Big Data-Based IoT Security Management Methods and Systems

Abstract

This invention discloses a big data-based IoT security management method and system, belonging to the field of IoT security management technology. Through a big data-based health prediction model, the system can collect real-time operational data of IoT devices and train a regression model to predict the remaining useful life (RUL) and time to failure (TTF) of the devices. Unlike traditional rule-based monitoring methods, this system provides deep learning analysis based on historical data to predict potential device failure times in advance, helping managers to perform timely preventative maintenance. This not only solves the lag problem in device health prediction but also improves device utilization efficiency and reliability, reducing the impact of device failures. Through a resource allocation and optimization unit, the system rationally allocates maintenance personnel, tools, and spare parts according to maintenance strategy decisions and assigns appropriate maintenance resources to different devices.

Description

Technical Field

[0001] This invention relates to the field of IoT security management technology, specifically to IoT security management methods and systems based on big data. Background Technology

[0002] The Internet of Things (IoT) connects various devices and objects through sensors, networks, and data processing technologies to achieve intelligent sensing and control. Within this broad field, IoT security management is a crucial subfield, dedicated to ensuring the security and stability of devices and networks within IoT systems, preventing data breaches, equipment malfunctions, and other potential risks.

[0003] Despite the widespread adoption of IoT technology and security management systems in recent years, current IoT security management systems still face numerous challenges, particularly in device health monitoring and fault prediction. Most traditional monitoring systems rely on rule-based judgments and cannot predict device health status in real time based on historical data.

[0004] This lack of effective prediction and monitoring often stems from the fact that most IoT devices rely on simple alarm systems or are monitored solely based on basic device status parameters, lacking deep learning and analysis based on historical data. This limitation makes it difficult to detect potential faults in real-world applications. To avoid these problems, intelligent predictive models based on big data offer a solution. Through real-time health prediction and data analysis, they can not only detect faults promptly but also enable preventative maintenance, thereby reducing unnecessary downtime and malfunctions. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the present invention provides a method and system for Internet of Things security management based on big data, which solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: a big data-based Internet of Things security management and control system, including a data acquisition module, a data preprocessing module, a feature engineering module, a health prediction model training module, a fault prediction and risk assessment module, and a decision support and optimization module;

[0007] The data acquisition module is responsible for collecting the operational data of IoT devices and forming a data operation set W;

[0008] The data preprocessing module cleans and standardizes the data runtime set W to obtain the data standard set WB.

[0009] The feature engineering module extracts features from the data standard set WB, including frequency variation Δf, temperature fluctuation ΔT, and power consumption variation ΔE, and forms a time series feature group F.

[0010] The health prediction model training module is based on the time series feature group F. It uses a regression model to train the prediction model and obtain the prediction result Y, which includes the remaining lifetime (RUL) and time to failure (TTF) of the IoT device.

[0011] The fault prediction and risk assessment module uses the remaining lifetime (RUL) and time to failure (TTF) to predict the status of IoT devices and assess fault risks, obtain the risk assessment value (RS) and failure probability (Pfa) of IoT devices, and generate adjustment suggestions.

[0012] The decision support and optimization module provides decision support based on the adjustment suggestions of IoT devices and adjusts resource allocation.

[0013] Preferably, the data acquisition module includes a sensor data acquisition unit and a device data acquisition unit;

[0014] The sensor data acquisition unit collects the temperature T, battery voltage V, and power consumption E of the IoT device through sensors; the temperature T is acquired through a temperature sensor, the battery voltage V is acquired through a voltage sensor, and the power consumption E is acquired through a power sensor.

[0015] The device data acquisition unit collects the usage frequency f and runtime L of the IoT device through the usage log of the working device; it fits the data collected by the sensor data acquisition unit and the data collected by the device data acquisition unit to obtain the data runtime set W.

[0016] Preferably, the data preprocessing module includes a data cleaning unit and a data standardization unit;

[0017] The data cleaning unit cleans the data run set W, including processing missing values, outliers and duplicate data, and obtains the cleaned data set WC.

