] respectively, the N number of data acquisition module fault history data sets are subjected to data cleaning and standardization processing to obtain N number of standard data acquisition module fault data sets; according to the application function standard of the N number of data acquisition key modules, a data acquisition module fault label system is constructed, the data acquisition module fault label system including fault type, fault degree and fault generation cause; the N number of standard data acquisition module fault data sets are subjected to label evaluation using the data acquisition module fault label system to obtain N number of data acquisition module fault label data sets; the N number of data acquisition module fault label data sets are subjected to fault feature extraction and quantitative cascade analysis respectively to generate N number of data acquisition module fault evaluation trees.

[0030] Preferably, the obtained N data acquisition module fault history data sets are first subjected to data cleaning and standardization processing respectively. The data cleaning stage mainly includes removing duplicate records, correcting missing values, and deleting abnormal values. The duplicate records can be removed by comparing timestamps. The missing values can be corrected by linear interpolation filling (short-term missing) or by filling with the mean or median of the data in the same environment (long-term missing). The abnormal values can be deleted by mean and variance. The standardization processing stage is responsible for unifying the N data acquisition module fault history data sets to the same dimension. This can be done by maximum-minimum normalization or Z-score standardization method, so as to map parameters of different units and different orders of magnitude to a unified scale interval, thereby eliminating the influence of dimension difference on subsequent analysis, and finally obtaining N standard data acquisition module fault data sets. Subsequently, according to the application function standard and operation mechanism of each data acquisition key module, a data acquisition module fault label system is constructed. This data acquisition module fault label system covers three aspects of fault type, fault degree, and fault generation reason. The fault type can be sensor drift, data loss, communication anomaly, power fluctuation, storage error, etc., which is selected according to the application function standard and operation mechanism of the data acquisition key module. The fault degree can be divided into first level, second level, third level (severity decreasing in turn), etc., which is used to reflect the influence degree of the fault on the module function. The fault generation reason is the possible reason for each fault type, such as environmental interference, hardware aging, software anomaly, external impact, etc. Then, the N standard data acquisition module fault data sets are labeled and evaluated using the data acquisition module fault label system. That is, the N standard data acquisition module fault data sets are traversed. In each traversal, the key name of each label in the data acquisition module fault label system is used to compare the contents of the traversal, and the data corresponding to each label is extracted, including fault type label data, fault degree label data, and fault generation reason label data. By organizing these data, N data acquisition module fault label data sets are obtained. Then, for each data acquisition module fault label data set, fault features are extracted, and association rules are mined according to these fault features. These rules are subjected to quantitative cascade analysis to form a data acquisition module fault evaluation tree that can be used for reasoning analysis. The nodes of the evaluation tree represent fault features, the edges represent the relationship between different features, and the leaf nodes represent fault categories, fault degrees, etc., providing an accurate modular decision model for subsequent real-time fault diagnosis and warning.

[0031] Further, the respective N data acquisition module fault label data sets are subjected to fault feature extraction and quantitative cascade analysis to generate N data acquisition module fault evaluation trees, including:

[0032] respectively, and each of the N number of data acquisition module fault feature sets is taken as a row and each of the corresponding fault features is taken as a column to generate N number of data acquisition module fault feature matrices; the N number of data acquisition module fault feature matrices are sequentially subjected to association rule mining to obtain N number of module fault feature association rule sets; and quantitative cascade analysis is performed based on the N number of module fault feature association rule sets to generate N number of data acquisition module fault evaluation trees.

