Intelligent Fault Warning System for Ship Environmental Monitoring Equipment

Abstract

The present invention discloses an intelligent fault warning system for ship environmental monitoring equipment, which relates to the technical field of data prediction and analysis. It includes a data transmission monitoring module, a computing resource monitoring module, a real-time evaluation module, a classification module, and a decision optimization module. The data transmission monitoring module uses timestamp technology to dynamically analyze the degree of network delay interference, and combines with the computing resource monitoring module to collect the CPU occupancy rate and fluctuation amplitude in real time. The real-time evaluation module comprehensively analyzes the degree of delay and computing resource anomalies to accurately evaluate the real-time performance ability of the system. The classification module divides the warnings into real-time and non-real-time warnings. The decision optimization module conducts in-depth analysis and optimizes decisions for non-real-time warnings, effectively improving the accuracy and response efficiency of fault warnings, achieving efficient and accurate warnings for ship faults in complex environments, and significantly improving the stability of the system and the reliability of operators' decisions.

Description

Technical Field

[0001] The present invention relates to the technical field of data prediction and analysis, and particularly to an intelligent fault warning system for ship environmental monitoring equipment. Background Art

[0002] The intelligent fault warning of ship environmental monitoring equipment refers to the use of modern intelligent technologies to analyze and predict the operating status of ship environmental monitoring equipment in real time, so as to issue a warning signal before a fault occurs. Through a sensor network, a data acquisition system, and intelligent algorithms (such as machine learning and big data analysis), this technology continuously monitors the key performance indicators of the equipment (such as temperature, humidity, gas concentration, or vibration). When the system detects an abnormal trend or a potential risk that may lead to a fault, it can automatically generate a warning notification to remind relevant personnel to take preventive measures.

[0003] This kind of intelligent fault warning has significant advantages. First of all, it can help ship operators discover potential problems in advance, reduce unexpected shutdowns or safety accidents caused by equipment failures, and improve operation efficiency and safety. Secondly, the warning system can provide detailed fault cause analysis and optimization suggestions for maintenance personnel, support accurate decision-making, and thus reduce maintenance costs and time investment. This intelligent method not only improves the reliability of monitoring equipment, but also provides technical support for the green shipping and efficient management of ships.

[0004] The existing technologies have the following deficiencies:

[0005] The fault warning system relies on high-frequency data transmission and real-time analysis. However, the limited ship network environment may cause abnormalities in the operating status of environmental monitoring equipment, resulting in data delay or packet loss. Coupled with insufficient edge computing capabilities, the analysis speed and efficiency of the system decline, and the warning cannot be issued in time at the initial stage of the fault, missing the best intervention opportunity. Moreover, operators may make wrong decisions due to the delayed warning signal, such as delaying shutdown or taking ineffective intervention measures, further deteriorating the fault situation. Summary of the Invention

[0006] The purpose of the present invention is to provide an intelligent fault warning system for ship environmental monitoring equipment to solve the deficiencies in the background art.

[0007] To achieve the above purpose, the present invention provides the following technical solution: An intelligent fault warning system for ship environmental monitoring equipment, including a data transmission monitoring module, a computing resource monitoring module, a real-time evaluation module, a classification module, and a decision optimization module;

[0008] Data Transmission Monitoring Module: Records the sending time and receiving time of sensor data from the acquisition end to the edge computing unit through timestamp technology, calculates the transmission delay of a single data packet, conducts several measurements under different network conditions, and determines the overall interference degree of the data packet transmission delay;

[0009] Computing Resource Monitoring Module: When the overall interference degree of the data packet transmission delay is high, uses system monitoring tools to collect the CPU occupancy rate of the edge computing device in real time, and analyzes the fluctuation range of the real-time collected CPU occupancy rate to determine the abnormal degree of the CPU occupancy rate of the edge computing device;

[0010] Real-time Evaluation Module: In the fault warning task, comprehensively analyzes the overall interference degree of the data packet transmission delay and the abnormal degree of the CPU occupancy rate of the edge computing device, and evaluates the real-time performance ability of the warning system in the fault warning task;

[0011] Classification Module: According to the evaluation results, classifies the real-time performance ability of the warning system in the fault warning task into real-time warning and non-real-time warning;

[0012] Decision Optimization Module: For non-real-time warnings, deeply analyzes the abnormal degree of the real-time performance ability of the warning system in the fault warning task within a fixed time period, and makes decision optimization according to the analysis results to improve the accuracy of fault warning.