[0018] Missing values ​​are handled by filling in missing values ​​using interpolation methods;

[0019] Outliers are detected in the data using the IQR method, and then corrected, removed, and marked.

[0020] The deduplication process checks for duplicate records in the data run set W, including removing duplicate data with the same timestamp.

[0021] The data standardization unit performs Z-score standardization on the cleaned data set WC, transforming data of different scales into the same standard range to obtain the data standard set WB.

[0022] Preferably, the feature engineering module includes a temporal feature extraction unit and a health index construction unit;

[0023] The timing feature extraction unit extracts the timing features of the device from the data standard set WB, including usage frequency change Δf, temperature fluctuation ΔT and power consumption change ΔE, forming a timing feature group F;

[0024] The frequency change Δf is obtained by the difference between the frequency of use of the IoT device at time t and the frequency of use of the IoT device at time t-1.

[0025] The temperature fluctuation ΔT is obtained by calculating the average temperature of the IoT devices;

[0026] The power consumption change ΔE is obtained by the difference between the power consumption of the IoT device at time t and the power consumption of the IoT device at time t-1.

[0027] Preferably, the health index construction unit calculates and obtains a comprehensive health index HI based on the time-series feature group F, performs standardization processing, and determines the health status of IoT devices through the health index HI.

[0028] The health index HI is obtained by weighted summation of usage frequency change Δf, temperature fluctuation ΔT, and power consumption change ΔE.

[0029] The health status of the IoT device is obtained through the following methods:

[0030] When 0.5 ≤ Health Index HI < 1, it indicates that the IoT device is in normal health condition;

[0031] When 0 < Health Index HI < 0.5, it indicates that the health status of the IoT device is abnormal.

[0032] Preferably, the health prediction model training module includes a feature selection unit and a training and prediction unit;

[0033] The feature selection unit uses the Pearson correlation coefficient R to select features in the time-series feature group F that are related to remaining lifetime RUL and time to failure TTF.

[0034] The training and prediction unit trains the device health prediction model using a regression model and predicts the remaining lifetime (RUL) and time to failure (TTF) of the IoT device.

[0035] Specifically, the mean squared error (MSE) is used as the loss function to train the regression model;

[0036] The remaining lifetime RUL is obtained by calculating the features in the time-series feature group F;

[0037] The time-of-failure (TTF) is obtained by calculating the features in the time-series feature group F.

[0038] Preferably, the fault prediction and risk assessment module includes a risk assessment and status prediction unit and a maintenance suggestion generation unit;

[0039] The risk assessment and status prediction unit calculates the risk assessment value RS of the IoT device based on the remaining lifetime RUL and time to failure TTF, and compares it with the preset risk threshold TRS to determine the risk status of the IoT device.

[0040] The risk assessment value RS is obtained by calculating the remaining useful life (RUL) and the time to failure (TTF).

[0041] The risk status of the IoT device is obtained through matching in the following way:

[0042] When the risk assessment value RS < the risk threshold TRS, it indicates that the risk status of the IoT device is normal and the IoT device is in a healthy state.

[0043] When the risk assessment value RS ≥ the risk threshold TRS, it indicates that the risk status of the IoT device is abnormal and the IoT device is in an unhealthy state.

[0044] Based on the health index HI, remaining lifetime RUL, and time to failure TTF of the IoT device, the failure probability Pfa of the IoT device is calculated.

[0045] The failure probability Pfa is obtained by weighting the health index HI, remaining lifetime RUL, and time to failure TTF. The shorter the remaining lifetime RUL of an IoT device, the shorter the potential time to failure TTF, the lower the health index HI, and the higher the failure probability. Conversely, the longer the remaining lifetime RUL, the longer the potential time to failure TTF, the higher the health index HI, and the lower the failure probability.