[0033] Optionally, the N number of data acquisition module fault label data set includes multiple fault types, multiple severity, multiple fault reasons of multiple fault record samples, each fault record sample can involve multiple dimensions of features, such as environmental conditions (temperature, humidity), working state (load, current, voltage) and the like. According to the preset fault feature key name, the key features representing the module fault state can be extracted from these fault record samples, which include but are not limited to the frequency of fault occurrence, fault duration, time period of fault occurrence, temperature, humidity, voltage, current, working frequency, signal strength, fault type, severity, communication delay time, packet loss rate, etc. By summarizing these key features, the N number of data acquisition module fault feature sets can be constructed. Then, each fault record sample in the N number of data acquisition module fault feature sets is taken as a row record, and the corresponding data acquisition module fault features are arranged as column fields to construct N number of structured data acquisition module fault feature matrices. The element value in each data acquisition module fault feature matrix represents the value of the sample under a specific feature. Subsequently, the association rule mining algorithm, such as Apriori or FP-Growth, is applied to each module's fault feature matrix in turn to analyze the co-occurrence patterns between different features in the fault samples. Taking Apriori as an example, all possible single feature items are first extracted from the N number of data acquisition module fault feature matrices, the support degree of each feature item in all samples is calculated, and the feature items with support degree higher than the set minimum support threshold are selected to form the frequent 1-item set. Then, based on the frequent k-item set (k≥1), candidate (k+1) item sets are generated by pairwise combination, and the support degree of each candidate item set is calculated to select the item sets that still meet the minimum support requirement to form the frequent (k+1) item set. This process continues until no new frequent item set can be generated. For each obtained frequent item set, possible antecedent (condition) and consequent (conclusion) division modes are enumerated to form multiple rules, such as "power supply voltage fluctuation exceeds threshold and communication delay rises" and "data loss rate significantly increases", and these rules are added to the corresponding association rule set to form N number of module fault feature association rule sets. Then, the module fault feature association rule sets are subjected to quantitative cascade analysis, i.e., the importance of the rules is quantified according to the confidence degree, and the high-value effective rules are selected. Then, the fault cascade analysis and identification are performed according to these effective rules to generate N number of data acquisition module fault evaluation trees. These data acquisition module fault evaluation trees can help identify the root cause of module failure, as well as the path and possible impact of fault propagation, effectively guiding fault warning and repair, and providing basis for fault prediction, detection and optimization of the beacon data acquisition system.

[0034] Further, the quantitative cascade analysis based on the N number of module fault feature association rule sets to generate N number of data acquisition module fault evaluation trees includes:

[0035] The N module fault feature association rule sets are subjected to confidence evaluation to obtain N module association rule confidence sets; the N module association rule confidence sets are subjected to screening according to the N module association rule confidence sets to obtain N module available fault association rule sets; the module names of the N data acquisition key modules are taken as root nodes, and the N module available fault association rule sets are subjected to fault cascade analysis respectively to generate N module fault cascade trees; each fault node in the N module fault cascade trees is subjected to fault degree identification respectively to generate N data acquisition module fault evaluation trees.

[0036] Optionally, after obtaining the N module fault feature association rule sets, the confidence of each rule in the module fault feature association rule set is calculated, that is, the confidence of each rule is calculated by using the frequency of occurrence of each association rule in historical fault data and the conditional satisfaction probability, to calculate the confidence degree of deriving the result feature from the premise feature, for example, for the rule "A→B", the confidence = support(A∩B) / support(A), wherein support(A∩B) refers to the proportion of the number of samples containing both features A and B in the total number of samples in all historical fault data, and support(A) refers to the proportion of the number of samples containing feature A in the total number of samples in all historical fault data. By storing the confidence of each rule according to the storage order of the rules, the N module association rule confidence sets are formed. Subsequently, each confidence in the N module association rule confidence sets is compared with a confidence threshold (for example, 0.7 or 0.8), and the rules lower than the threshold are removed to reduce noise interference and misjudgment risk, thereby obtaining the N module available fault association rule sets. Then, the module name of each data acquisition key module is taken as a root node, and the module available fault association rule set of the data acquisition key module is subjected to fault cascade analysis, in which process, according to the causal relationship between the rules, multiple fault features are hierarchically organized in a logical order, that is, the upper layer nodes represent the first fault features or key trigger conditions, and the lower layer nodes represent the secondary faults or result faults possibly caused by the upper layer nodes, for example, the "power supply voltage fluctuation" rule node of a certain module can trigger "communication delay rise", which further causes "data loss rate rise", and in this way, the available rule set of each module is converted into a module fault cascade tree, thereby obtaining the N module fault cascade trees. Then, according to the fault degree label data, each fault node in each module fault cascade tree is subjected to fault degree identification, and in the identification, color coding, numerical level or weight coefficient, etc. can be used for labeling. Finally, the N module fault cascade trees after labeling are taken as N data acquisition module fault evaluation trees. These data acquisition module fault evaluation trees not only reflect the causal link and severity of the internal faults of the modules, but also provide a directly usable diagnostic tool for subsequent cross-module coupling analysis and system-level early warning, to realize precise early warning.

[0037] Further, the navigation beacon data acquisition system fault evaluation forest is built, comprising:

[0038] Collecting navigation beacon data acquisition system fault influence data set; based on the navigation beacon data acquisition system fault influence data set, the N data acquisition module fault evaluation tree is analyzed and the fault propagation probability is evaluated across the module fault propagation, and the data acquisition module fault propagation directed graph is constructed; according to the data acquisition module fault propagation directed graph, the N data acquisition module fault evaluation tree is coupled and encapsulated across the module and identified, and the navigation beacon data acquisition system fault evaluation forest is built.