[0013] Preferably, in the data transmission monitoring module, after analyzing the overall change trend of the data delay, a data delay drift index is generated. The method for obtaining the data delay drift index is as follows:

[0014] Continuously measure the transmission delay of the data packet multiple times; Obtain the delay sample value set: {Delay1, Delay2,..., Delay n}, calculate the mean μ of the delay samples, and the expression is: Calculate the standard deviation σ of the delay: In the formula, Delay i is the transmission delay of the i-th data packet, and n is the number of samples; Calculate the distribution symmetry S of the delay data, and the expression is: σ 3 represents normalizing the deviation value; Select the skewness coefficient S baseline of the delay distribution in the reference time period as the reference value for judging the current delay drift, and calculate the data delay drift index, and the expression is: HK = |S - S baseline |; In the formula, HK is the data delay drift index.

[0015] Preferably, in the computing resource monitoring module, after analyzing the fluctuation range of the real-time collected CPU occupancy rate, a CPU occupancy rate fluctuation index is generated. The method for obtaining the CPU occupancy rate fluctuation index is as follows:

[0016] Record the CPU occupancy rate data in real time to form a time series: {CPU t}, t = 1, 2, …, N; N is the total number of sampling points. Remove the mean component from the data to eliminate the interference of the DC component on the frequency domain analysis: CPU t ′ = CPU t - μe, CPU t ′ is the processed CPU occupancy rate, and μe is the mean of the CPU occupancy rate data in the time series; Apply the fast Fourier transform to the time series {CPU t ′} to obtain the frequency domain complex signal: In the formula, F(f k ) is the complex value of the frequency domain signal at frequency f k ; Calculate the power spectral density P(f k ), and the expression is: P(f k ) = |F(f k )| 2 ; Set the frequency threshold f threshold , and regard the part higher than the frequency, that is, the high-frequency component, as the fluctuation component: Calculate the total energy E high of the high-frequency component, and the expression is: Calculate the total energy E total , and the expression is: Calculate the high-frequency energy ratio as the CPU occupancy rate fluctuation index, and the expression is: GH is the CPU occupancy rate fluctuation index.

[0017] Preferably, in the real-time evaluation module, comprehensively analyze the overall interference degree of the data packet transmission delay and the abnormal degree of the CPU occupancy rate of the edge computing device to evaluate the real-time performance ability of the early warning system in the fault early warning task. Specifically:

[0018] Convert the data delay drift index and the CPU occupancy rate fluctuation index into a comprehensive feature vector, and use the comprehensive feature vector as the input of the machine learning model. The machine learning model takes the prediction of the real-time performance ability value label of the early warning system in the fault early warning task for each group of comprehensive feature vectors as the prediction target, and takes minimizing the sum of the prediction errors of the real-time performance ability value labels of all early warning systems in the fault early warning task as the training target. Train the machine learning model until the sum of the prediction errors reaches convergence and then stop the model training. Determine the real-time performance ability value of the early warning system in the fault early warning task according to the model output result, where the machine learning model is a polynomial regression model.

[0019] Preferably, in the classification module, according to the evaluation results, the fault timeliness of the early warning system in the fault early warning task is divided into real-time early warning and non-real-time early warning, specifically as follows:

[0020] Compare the obtained real-time performance ability value of the early warning system in the fault early warning task with the reference threshold of the real-time performance ability value set under the normal state according to historical data. If the real-time performance ability value of the early warning system in the fault early warning task is greater than or equal to the reference threshold of the real-time performance ability value, it indicates that the real-time performance of the early warning system in the fault early warning task is high. At this time, a high real-time signal is generated, and the fault timeliness of the early warning system in the fault early warning task is divided into real-time early warning; if the real-time performance ability value of the early warning system in the fault early warning task is less than the reference threshold of the real-time performance ability value, it indicates that the real-time performance of the early warning system in the fault early warning task is low. At this time, a low real-time signal is generated, and the fault timeliness of the early warning system in the fault early warning task is divided into non-real-time early warning.