[0046] Preferably, the maintenance recommendation generation unit generates recommendations for preventative maintenance and replacement of IoT devices based on the risk assessment value RS and failure probability Pfa of the IoT devices;

[0047] When an IoT device is in an abnormal state, maintenance and replacement adjustment suggestions are generated and the administrator is notified via SMS and email.

[0048] The adjustment recommendations are obtained through matching in the following ways:

[0049] When the risk assessment value RS≥0.8 or the failure probability Pfa≥0.7, it indicates that the IoT device is in the first risk state, the first adjustment suggestion is initiated, and preventive maintenance suggestions are generated.

[0050] When the risk assessment value RS≥0.9 or the failure probability Pfa≥0.8, it indicates that the IoT device is in the second risk state, and the second adjustment recommendation is initiated, which is to replace the IoT device.

[0051] When the risk assessment value RS < 0.8 or the failure probability Pfa < 0.7, it indicates that the IoT device is in the third risk state. Activation indicates the third adjustment recommendation. The IoT device is operating normally and does not require maintenance.

[0052] Preferably, the decision support and optimization module includes a maintenance strategy decision-making unit and a resource allocation and optimization unit;

[0053] The maintenance strategy decision-making unit provides maintenance strategy decisions based on the obtained adjustment suggestions, specifically including:

[0054] When IoT devices are in a state of first risk, they require planned maintenance, and a maintenance plan should be generated for each IoT device.

[0055] When an IoT device is in a second-risk state, IoT device maintenance and replacement are required.

[0056] The resource allocation and optimization unit allocates appropriate maintenance resources, including maintenance personnel, tools, and spare parts, to different IoT devices based on the first and second risk states determined by the maintenance strategy. It also schedules maintenance windows and maintenance cycles for IoT devices based on the risk assessment value RS and the failure probability Pfa, and rationally allocates maintenance personnel and resources.

[0057] The IoT security management method based on big data includes the following steps:

[0058] Step 1: The data acquisition module is responsible for collecting the operational data of IoT devices and forming a data operation set W;

[0059] Step 2: The data preprocessing module cleans and standardizes the data run set W to obtain the data standard set WB;

[0060] Step 3: The feature engineering module extracts features from the data standard set WB, including frequency variation Δf, temperature fluctuation ΔT, and power consumption variation ΔE, and forms a time series feature group F;

[0061] Step 4: The health prediction model training module is based on the time series feature group F. It uses a regression model to train the prediction model and obtain the prediction result Y, including the remaining lifetime (RUL) and time to failure (TTF) of the IoT device.

[0062] Step 5: The Fault Prediction and Risk Assessment module uses Remaining Life (RUL) and Time to Fail (TTF) to predict the status of IoT devices and assess fault risks, obtain the risk assessment value (RS) and failure probability (Pfa) of IoT devices, and generate adjustment suggestions.

[0063] Step Six: The Decision Support and Optimization module provides decision support based on the adjustment suggestions from IoT devices and adjusts resource allocation.

[0064] This invention provides a method and system for IoT security management based on big data, which has the following beneficial effects:

[0065] (1) During system operation, the system can collect real-time operating data of IoT devices through a big data-based health prediction model and use regression models for training to predict the remaining useful life (RUL) and time to failure (TTF) of the devices. Unlike traditional rule-based monitoring methods, this system can provide deep learning analysis based on historical data to predict the possible failure time of devices in advance, helping managers to carry out timely preventive maintenance. This not only solves the problem of lag in device health prediction, but also improves the efficiency and reliability of device use and reduces the impact of device failure.

[0066] Through its fault prediction and risk assessment module, the system can predict the health status of equipment by combining its remaining useful life (RUL) and time to failure (TTF), and calculate the risk assessment value (RS) and probability of failure (Pfa). This function addresses the lack of dynamic risk assessment and early warning in traditional monitoring systems.