[0039] Optionally, when building the beacon data acquisition system fault evaluation forest, first, the fault influence data set of the beacon data acquisition system is collected from the historical fault records of the beacon data acquisition system. This fault influence data set of the beacon data system not only includes the running parameters and fault records of a single module, but also includes the interaction and influence between modules in actual operation. For example, when the communication module appears delay or interruption, it may cause the data of the wave sensor, current meter or multi-parameter water quality instrument to be unable to upload in time, thereby causing data loss; when the output voltage of the power supply module is unstable, it may simultaneously cause the accuracy of multiple sensors to decrease or restart. Subsequently, based on the collected fault influence data set of the beacon data acquisition system, cross-module fault propagation analysis is performed on the N data acquisition module fault evaluation trees. Specifically, the dependency relationship and data flow between different modules are first identified, for example, the power supply module provides energy for multiple measurement modules, and the communication module provides a data transmission channel between each sensor and the server. Then, by using a graph analysis method or a Bayesian network inference model, in combination with the historical fault trigger sequence in the fault influence data set of the beacon data acquisition system, the probability that module A causes module B to have a related fault after a certain fault of module A is calculated, thereby obtaining a cross-module fault propagation probability matrix. Subsequently, the modules are taken as nodes of a graph, the possible fault propagation paths are taken as directed edges, and the propagation probability is taken as the weight of the edges, to construct a data acquisition module fault propagation directed graph. This data acquisition module fault propagation directed graph can directly represent the fault transmission path and influence strength between modules. Then, according to the constructed fault propagation directed graph, the N data acquisition module fault evaluation trees are coupled and encapsulated, that is, the originally independent N data acquisition module fault evaluation trees are linked according to the propagation relationship shown in the directed graph, so that when a specific node of one module evaluation tree is triggered, the corresponding node in another module evaluation tree can be triggered synchronously or its fault probability can be adjusted. After coupling, a module unique identifier, coupling interface information and applicable condition description are added to each coupled evaluation tree, so that the correct coupled model can be called during the operation of the beacon data acquisition system. Through the above steps, all coupled and encapsulated module fault evaluation trees are finally collected together to build a beacon data acquisition system fault evaluation forest, thereby realizing cross-module, multi-level and global fault diagnosis and reasoning analysis capability, helping to identify the fault propagation path, and providing effective basis for fault diagnosis, decision support and early warning system.

[0040] The N data acquisition module working data streams of the N key data acquisition modules are detected, and the N data acquisition module working data streams are matched and mapped and fault evaluated based on the beacon data acquisition system fault evaluation forest, to obtain a N data acquisition module fault feature set.

[0041] In one embodiment, first, N number of data acquisition key modules are detected, and N number of data acquisition module working data streams of the N number of data acquisition key modules are obtained, which include sensor measurement values, communication delays, packet loss rates, signal qualities, error codes, currents, voltages, etc., and are synchronized and arranged according to unified time stamps and data formats. Finally, the N number of data acquisition module working data streams are matched and mapped to corresponding module fault assessment trees on the beacon data acquisition system fault assessment forest, and the module fault assessment trees are used to identify and evaluate the data acquisition module working data streams. Then, the beacon data acquisition system fault assessment forest is used to correct the identification and evaluation results, and N number of data acquisition module fault feature sets are obtained, which are used as basic data for subsequent fault fusion analysis and hierarchical early warning, and provide reliable basis for realizing real-time and accurate beacon data acquisition system fault detection.

[0042] Further, the N number of data acquisition module fault feature sets are obtained, including:

[0043] Based on the matching and mapping of the N number of data acquisition module working data streams by the beacon data acquisition system fault assessment forest, N number of matching module fault assessment trees are loaded; the N number of matching module fault assessment trees are used for feature windowing and fault assessment of the N number of data acquisition module working data streams, and N number of module working fault feature sets are obtained; based on the beacon data acquisition system fault assessment forest, cross-module influence correction is performed on the N number of module working fault feature sets, and the N number of data acquisition module fault feature sets are obtained.