[0021] Preferably, in the decision optimization module, for non-real-time early warning, deeply analyze the abnormal degree of the real-time performance ability of the early warning system in the fault early warning task within a fixed time period, and make decision optimization according to the analysis results, specifically as follows:

[0022] For non-real-time early warning, that is, the real-time performance ability value of the early warning system generated within a fixed time period is less than the reference threshold of the real-time performance ability value. Collect the real-time performance ability values of the early warning system generated within the subsequent fixed time period that are less than the reference threshold of the real-time performance ability value, and establish a corresponding data set. Calculate the mean and standard deviation of the data set, analyze it, and make decision optimization according to the analysis results to improve the accuracy of fault early warning.

[0023] Preferably, if the mean value of the real-time performance ability values in the data set is greater than or equal to the reference threshold of the mean value of the real-time performance ability values, and the standard deviation of the real-time performance ability values is less than the reference threshold of the standard deviation of the real-time performance ability values, it indicates that the overall real-time performance of the early warning system is good and the fluctuation is small. At this time, the system operation should be continuously monitored to ensure that the real-time performance remains stable;

[0024] If the mean value of the real-time performance ability values is greater than or equal to the reference threshold of the mean value of the real-time performance ability values, and the standard deviation of the real-time performance ability values is greater than or equal to the reference threshold of the standard deviation of the real-time performance ability values, the overall real-time performance of the early warning system is good but the fluctuation is large. At this time, high-fluctuation periods or sudden tasks should be identified, and the system load should be dynamically optimized;

[0025] If the mean value of the real-time performance ability is less than the reference threshold of the mean value of the real-time performance ability, and the standard deviation of the real-time performance ability is greater than or equal to the reference threshold of the standard deviation of the real-time performance ability, the real-time performance of the early warning system is insufficient and the fluctuation is severe. At this time, the system performance bottleneck should be checked and the task scheduling strategy should be optimized;

[0026] If the mean value of the real-time performance ability is less than the reference threshold of the mean value of the real-time performance ability, and the standard deviation of the real-time performance ability is less than the reference threshold of the standard deviation of the real-time performance ability, the real-time performance of the early warning system is insufficient but the performance is stable. At this time, all factors affecting real-time performance should be comprehensively optimized to improve the basic processing ability of the system and restore real-time performance.

[0027] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0028] 1. The present invention can effectively solve problems such as early warning delay, data packet loss, and decision-making errors caused by network limitations and insufficient edge computing capabilities in the prior art. The data transmission monitoring module records and analyzes the transmission delay of data packets, the computing resource monitoring module real-time collects and evaluates the CPU occupancy rate fluctuation of the edge computing device, and the real-time performance evaluation module comprehensively analyzes the delay drift index and the CPU fluctuation index using a machine learning model to accurately evaluate the real-time performance ability of the system in the fault early warning task. The classification module divides the real-time performance of the early warning task into real-time early warning and non-real-time early warning, and combines the decision optimization module to deeply analyze and optimize the non-real-time early warning, effectively improving the accuracy and response efficiency of the fault early warning.

[0029] 2. Through multi-index monitoring and analysis, the system of the present invention can dynamically evaluate and classify the real-time performance of early warnings, provide flexible coping strategies, and avoid misjudgment and system performance degradation caused by delays and fluctuations. Secondly, the decision optimization module statistically analyzes the real-time performance ability of non-real-time early warnings, and puts forward targeted optimization suggestions according to the characteristics of the mean value and the standard deviation of fluctuations, ensuring the stability and reliability of the early warning system. At the same time, the present invention uses a machine learning model to improve the evaluation accuracy and combines a dynamic adjustment strategy, greatly improving the real-time performance and accuracy of the fault early warning task, and meeting the requirements of efficient fault management under complex shipboard environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0031] Figure 1 It is the system module diagram of the present invention. Detailed Implementation Modes

[0032] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0033] For the embodiments, please refer to Figure 1 As shown, the intelligent fault warning system of the ship environmental monitoring device in this embodiment includes a data transmission monitoring module, a computing resource monitoring module, a real-time evaluation module, a classification module, and a decision optimization module;