[0067] (2) The time-series feature extraction unit processes the data standard set WB to extract the time-series features of equipment usage frequency change Δf, temperature fluctuation ΔT, and power consumption change ΔE, forming a time-series feature group F = {Δf, ΔT, ΔE}. These time-series features can reflect the changes in the equipment's state during operation, especially capturing subtle fluctuations in equipment health, thereby helping to predict the equipment's remaining useful life (RUL) and time to failure (TTF). Compared with traditional static features, time-series features better reflect the trends and potential failure risks of equipment in long-term operation, improving the accuracy and timeliness of failure prediction. The health index construction unit comprehensively analyzes the time-series features of equipment usage frequency change Δf, temperature fluctuation ΔT, and power consumption change ΔE to construct the equipment health index HI, which serves as a quantitative indicator of equipment health status. This health index can reflect the equipment's health status in real time during operation and provide basic data for subsequent failure prediction and risk assessment. The introduction of the health index solves the problem that traditional monitoring systems cannot comprehensively assess equipment health status, improving the system's sensitivity and predictive ability to potential equipment failures.

[0068] (3) The system automatically generates corresponding adjustment suggestions based on the risk assessment value (RS) and failure probability (Pfa), and promptly notifies the administrator via SMS and email. This reduces manual intervention and improves the system's response speed and efficiency. Simultaneously, timely notifications help administrators make quick decisions, avoiding equipment failures and production losses due to information delays. Through the resource allocation and optimization unit, the system rationally allocates maintenance personnel, tools, and spare parts according to maintenance strategies, and assigns appropriate maintenance resources to different equipment. At the same time, the system schedules maintenance time windows and equipment maintenance cycles based on the equipment's risk assessment value and failure probability. This ensures efficient use of maintenance resources, avoids resource waste, and ensures that equipment can be repaired or replaced in the shortest possible time. Compared with traditional resource allocation methods, this improvement offers greater flexibility and response speed, enhancing overall equipment management efficiency.

[0069] (4) The Health Index (HI) is obtained by weighted summation of usage frequency change (Δf), temperature fluctuation (ΔT), and power consumption change (ΔE). This method effectively integrates changes in multiple time-series characteristics, enabling a more comprehensive assessment of the device's health status. Compared to health assessments based on a single indicator, the comprehensive health index can more accurately reflect the overall operating status of the device and provide a more accurate basis for subsequent fault prediction. This improvement effectively solves the problem of single and one-sided assessment of device health in traditional IoT device monitoring systems, making device management more scientific and comprehensive.

[0070] By weighting different features, the system can adjust the degree of influence of each feature according to its actual importance in device health. Attached Figure Description

[0071] Figure 1 This is a schematic diagram of the block diagram of the IoT security management and control system based on big data according to the present invention;

[0072] Figure 2 This is a schematic diagram illustrating the steps of the IoT security management method based on big data according to the present invention. Detailed Implementation

[0073] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0074] Example 1

[0075] This invention provides an IoT security management and control system based on big data. Please refer to [link / reference]. Figure 1 It includes a data acquisition module, a data preprocessing module, a feature engineering module, a health prediction model training module, a fault prediction and risk assessment module, and a decision support and optimization module;

[0076] The data acquisition module is responsible for collecting the operational data of IoT devices and forming a data operation set W;

[0077] The data preprocessing module cleans and standardizes the data runtime set W to obtain the data standard set WB.

[0078] The feature engineering module extracts features from the data standard set WB, including frequency variation Δf, temperature fluctuation ΔT, and power consumption variation ΔE, and forms a time series feature group F.

[0079] The health prediction model training module is based on the time series feature group F. It uses a regression model to train the prediction model and obtain the prediction result Y, which includes the remaining lifetime (RUL) and time to failure (TTF) of the IoT device.

[0080] The fault prediction and risk assessment module uses the remaining lifetime (RUL) and time to failure (TTF) to predict the status of IoT devices and assess fault risks, obtain the risk assessment value (RS) and failure probability (Pfa) of IoT devices, and generate adjustment suggestions.

[0081] The decision support and optimization module provides decision support based on the adjustment suggestions of IoT devices and adjusts resource allocation.