[0044] Preferably, the N number of data acquisition module working data streams collected are first matched and mapped to the corresponding module fault assessment tree in the beacon data acquisition system fault assessment forest according to the module name, and then the mapped module fault assessment tree is loaded to obtain N matching module fault assessment trees. Subsequently, each matching module fault assessment tree will perform feature windowing and fault assessment on the internally mapped data acquisition module working data stream, that is, key feature data such as signal peak value, fluctuation amplitude, transmission delay, packet loss rate, voltage, temperature, electromagnetic interference, etc. is extracted within the sliding feature window set based on the sampling period, and these window features are triggered and probabilistically inferred layer by layer to calculate the fault type, fault degree, fault generation reason, etc. of each candidate fault node in the window, and then combine these information with the key feature data to form N module working fault feature sets. Subsequently, based on the data acquisition module fault propagation directed graph and propagation probability in the beacon data acquisition system fault assessment forest, the potential impact of a module fault on other modules is analyzed, and the N module working fault feature sets obtained are corrected for impact, for example, if the voltage fluctuation fault node of the power supply module is triggered, and the propagation probability model shows that this fault will cause the communication module delay to rise with a high probability, then the corresponding impact weight is added to the fault feature set of the communication module or its fault probability is adjusted. After cross-module impact correction, N number of data acquisition module fault feature sets are obtained, which provide more accurate and comprehensive basic data for subsequent global fault fusion analysis and hierarchical early warning.

[0045] The N number of data acquisition module fault feature sets are integrated and fused to generate a beacon data acquisition system fault detection result, and a hierarchical early warning mechanism is triggered to match and perform hierarchical early warning on the beacon data acquisition system fault detection result.

[0046] In one embodiment, after obtaining the N number of data acquisition module fault feature sets, integrated fusion analysis is performed on the N number of data acquisition module fault feature sets, that is, the fault features of each module are weighted and fused to form a unified fault fusion feature set, and then fault mode recognition and root cause analysis are performed according to these fault fusion feature sets to determine whether the current beacon data acquisition system has a single module fault, a multi-module coupled fault, or a chain fault, and to identify the main cause of the fault, such as environmental factors, hardware aging, communication interference, or energy shortage. By integrating these analysis results, the beacon data acquisition system fault detection result can be obtained. Then, based on the fault detection result, a hierarchical warning mechanism is triggered, and in this process, the beacon data acquisition system fault detection result in the beacon data acquisition system fault detection result is compared with the preset level in the hierarchical warning mechanism to obtain the warning scheme corresponding to the current detection, for example, a first-level warning is used to prompt a serious fault that may affect the core function or data integrity of the system, and emergency measures need to be taken immediately; a second-level warning is used to prompt a moderate fault that may affect part of the function, and on-site or remote maintenance is recommended; a third-level warning is used to prompt a slight anomaly, record logs and continuously monitor. Finally, the matched warning scheme is sent in real time to the monitoring center or operation and maintenance terminal through the communication link, and is presented on the client interface of the Internet of Things platform in the form of highlighting, sound and light alarm, etc., so that the operation and maintenance personnel can obtain key information and take corresponding measures in the shortest time.

[0047] Further, the beacon data acquisition system fault detection result is generated, including:

[0048] The N number of data acquisition module fault feature sets are spatio-temporally aligned and weighted fused to obtain a data acquisition system fault fusion feature set; the data acquisition system fault fusion feature set is subjected to fault mode clustering and root cause analysis to determine a data acquisition system fault mode set and a data acquisition system fault root cause set; the data acquisition system fault fusion feature set is subjected to fault degree summation quantification to obtain a data acquisition system fault degree level, and the data acquisition system fault mode set and the data acquisition system fault root cause set are integrated with the data acquisition system fault degree level to generate the beacon data acquisition system fault detection result.