[0034] Data transmission monitoring module: Record the sending time and receiving time of sensor data from the acquisition end to the edge computing unit through the timestamp technology, calculate the transmission delay of a single data packet, and perform several measurements under different network conditions to judge the overall interference degree of the data packet transmission delay;

[0035] Computing resource monitoring module: When the overall interference degree of the data packet transmission delay is high, use the system monitoring tool to collect the CPU occupancy rate of the edge computing device in real time, and analyze the fluctuation range of the CPU occupancy rate collected in real time to judge the abnormal degree of the CPU occupancy rate of the edge computing device;

[0036] Real-time evaluation module: In the fault warning task, comprehensively analyze the overall interference degree of the data packet transmission delay and the abnormal degree of the CPU occupancy rate of the edge computing device to evaluate the real-time performance ability of the warning system in the fault warning task;

[0037] Classification module: According to the evaluation results, divide the real-time performance ability of the warning system in the fault warning task into real-time warning and non-real-time warning;

[0038] Decision optimization module: For non-real-time warnings, deeply analyze the abnormal degree of the real-time performance ability of the warning system in the fault warning task within a fixed time period, and make decision optimization according to the analysis results to improve the accuracy of fault warning.

[0039] In the data transmission monitoring module, at the sensor data acquisition end, a high-precision timestamp Tsend (the unit can be milliseconds or microseconds) is generated before each data packet is sent. The timestamp can be obtained through the system clock or a synchronous clock protocol (such as NTP, Network Time Protocol) to ensure the time consistency between the acquisition end and the receiving end. When the edge computing unit receives a data packet, it immediately records the arrival timestamp Treceive. Ensure the clock synchronization of the receiving end device to eliminate the error introduced by time differences. For each data packet, calculate its transmission delay: Delay = Treceive - Tsend; the delay value directly reflects the transmission time of the data packet in the network.

[0040] After analyzing the overall change trend of the data delay, a data delay drift index is generated to judge the overall interference degree of the data packet transmission delay. The method for obtaining the data delay drift index is as follows:

[0041] Under different network conditions (such as signal strength changes, bandwidth fluctuations), continuously measure the transmission delays of multiple data packets for multiple times; obtain the delay sample value set: {Delay1, Delay2,..., Delay n}, calculate the mean μ of the delay samples, and the expression is: Calculate the standard deviation σ of the delay, which reflects the instability of the delay: In the formula, Delay i is the transmission delay of the i-th data packet, and n is the number of samples; calculate the distribution symmetry S of the delay data, and the expression is: σ 3 represents normalizing the deviation value to eliminate the influence of the data magnitude on the result; select the skewness coefficient S baseline of the delay distribution in the reference time period as the reference value for judging the current delay drift, and calculate the data delay drift index. The expression is: HK = |S - S baseline |; in the formula, HK is the data delay drift index.

[0042] When the data delay drift index is larger, it indicates that the distribution of the data packet transmission delay deviates from the reference state, and the asymmetry of the delay data increases significantly. This usually means that the instability of the network performance intensifies, such as the frequent occurrence of delay peaks, or the transmission time of some data packets deviates greatly from the average value. In the case of a large drift index, the overall interference degree is higher, which may lead to a decrease in the transmission reliability of real-time data, and further affect the response speed and accuracy of the fault warning system.

[0043] On the contrary, when the data delay drift index is smaller, it indicates that the distribution of packet transmission delay is close to the reference state, with less delay fluctuation and more stable network performance. This shows that the degree of interference to data transmission is relatively low, and the delay change tends to be within a predictable range. In the case of a small drift index, the system can maintain high real-time performance, and the transmission and processing of warning signals are more efficient, providing stable support for intelligent fault warning.

[0044] Computing resource monitoring module: When the overall degree of interference to packet transmission delay is high, use the system monitoring tool to collect the CPU occupancy rate of the edge computing device in real time, and analyze the fluctuation amplitude of the CPU occupancy rate collected in real time to judge the abnormal degree of the CPU occupancy rate of the edge computing device.

[0045] Use system monitoring tools, including: Common tools: In the Linux environment: commands such as top, htop, mpstat, sar, vmstat, etc. Special monitoring tools: Prometheus, Zabbix or built-in monitoring API. Programming interface: Obtain the CPU occupancy rate through the psutil library in Python.