[0082] In this embodiment, through a health prediction model based on big data, the system can collect real-time operational data from IoT devices and train a regression model to predict the device's Remaining Life (RUL) and Time to Failure (TTF). Unlike traditional rule-based monitoring methods, this system provides deep learning analysis based on historical data to predict potential device failure times in advance, helping managers to perform timely preventative maintenance. This not only solves the lag problem in device health prediction but also improves device efficiency and reliability, reducing the impact of device failures.

[0083] Through its fault prediction and risk assessment module, the system can predict the health status of equipment by combining its remaining useful life (RUL) and time to failure (TTF), and calculate the risk assessment value (RS) and probability of failure (Pfa). This function addresses the lack of dynamic risk assessment and early warning in traditional monitoring systems. Based on these assessment results, the system can promptly identify potential faults and safety risks, providing managers with a scientific basis for decision-making, reducing sudden equipment failures, and mitigating the risk of production system interruptions.

[0084] The system's decision support and optimization module provides decision support based on the risk assessment value (RS) and failure probability (Pfa) of IoT devices, and adjusts resource allocation accordingly. Managers can rationally allocate maintenance resources based on the system's intelligent suggestions, thereby achieving optimal equipment maintenance strategies. By combining historical data and intelligent prediction, the system can proactively identify potential equipment failures and provide adjustment suggestions, enabling equipment to be repaired or replaced before failures occur, thus reducing downtime and maintenance costs caused by sudden failures. Furthermore, by optimizing resource allocation, over-repair or unnecessary equipment replacement can be avoided, effectively reducing maintenance costs and improving system operational efficiency.

[0085] Example 2

[0086] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically: the data acquisition module includes a sensor data acquisition unit and a device data acquisition unit;

[0087] The sensor data acquisition unit collects the temperature T, battery voltage V, and power consumption E of the IoT device through sensors; the temperature T is acquired through a temperature sensor, the battery voltage V is acquired through a voltage sensor, and the power consumption E is acquired through a power sensor.

[0088] The device data acquisition unit collects the usage frequency f and runtime L of the IoT device through the usage log of the working device; it fits the data collected by the sensor data acquisition unit and the data collected by the device data acquisition unit to obtain the data runtime set W.

[0089] The data preprocessing module includes a data cleaning unit and a data standardization unit;

[0090] The data cleaning unit cleans the data run set W, including processing missing values, outliers and duplicate data, and obtains the cleaned data set WC.

[0091] Missing values ​​are handled by filling in missing values ​​using interpolation methods;

[0092] Outliers are detected in the data using the IQR method, and then corrected, removed, and marked.

[0093] The deduplication process checks for duplicate records in the data run set W, including removing duplicate data with the same timestamp.

[0094] The data standardization unit performs Z-score standardization on the cleaned data set WC, transforming data of different scales into the same standard range to obtain the data standard set WB.

[0095] The data standard set WB is obtained using the following formula:

[0096]

[0097] In the formula, WCd represents the d-th data item in the data cleansing set WC, μd represents the mean of the d-th data item in the data cleansing set WC, and σ represents the standard deviation of the d-th data item in the data cleansing set WC.

[0098] The feature engineering module includes a temporal feature extraction unit and a health index construction unit;

[0099] The timing feature extraction unit extracts the timing features of the device from the data standard set WB, including usage frequency change Δf, temperature fluctuation ΔT and power consumption change ΔE, forming a timing feature group F = {Δf, ΔT, ΔE}.

[0100] The frequency change Δf is obtained by the difference between the frequency of use of the IoT device at time t and the frequency of use of the IoT device at time t-1.

[0101] The temperature fluctuation ΔT is obtained by calculating the average temperature of the IoT devices;

[0102] The temperature fluctuation ΔT is obtained using the following formula:

[0103]

[0104] In the formula, n represents the total number of time points, Tt represents the temperature of the IoT device at time point t, and μT represents the average temperature of the IoT device at time point t.