[0049] Preferably, after obtaining the N number of data acquisition module fault feature sets, the N number of data acquisition module fault feature sets are synchronized according to a unified time axis, and the geographical position, data acquisition period, event triggering time and the like are uniformly calibrated to ensure the comparability between the module features. Subsequently, according to the functional importance of each module in the navigation beacon data acquisition system, the frequency of fault occurrence, the influence range of module performance and the like, the corresponding weight coefficient is set, and the fault probability and fault degree of each module are weighted and calculated, so as to obtain the data acquisition system fault fusion feature set reflecting the overall operation state of the navigation beacon data acquisition system. Then, the clustering algorithm (such as K-means, DBSCAN or hierarchical clustering) is used to cluster the data acquisition system fault fusion feature set, identify the fault type group with similar feature distribution, and form the data acquisition system fault mode set. At the same time, combined with historical data and causal analysis method (such as Bayesian network analysis, causal inference), the source conditions, triggering link and main causes of fault occurrence are further traced in the data acquisition system fault mode set, and the data acquisition system fault root cause set is generated. For example, the elbow rule or contour coefficient method is used to determine the appropriate number of clusters K, and K cluster center points are randomly initialized from the data acquisition system fault fusion feature set. Then, the data acquisition system fault fusion features are assigned to the cluster corresponding to the center point with the nearest Euclidean distance, and the center point coordinates are updated according to the mean position of the samples in each cluster. This process is repeated until the center point position changes less than a predetermined threshold or the maximum iteration number is reached, so as to obtain the data acquisition system fault mode set. Each mode in the data acquisition system fault mode set represents a typical fault performance in the operation process. After obtaining the data acquisition system fault mode set, the Bayesian network is used for causal reasoning of the data acquisition system fault mode set. In this process, the directed acyclic graph structure of the Bayesian network is initially constructed based on expert experience and data statistics, wherein the nodes represent variables and the edges represent potential causal relationships. Then, the structure learning algorithm is used to optimize the preliminary network structure, remove redundant edges and supplement missing relationships. After the structure is determined, the conditional probability distribution of each node is calculated by maximum likelihood estimation or Bayesian estimation method. Then, the current data acquisition system fault mode set is input into the trained Bayesian network, and the variable elimination method and other reasoning algorithms are used to calculate the posterior probability of each potential cause node, to determine the most likely root cause and causal link, and to form the data acquisition system fault root cause set. This data acquisition system fault root cause set not only reveals the direct cause of the fault, but also identifies indirect triggering factors across modules, providing a scientific basis for accurate early warning and efficient operation and maintenance of the navigation beacon data acquisition system. Then, the fault degree of the data acquisition system fault fusion feature set is quantitatively processed, that is, the weighted total score of the fault degree of each module in the data acquisition system fault fusion feature set is calculated, and the weighted total score is compared with the fault degree level mapping table to determine the fault degree level of the current data acquisition system.Finally, the number of acquisition system fault mode set, the number of acquisition system fault root cause set and the number of acquisition system fault degree are integrated to generate a complete navigation beacon data acquisition system fault detection result, and to provide comprehensive and executable decision basis for operation and maintenance personnel.

[0050] To sum up, the embodiments of the application have at least the following technical effects:

[0051] Firstly, the target navigation beacon data acquisition system is split into N number of key modules, and N number of data acquisition module fault historical data sets are constructed based on the N number of data acquisition key modules. Subsequently, N number of data acquisition module fault evaluation trees are generated by performing fault feature extraction and quantitative cascade analysis on the N number of data acquisition module fault historical data sets, and the N number of data acquisition module fault evaluation trees are encapsulated and identified to build a navigation beacon data acquisition system fault evaluation forest. Further, N number of data acquisition module working data streams of the N number of data acquisition key modules are detected, and the N number of data acquisition module working data streams are matched and mapped based on the navigation beacon data acquisition system fault evaluation forest for fault evaluation to obtain N number of data acquisition module fault feature sets. Finally, the N number of data acquisition module fault feature sets are integrated and fused to generate a navigation beacon data acquisition system fault detection result, and a hierarchical early warning mechanism is triggered to match and grade the navigation beacon data acquisition system fault detection result. The problem of early warning lag and high false alarm rate caused by the difficulty in tracing the fault of the navigation beacon data acquisition system in the complex marine environment due to the coupling of multiple sensors is solved, and the effect of realizing the cooperative diagnosis of multi-dimensional fault features by constructing a modular fault evaluation forest and an integrated fusion analysis mechanism is achieved, and the fault detection sensitivity and accuracy of the navigation beacon data acquisition system in the complex marine environment are improved.

[0052] Embodiment two, based on the same inventive concept as the fault detection and early warning method of the navigation beacon data acquisition system in the foregoing embodiments, such as Figure 2As shown, the application provides a fault detection and early warning system for a beacon number acquisition system, wherein the system comprises: a fault data mining unit 11: performing key module splitting on a target beacon number acquisition system to obtain N number acquisition key modules, performing fault data mining based on the N number acquisition key modules, and constructing N number acquisition module fault historical data sets; an encapsulation identification unit 12: respectively performing fault feature extraction and quantitative cascade analysis on the N number acquisition module fault historical data sets, generating N number acquisition module fault evaluation trees, and encapsulating and identifying the N number acquisition module fault evaluation trees to build a beacon number acquisition system fault evaluation forest; a fault evaluation unit 13: detecting N number acquisition module working data streams of the N number acquisition key modules, performing matching mapping and fault evaluation on the N number acquisition module working data streams based on the beacon number acquisition system fault evaluation forest, and obtaining N number acquisition module fault feature sets; and a matching hierarchical early warning unit 14: performing integrated fusion analysis on the N number acquisition module fault feature sets, generating a beacon number acquisition system fault detection result, and triggering a hierarchical early warning mechanism to perform matching hierarchical early warning on the beacon number acquisition system fault detection result.