[0046] Set a reasonable collection interval, for example, collect once per second, to ensure sufficient accuracy and real-time performance. Record the CPU occupancy rate value for each collection to form time series data: {CPU t1 , CPU t2 ,..., CPU tn}; where CPU ti is the CPU occupancy rate for the i-th collection (expressed as a percentage). Store the collected CPU occupancy rate data in a local log file or a remote database for subsequent analysis and monitoring of trends.

[0047] After analyzing the fluctuation amplitude of the CPU occupancy rate collected in real time, generate a CPU occupancy rate fluctuation index to judge the abnormal degree of the CPU occupancy rate of the edge computing device. The method for obtaining the CPU occupancy rate fluctuation index is as follows:

[0048] Record the CPU occupancy rate data in real time to form a time series: {CPU t}, t = 1, 2,..., N; N is the total number of sampling points. Remove the mean component from the data to eliminate the interference of the DC component on the frequency domain analysis: CPU t ′ = CPU t - μe, CPU t ′ is the processed CPU occupancy rate, and μe is the mean of the CPU occupancy rate data in the time series. Use a Hamming or Hanning window to smooth the data to reduce spectral leakage.

[0049] For the time series {CPUt Apply the fast Fourier transform to obtain the complex signal in the frequency domain: where F(f k ) is the complex value of the frequency-domain signal at frequency f k ; Calculate the power spectral density P(f k ) of the frequency-domain signal, and the expression is: P(f k ) = |F(f k )| 2 ; Set the frequency threshold f threshold , and regard the part higher than the frequency, that is, the high-frequency component, as the fluctuation component: f threshold is usually selected according to actual needs. For example, 0.1 Hz represents the fluctuation within every 10 seconds. Calculate the total energy E high of the high-frequency component, and the expression is: Calculate the total energy E total , and the expression is: Calculate the high-frequency energy ratio as the CPU occupancy fluctuation index, and the expression is: GH is the CPU occupancy fluctuation index.

[0050] When the CPU occupancy fluctuation index is larger, it indicates that the proportion of high-frequency fluctuation components in CPU usage is higher, and the abnormal degree of the CPU occupancy of the edge computing device is more serious. This usually means that there is significant instability in CPU usage, which may be caused by sudden tasks, resource competition, or unbalanced task scheduling. In the case of a large fluctuation index, the CPU load shows drastic changes, which may lead to delays or failures of computing tasks, affecting the real-time performance and reliability of the system.

[0051] On the contrary, when the CPU occupancy fluctuation index is smaller, it indicates that there are fewer high-frequency fluctuations in CPU usage, the CPU occupancy of the edge computing device is relatively stable, and the abnormal degree is lower. This shows that the task execution distribution of the system is more balanced, the load is relatively stable, and it is not easy to cause performance bottlenecks or resource exhaustion due to short-term fluctuations. In the case of a small fluctuation index, the device can complete real-time computing tasks more efficiently, ensuring the stability and accuracy of the fault warning system.

[0052] Real-time evaluation module: In the fault warning task, comprehensively analyze the overall interference degree of the data packet transmission delay and the abnormal degree of the CPU occupancy of the edge computing device to evaluate the real-time performance of the warning system in the fault warning task.

[0053] Convert the data delay drift index and the CPU occupancy rate fluctuation index into a comprehensive feature vector, and use the comprehensive feature vector as the input of the machine learning model. The machine learning model takes the real-time performance ability value label of the early warning system in the fault early warning task predicted by each group of comprehensive feature vectors as the prediction target, and takes minimizing the sum of the prediction errors of the real-time performance ability value labels of all early warning systems in the fault early warning task as the training target to train the machine learning model until the sum of the prediction errors reaches convergence and then stop the model training. Determine the real-time performance ability value of the early warning system in the fault early warning task according to the model output result, where the machine learning model is a polynomial regression model.

[0054] The method for obtaining the real-time performance ability value of the early warning system in the fault early warning task is as follows: Obtain the corresponding function expression from the comprehensive feature vector training data of the trained machine learning model: LR = F(HK, GH); where F is the output function of the model, HK is the data delay drift index, GH is the CPU occupancy rate fluctuation index, and LR is the real-time performance ability value of the early warning system in the fault early warning task.