[0105] The power consumption change ΔE is obtained by the difference between the power consumption of the IoT device at time t and the power consumption of the IoT device at time t-1.

[0106] In this embodiment, the data acquisition module comprehensively collects multi-dimensional data from IoT devices through a sensor data acquisition unit and a device data acquisition unit. The sensor data acquisition unit can acquire the IoT device's temperature T, battery voltage V, and power consumption E in real time, while the device data acquisition unit collects the IoT device's usage frequency f and runtime L through the device's usage logs. This multi-source data acquisition provides more comprehensive and accurate basic data for subsequent health prediction and fault analysis.

[0107] The data preprocessing module ensures improved data quality through data cleaning and standardization units. The cleaning unit employs interpolation to fill in missing values, IQR to detect and correct outliers, and remove duplicate data, significantly improving the quality of the original data and preventing model training failures or errors due to poor data quality. The standardization unit uses Z-score standardization to transform the data into a uniform standard range, allowing for consistent analysis of numerical data from different sources and scales. This not only avoids inconsistencies caused by different data scales but also provides more unified and standardized input data for subsequent feature extraction and model training.

[0108] The time-series feature extraction unit processes the standard data set WB to extract time-series features of equipment usage frequency variation (Δf), temperature fluctuation (ΔT), and power consumption variation (ΔE), forming a time-series feature set F = {Δf, ΔT, ΔE}. These time-series features reflect changes in the equipment's state during operation, especially capturing subtle fluctuations in equipment health, thereby helping to predict the equipment's remaining useful life (RUL) and time to failure (TTF). Compared to traditional static features, time-series features better reflect the trends and potential failure risks of equipment during long-term operation, improving the accuracy and timeliness of failure prediction.

[0109] The calculation of temperature fluctuation ΔT, by analyzing the average temperature of the equipment over a period of time, can reveal whether the equipment is fluctuating within an abnormal temperature range. Temperature fluctuation is a significant factor affecting equipment health, and abnormal temperature fluctuations are often a precursor to equipment failure. The health index construction unit constructs the equipment health index HI as a quantitative indicator of equipment health status by comprehensively analyzing the time-series characteristics of equipment usage frequency change Δf, temperature fluctuation ΔT, and power consumption change ΔE. This health index can reflect the equipment's health status in real time during operation and provides basic data for subsequent fault prediction and risk assessment. The introduction of the health index solves the problem that traditional monitoring systems cannot comprehensively assess equipment health status, improving the system's sensitivity and predictive ability for potential equipment failures.

[0110] Example 3

[0111] This embodiment is an explanation based on Embodiment 2. Please refer to it. Figure 1 Specifically: the health index construction unit calculates and obtains a comprehensive health index HI based on the time-series feature group F, performs standardization processing, and uses the health index HI to determine the health status of IoT devices;

[0112] The health index HI is obtained by weighted summation of usage frequency change Δf, temperature fluctuation ΔT, and power consumption change ΔE.

[0113] The health index HI is obtained using the following formula:

[0114] HI=ω1*Δf+ω2*ΔT+ω3*ΔE;

[0115] In the formula, ω1, ω2 and ω3 represent the preset weight values ​​for the frequency change Δf, temperature fluctuation ΔT and power consumption change ΔE, respectively;

[0116] The health status of the IoT device is obtained through the following methods:

[0117] When 0.5 ≤ Health Index HI < 1, it indicates that the IoT device is in normal health condition;

[0118] When 0 < Health Index HI < 0.5, it indicates that the health status of the IoT device is abnormal.

[0119] The health prediction model training module includes a feature selection unit and a training and prediction unit;

[0120] The feature selection unit uses the Pearson correlation coefficient R to select features in the time-series feature group F that are related to remaining lifetime RUL and time to failure TTF.