[0053] Further, the fault data mining unit 11 is used to perform the following method:

[0054] Based on the application function information of the target beacon number acquisition system, module boundary division is performed to obtain M number acquisition independent function modules; based on the working environment of the target beacon number acquisition system, an environment simulation parameter table is designed; based on the environment simulation parameter table, the M number acquisition independent function modules are analyzed and simulated to obtain M function module performance parameters; based on the M function module performance parameters, the M number acquisition independent function modules are disassembled and subdivided to obtain the N number acquisition key modules, wherein N≥M.

[0055] Further, the encapsulation identification unit 12 is used to perform the following method:

[0056] The N number acquisition module fault historical data sets are respectively subjected to data cleaning and standardization processing to obtain N standard number acquisition module fault data sets; based on the application function standards of the N number acquisition key modules, a number acquisition module fault label system is constructed, the number acquisition module fault label system comprising fault types, fault degrees, and fault generation reasons; the N standard number acquisition module fault data sets are subjected to label evaluation using the number acquisition module fault label system to obtain N number acquisition module fault label data sets; the N number acquisition module fault label data sets are respectively subjected to fault feature extraction and quantitative cascade analysis to generate N number acquisition module fault evaluation trees.

[0057] Further, the encapsulation identification unit 12 is used to perform the following method:

[0058] Fault features are extracted from the fault label datasets of the N data acquisition modules respectively to obtain N fault feature sets of data acquisition modules; each sample in the fault feature sets of the N data acquisition modules is used as a row and the corresponding fault feature is used as a column to generate N fault feature matrices of data acquisition modules; association rule mining is performed on the fault feature matrices of the N data acquisition modules in sequence to obtain N association rule sets of module fault features; quantitative cascade analysis is performed on the association rule sets of module fault features to generate N fault evaluation trees of data acquisition modules.

[0059] Furthermore, the encapsulation identification unit 12 is used to perform the following method:

[0060] The confidence level of the N module fault feature association rule sets is evaluated to obtain N module association rule confidence sets; the N module fault feature association rule sets are then filtered according to the N module association rule confidence sets to obtain N module usable fault association rule sets; the module names of the N key data acquisition modules are used as root nodes, and fault cascading analysis is performed on the N module usable fault association rule sets to generate N module fault cascading trees; the fault degree of each fault node in the N module fault cascading trees is identified to generate N data acquisition module fault evaluation trees.

[0061] Furthermore, the encapsulation identification unit 12 is used to perform the following method:

[0062] Collect a dataset of the impact of navigation aid data acquisition system failures; based on the dataset, perform cross-module failure propagation analysis and failure propagation probability assessment on the failure assessment trees of the N data acquisition modules, and construct a directed graph of data acquisition module failure propagation; according to the directed graph of data acquisition module failure propagation, perform cross-module coupling and encapsulation identification on the failure assessment trees of the N data acquisition modules, and build a navigation aid data acquisition system failure assessment forest.

[0063] Furthermore, the fault assessment unit 13 is used to perform the following method:

[0064] Collect a dataset of the impact of navigation aid data acquisition system failures; based on the dataset, perform cross-module failure propagation analysis and failure propagation probability assessment on the failure assessment trees of the N data acquisition modules, and construct a directed graph of data acquisition module failure propagation; according to the directed graph of data acquisition module failure propagation, perform cross-module coupling and encapsulation identification on the failure assessment trees of the N data acquisition modules, and build a navigation aid data acquisition system failure assessment forest.

[0065] Furthermore, the matching hierarchical early warning unit 14 is used to perform the following method:

[0066] The N number of sampling module fault feature sets are spatiotemporally aligned and weightedly fused to obtain a number of sampling system fault fusion feature sets; the number of sampling system fault fusion feature sets are subjected to fault mode clustering and root cause analysis to determine a number of sampling system fault mode sets and a number of sampling system fault root cause sets; the number of sampling system fault fusion feature sets are subjected to fault degree summation quantification to obtain a number of sampling system fault degree grades, and the number of sampling system fault mode sets and the number of sampling system fault root cause sets, and the number of sampling system fault degree grades are integrated to generate the navigation beacon number of sampling system fault detection result.