[0055] Classification module: According to the evaluation result, classify the fault real-time performance of the early warning system in the fault early warning task into real-time early warning and non-real-time early warning.

[0056] Compare the obtained real-time performance ability value of the early warning system in the fault early warning task with the reference threshold of the real-time performance ability value set under the normal state according to historical data. If the real-time performance ability value of the early warning system in the fault early warning task is greater than or equal to the reference threshold of the real-time performance ability value, it indicates that the real-time performance of the early warning system in the fault early warning task is high. At this time, generate a high real-time signal and classify the fault real-time performance of the early warning system in the fault early warning task into real-time early warning; if the real-time performance ability value of the early warning system in the fault early warning task is less than the reference threshold of the real-time performance ability value, it indicates that the real-time performance of the early warning system in the fault early warning task is low. At this time, generate a low real-time signal and classify the fault real-time performance of the early warning system in the fault early warning task into non-real-time early warning.

[0057] For real-time early warning, it indicates that the system can respond and process the fault early warning task in a timely manner in the current state without additional optimization measures. In this case, it is necessary to ensure that the system maintains a high real-time operating state, continuously monitor the data delay drift index and the CPU occupancy rate fluctuation index to prevent performance degradation, and record the key parameters of the high real-time operation as a reference for subsequent performance evaluation and optimization.

[0058] Decision Optimization Module: For non-real-time warnings, deeply analyze the abnormality degree of the real-time performance ability of the warning system in the fault warning task within a fixed time period, and make decision optimization according to the analysis results to improve the accuracy of fault warning.

[0059] For non-real-time warnings, that is, the real-time performance ability value of the warning system generated within a fixed time period in the fault warning task is less than the reference threshold of the real-time performance ability value. Collect the real-time performance ability values of the warning system generated within the subsequent fixed time period that are less than the reference threshold of the real-time performance ability value, and establish a corresponding data set. Calculate the mean and standard deviation of the data set, analyze it, and make decision optimization according to the analysis results to improve the accuracy of fault warning.

[0060] If the mean value of the real-time performance ability values in the data set is greater than or equal to the reference threshold of the mean value of the real-time performance ability values, and the standard deviation of the real-time performance ability values is less than the reference threshold of the standard deviation of the real-time performance ability values, it indicates that the overall real-time performance of the warning system is good and the fluctuation is small, and the system has acceptable stability in the non-real-time warning state. At this time, the system operation should be continuously monitored, and the scheduling of non-critical tasks should be appropriately optimized to ensure that the real-time performance remains stable.

[0061] If the mean value of the real-time performance ability values is greater than or equal to the reference threshold of the mean value of the real-time performance ability values, and the standard deviation of the real-time performance ability values is greater than or equal to the reference threshold of the standard deviation of the real-time performance ability values, the overall real-time performance of the warning system is good but the fluctuation is large, and there is a risk of short-term performance degradation. At this time, high-fluctuation periods or sudden tasks should be identified, and the system load should be dynamically optimized, such as adjusting the priority of critical tasks or balancing the calculation resource allocation.

[0062] If the mean value of the real-time performance ability values is less than the reference threshold of the mean value of the real-time performance ability values, and the standard deviation of the real-time performance ability values is greater than or equal to the reference threshold of the standard deviation of the real-time performance ability values, the real-time performance of the warning system is insufficient and the fluctuation is severe, and the performance ability is unstable. At this time, the system performance bottlenecks (such as network latency, calculation load) should be checked, the task scheduling strategy should be optimized, and if necessary, backup calculation resources should be introduced or the hardware performance should be improved.

[0063] If the mean value of the real-time performance ability values is less than the reference threshold of the mean value of the real-time performance ability values, and the standard deviation of the real-time performance ability values is less than the reference threshold of the standard deviation of the real-time performance ability values, the real-time performance of the warning system is insufficient but the performance is relatively stable, indicating that the system performance is limited and there is a large optimization space. At this time, the main factors affecting real-time performance (such as latency, algorithm efficiency) should be comprehensively optimized to improve the basic processing ability of the system and restore the real-time performance.

[0064] In this embodiment, there is a data transmission monitoring module that records the sending and receiving times of sensor data through timestamp technology, calculates the transmission delay, and determines the degree of network interference; a computing resource monitoring module that, when the degree of transmission delay interference is relatively high, monitors the CPU occupancy rate of the edge computing device in real time and analyzes its fluctuation range; a real-time performance evaluation module that evaluates the real-time performance of the system based on the comprehensive data transmission delay and CPU occupancy anomaly degree; a classification module that divides the fault warnings into real-time warnings and non-real-time warnings according to the real-time performance evaluation results; and a decision-making optimization module that deeply analyzes the abnormal manifestations of non-real-time warnings and optimizes the system operation according to the results to improve the warning accuracy. These modules jointly ensure the real-time performance and reliability of the warning system.

[0065] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula that is closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0066] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that the computer can access, or a data storage device such as a server or data center that includes one or more collections of available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0067] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present application, and all of them should be covered by the protection scope of the present application.

Claims

1. Intelligent fault warning system for ship environment monitoring equipment, characterized by: It includes data transmission monitoring module, computing resource monitoring module, real-time evaluation module, classification module and decision optimization module; Data transmission monitoring module: records the sending and receiving time of sensor data from the acquisition end to the edge computing unit through timestamp technology, calculates the transmission delay of a single data packet, and performs several measurements under different network conditions to determine the overall interference degree of data packet transmission delay; Computing resource monitoring module: When the overall interference level of data packet transmission delay is high, the system monitoring tool is used to collect the CPU occupancy rate of edge computing devices in real time, and the fluctuation range of the CPU occupancy rate collected in real time is analyzed to determine the abnormality level of the CPU occupancy rate of edge computing devices; Real-time evaluation module: In the fault warning task, the overall interference degree of data packet transmission delay and the abnormal degree of CPU occupancy of edge computing devices are comprehensively analyzed to evaluate the real-time performance capability of the warning system in the fault warning task; Classification module: Based on the evaluation results, the early warning system's real-time performance in the fault warning task is divided into real-time warning and non-real-time warning; Decision optimization module: For non-real-time warnings, the abnormal degree of the real-time performance of the warning system in the fault warning task within a fixed time period is deeply analyzed, and decision optimization is performed based on the analysis results, specifically: For non-real-time warnings, that is, the real-time performance capability value of the warning system generated within a fixed time period in the fault warning task is less than the reference threshold of the real-time performance capability value, the real-time performance capability values ​​of the warning system in the fault warning task that are less than the reference threshold of the real-time performance capability value generated within a subsequent fixed time period are collected, and a corresponding data set is established, the mean and standard deviation of the data set are calculated, and after analysis, decision optimization is performed based on the analysis results.

2. The intelligent fault early warning system for ship environment monitoring equipment according to claim 1 is characterized in that: In the data transmission monitoring module, the data delay drift index is generated after analyzing the overall change trend of the data delay. The method for obtaining the data delay drift index is as follows: Continuously measure the transmission delay of multiple data packets; obtain the delay sample value set: {Delay1, Delay2, ..., Delay n }, calculate the mean μ of the delayed samples, the expression is: Calculate the standard deviation σ of the delay: Where Delay i is the transmission delay of the ith data packet, n is the number of samples; the distribution symmetry S of the delay data is calculated, and the expression is: σ 3 Indicates normalizing the deviation value; Select the delay distribution skewness coefficient S of the reference time period baseline , as the reference value for judging the current delay drift, calculate the data delay drift index, the expression is: HK=|SS baseline |; Where HK is the data delay drift index.