[0067] In the same inventive concept as the navigation beacon number of sampling system fault detection and early warning method in the foregoing embodiments, the application provides a medium, and the medium stores a computer program. When the processor executes the computer program, the following steps are implemented: the target navigation beacon number of sampling system is subjected to key module splitting to obtain N number of sampling key modules, fault data mining is performed based on the N number of sampling key modules to construct N number of sampling module fault historical data sets; the N number of sampling module fault historical data sets are respectively subjected to fault feature extraction and quantification cascade analysis to generate N number of sampling module fault evaluation trees, and the N number of sampling module fault evaluation trees are packaged and identified to build a navigation beacon number of sampling system fault evaluation forest; N number of sampling module working data streams of the N number of sampling key modules are detected, the N number of sampling module working data streams are subjected to matching mapping and fault evaluation based on the navigation beacon number of sampling system fault evaluation forest to obtain N number of sampling module fault feature sets; the N number of sampling module fault feature sets are subjected to integrated fusion analysis to generate a navigation beacon number of sampling system fault detection result, and a hierarchical early warning mechanism is triggered to match and grade the navigation beacon number of sampling system fault detection result.

[0068] It should be noted that the above-mentioned sequence of the embodiments of the application is only for description, and does not represent the advantages and disadvantages of the embodiments. The above describes a specific embodiment of the present application. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0069] The above only describes the preferred embodiments of the application and does not limit the application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application shall be included in the protection scope of the application.

[0070] The specification and drawings are, of course, to be regarded in an illustrative rather than a restrictive sense. It is to be understood that any such modifications, variations, combinations or equivalents that fall within the scope of the application are intended to be embraced herein.

Claims

1. A fault detection and early warning method for a beacon number collection system, characterized in that, The method includes: The target navigation beacon data acquisition system is broken down into key modules to obtain N key data acquisition modules. Fault data mining is performed based on the N key data acquisition modules to construct a historical dataset of faults for the N data acquisition modules. Fault feature extraction and quantitative cascade analysis are performed on the historical fault datasets of the N data acquisition modules respectively to generate N data acquisition module fault evaluation trees. The N data acquisition module fault evaluation trees are then encapsulated and labeled to build a fault evaluation forest for the navigation beacon data acquisition system. The working data streams of the N data acquisition modules of the N key data acquisition modules are detected. Based on the fault assessment forest of the navigation beacon data acquisition system, the working data streams of the N data acquisition modules are matched, mapped and assessed for faults to obtain the fault feature set of the N data acquisition modules. The fault feature sets of the N data acquisition modules are integrated and analyzed to generate fault detection results of the navigation beacon data acquisition system, and a graded early warning mechanism is triggered to match and classify the fault detection results of the navigation beacon data acquisition system for early warning. The obtained N key data acquisition modules include: Based on the application function information of the target navigation beacon data acquisition system, module boundaries are divided to obtain M independent data acquisition function modules; Based on the working environment of the target navigation beacon data acquisition system, design an environment simulation parameter table; The M independent data acquisition functional modules are analyzed and simulated according to the environmental simulation parameter table to obtain the performance parameters of the M functional modules. Based on the performance parameters of the M functional modules, the M independent data acquisition functional modules are decomposed and subdivided to obtain the N key data acquisition modules, where N≥M; The generation of N data acquisition module fault assessment trees includes: Data cleaning and standardization processes are performed on the N historical fault datasets of the data acquisition modules respectively to obtain N standard fault datasets of the data acquisition modules. Based on the application function standards of the N key data acquisition modules, a fault labeling system for data acquisition modules is constructed. The fault labeling system for data acquisition modules includes fault type, fault degree, and fault generation cause. The fault labeling system of the data acquisition module is used to evaluate the faults of the N standard data acquisition modules to obtain N fault label datasets of data acquisition modules. Fault feature extraction and quantization cascade analysis are performed on the fault label datasets of the N data acquisition modules respectively to generate N data acquisition module fault evaluation trees; The step involves extracting fault features and performing quantized cascade analysis on the fault label datasets of the N data acquisition modules to generate N fault assessment trees for the data acquisition modules, including: Fault features are extracted from the fault label datasets of the N data acquisition modules respectively to obtain N fault feature sets of data acquisition modules; The N data acquisition module fault feature matrices are generated by taking each sample in the N data acquisition module fault feature sets as rows and the corresponding fault features as columns. The association rules are sequentially mined from the fault feature matrices of the N data acquisition modules to obtain the N module fault feature association rule sets; Based on the set of association rules for the fault features of the N modules, a quantitative cascade analysis is performed to generate N fault assessment trees for the data acquisition modules.

2. The method of claim 1, wherein the method further comprises: The quantitative cascade analysis based on the N module fault feature association rule set generates N number of acquisition module fault evaluation trees, including: The confidence evaluation is performed on the N module fault feature association rule set to obtain an N module association rule confidence set; The N module fault feature association rule set is screened according to the N module association rule confidence set to obtain an N module available fault association rule set; The module name of the N number of acquisition key module is taken as a root node, and the fault cascade analysis is performed on the N module available fault association rule set to generate an N module fault cascade tree; The fault degree identification is performed on each fault node in the N module fault cascade tree to generate an N number of acquisition module fault evaluation tree.

3. The method of claim 2, wherein the method further comprises: determining whether the beacon number acquisition system is in a normal state or an abnormal state based on the comparison result; and outputting a warning signal when the beacon number acquisition system is in the abnormal state. The fault evaluation forest of the beacon number acquisition system is built, including: The beacon number acquisition system fault influence data set is collected; Based on the beacon number acquisition system fault influence data set, the cross-module fault propagation analysis and fault propagation probability evaluation are performed on the N number of acquisition module fault evaluation tree to construct a number acquisition module fault propagation directed graph; According to the number acquisition module fault propagation directed graph, the cross-module coupling and encapsulation identification are performed on the N number of acquisition module fault evaluation tree to build a beacon number acquisition system fault evaluation forest.

4. The method of claim 1, wherein the method further comprises: determining whether the beacon number acquisition system is in a failure state based on the comparison result; and outputting a warning signal when the beacon number acquisition system is in the failure state. The N number of acquisition module fault feature set is obtained, including: Based on the beacon number acquisition system fault evaluation forest, the N number of acquisition module working data flow is matched and mapped to load N matching module fault evaluation trees; The feature windowing and fault evaluation are performed on the N number of acquisition module working data flow by using the N matching module fault evaluation trees to obtain an N module working fault feature set; Based on the beacon number acquisition system fault evaluation forest, the cross-module influence correction is performed on the N module working fault feature set to obtain the N number of acquisition module fault feature set.

5. The method of claim 1, wherein the method further comprises: determining whether the beacon number acquisition system is in a failure state based on the comparison result; and outputting a warning signal when the beacon number acquisition system is in the failure state. The beacon number acquisition system fault detection result is generated, including: The N number of acquisition module fault feature set is spatio-temporally aligned and weighted fused to obtain a number acquisition system fault fusion feature set; The fault mode clustering and root cause analysis are performed on the number acquisition system fault fusion feature set to determine a number acquisition system fault mode set and a number acquisition system fault root cause set; The fault degree summation quantization is performed on the number acquisition system fault fusion feature set to obtain a number acquisition system fault degree level, and the number acquisition system fault mode set and the number acquisition system fault root cause set, and the number acquisition system fault degree level are integrated to generate the beacon number acquisition system fault detection result.

6. The fault detection and early warning system of the beacon number collection system, characterized in that, The fault detection and early warning method for the beacon number acquisition system according to any one of claims 1-5, the fault detection and early warning system comprising: A fault data mining unit: the target beacon number acquisition system is split into N number of acquisition key modules, and the fault data mining is performed based on the N number of acquisition key modules to construct N number of acquisition module fault historical data set; The encapsulation identification unit performs fault feature extraction and quantitative cascade analysis on the N number of data acquisition module fault history data sets respectively, generates N number of data acquisition module fault evaluation trees, encapsulates and identifies the N number of data acquisition module fault evaluation trees, and builds a beacon data acquisition system fault evaluation forest; The fault evaluation unit detects N number of data acquisition module working data streams of the N number of data acquisition key modules, performs matching mapping and fault evaluation on the N number of data acquisition module working data streams based on the beacon data acquisition system fault evaluation forest, and obtains N number of data acquisition module fault feature sets; The matching hierarchical early warning unit performs integrated fusion analysis on the N number of data acquisition module fault feature sets, generates a beacon data acquisition system fault detection result, and triggers a hierarchical early warning mechanism to perform matching hierarchical early warning on the beacon data acquisition system fault detection result.

7. A computer readable medium having stored thereon a computer program, characterized in that The program is executed by the processor to realize the fault detection and early warning method of the beacon data acquisition system as claimed in any one of claims 1-5.

Images