3. The intelligent fault warning system for ship environment monitoring equipment according to claim 2 is characterized in that: In the computing resource monitoring module, the fluctuation range of the CPU usage collected in real time is analyzed to generate a CPU usage fluctuation index. The CPU usage fluctuation index is obtained as follows: Record CPU usage data in real time to form a time series: {CPU t }, t=1,2,…,N; N is the total number of sampling points. Remove the mean component of the data to eliminate the interference of the DC component on the frequency domain analysis: CPU t ′=CPU t -μe, CPU t ′ is the CPU occupancy rate after processing, μe is the mean value of the CPU occupancy rate data in the time series; t ′} Apply fast Fourier transform to obtain the complex signal in frequency domain: In the formula, F(f k ) is the frequency domain signal at frequency f k The complex value on the frequency domain signal is calculated as the power spectral density P(f k ), the expression is: P(f k )=|F(f k )| 2 ; Set the frequency threshold f threshold , the part higher than the frequency, that is, the high-frequency component, is regarded as the wave component: calculate the total energy E of the high-frequency component high , the expression is: Calculate the total energy E total , the expression is: Calculate the high-frequency energy ratio as the CPU occupancy fluctuation index. The expression is: GH is the CPU usage fluctuation index.

4. The intelligent fault warning system for ship environment monitoring equipment according to claim 3 is characterized by: In the real-time evaluation module, the overall interference degree of data packet transmission delay and the abnormal degree of CPU occupancy of edge computing devices are comprehensively analyzed to evaluate the real-time performance of the early warning system in the fault early warning task, specifically: The data delay drift index and the CPU occupancy rate fluctuation index are converted into comprehensive feature vectors, and the comprehensive feature vectors are used as the input of the machine learning model. The machine learning model uses each group of comprehensive feature vectors to predict the real-time performance capability value label of the early warning system in the fault warning task as the prediction target, and takes minimizing the sum of the prediction errors of the real-time performance capability value labels of all early warning systems in the fault warning task as the training target. The machine learning model is trained until the sum of the prediction errors converges and the model training is stopped. The real-time performance capability value of the early warning system in the fault warning task is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

5. The intelligent fault warning system for ship environment monitoring equipment according to claim 4 is characterized in that: In the classification module, according to the evaluation results, the fault real-time nature of the early warning system in the fault warning task is divided into real-time warning and non-real-time warning, specifically: The obtained real-time performance capability value of the early warning system in the fault early warning task is compared with the reference threshold of the real-time performance capability value set under the normal state according to the historical data. If the real-time performance capability value of the early warning system in the fault early warning task is greater than or equal to the reference threshold of the real-time performance capability value, it means that the real-time performance of the early warning system in the fault early warning task is high. At this time, a high real-time signal is generated, and the fault real-time performance of the early warning system in the fault early warning task is classified as a real-time warning; If the real-time performance capability value of the early warning system in the fault warning task is less than the reference threshold of the real-time performance capability value, it means that the real-time performance of the early warning system in the fault warning task is low. At this time, a low real-time signal is generated, and the fault real-time performance of the early warning system in the fault warning task is classified as a non-real-time warning.

6. The intelligent fault warning system for ship environment monitoring equipment according to claim 1 is characterized in that: If the mean of the real-time performance capability values ​​in the data set is greater than or equal to the reference threshold of the mean of the real-time performance capability values, and the standard deviation of the real-time performance capability values ​​is less than the reference threshold of the standard deviation of the real-time performance capability values, it means that the real-time performance of the early warning system is generally good with small fluctuations. At this time, the system operation should be continuously monitored to ensure that the real-time performance remains stable; If the mean of the real-time performance capability value is greater than or equal to the reference threshold of the mean of the real-time performance capability value, and the standard deviation of the real-time performance capability value is greater than or equal to the reference threshold of the standard deviation of the real-time performance capability value, the overall real-time performance of the early warning system is good but the fluctuation is large. At this time, the high fluctuation period or sudden task should be identified to dynamically optimize the system load; If the mean of the real-time performance capability value is less than the reference threshold of the mean of the real-time performance capability value, and the standard deviation of the real-time performance capability value is greater than or equal to the reference threshold of the standard deviation of the real-time performance capability value, the real-time performance of the early warning system is insufficient and fluctuates violently. At this time, the system performance bottleneck should be checked and the task scheduling strategy should be optimized; If the mean of the real-time performance capability value is less than the reference threshold of the mean of the real-time performance capability value, and the standard deviation of the real-time performance capability value is less than the reference threshold of the standard deviation of the real-time performance capability value, the early warning system is not real-time enough but performs stably. At this time, comprehensive optimization should be carried out on the factors affecting the real-time performance to restore the real-time performance.

Images