[0121] The Pearson correlation coefficient R is obtained using the following formula:

[0122]

[0123] Y represents the target variables including Remaining Life (RUL) and Time to Failure (TTF), R represents the Pearson correlation coefficient, specifically representing the degree of linear correlation between the time series feature group F and the target variable Y, Fi represents the i-th feature in the time series feature group F, Yj represents the j-th target variable, σFi represents the standard deviation of the i-th feature in the time series feature group F, σYj represents the standard deviation of the j-th target variable, and cov(Fi, Yj) represents the covariance between the i-th feature in the time series feature group F and the j-th target variable.

[0124] The training and prediction unit trains the device health prediction model using a regression model and predicts the remaining lifetime (RUL) and time to failure (TTF) of the IoT device.

[0125] Specifically, the mean squared error (MSE) is used as the loss function to train the regression model;

[0126]

[0127] In the formula, yia represents the actual remaining lifetime or failure time of the ia-th sample, Gyia represents the predicted remaining lifetime or failure time of the ia-th sample, and N represents the total number of samples.

[0128] The remaining lifetime RUL is obtained by calculating the features in the time-series feature group F;

[0129]

[0130] In the formula, φ2 represents the weight coefficients obtained during the training of the regression model, which are used to adjust the impact of temperature fluctuations on the remaining lifespan, and e represents a constant;

[0131] The time-of-failure (TTF) is obtained by calculating the features in the time-series feature group F.

[0132]

[0133] In the formula, and This represents the weight coefficients obtained during the training of the regression model, used to adjust the impact of changes in usage frequency on potential failure time.

[0134] In this embodiment, the health index HI is obtained by weighted summation of usage frequency change Δf, temperature fluctuation ΔT, and power consumption change ΔE. This method effectively integrates changes in multiple time-series characteristics, enabling a more comprehensive assessment of the device's health status. Compared to health assessments based on a single indicator, the comprehensive health index can more accurately reflect the overall operating status of the device and provide a more accurate basis for subsequent fault prediction. This improvement effectively solves the problem of single and one-sided assessment of device health in traditional IoT device monitoring systems, making device management more scientific and comprehensive.

[0135] By weighting different features, the system can adjust the degree of influence of each feature according to its actual importance in equipment health. Temperature fluctuations ΔT may have a greater impact on equipment lifespan, so they can be given a higher weight; while frequency changes Δf may be more sensitive to the prediction of time to failure (TTF). This weighting mechanism can provide customized health assessments for different equipment and different operating environments, solving the problems of fixed weights or overly simplistic assessment standards in traditional systems, and improving the accuracy and relevance of the assessment results.

[0136] When the Health Index (HI) is between 0.5 and 1, the equipment is considered to be in normal condition; when the HI is between 0 and 0.5, the equipment's health status is deemed abnormal. This health status assessment is not only intuitive and clear but also provides administrators with a clear basis for decision-making. The health prediction model training module uses the Pearson correlation coefficient (R) for feature selection. Based on the linear correlation between time-series features and target variables, including Remaining Life (RUL) and Time to Failure (TTF), it automatically selects the most relevant features for modeling. This method effectively filters out features with low correlation to the target variables, improving the accuracy and training efficiency of the prediction model.

[0137] The system uses a regression model to calculate the features in the time-series feature group F, accurately predicting the remaining useful life (RUL) and time to failure (TTF) of the equipment. By adjusting the weight coefficients of the features, the regression model can accurately capture the health changes of the equipment during use, especially the impact of temperature fluctuations ΔT on remaining useful life and the impact of usage frequency changes Δf on time to failure. This accurate prediction of time to failure helps managers to implement preventative maintenance, reduce sudden equipment failures, thereby reducing equipment downtime and production losses, and improving production efficiency.

[0138] Example 4

[0139] This embodiment is an explanation based on Embodiment 3. Please refer to it. Figure 1 Specifically: the fault prediction and risk assessment module includes a risk assessment and status prediction unit and a maintenance suggestion generation unit;

[0140] The risk assessment and status prediction unit calculates the risk assessment value RS of the IoT device based on the remaining lifetime RUL and time to failure TTF, and compares it with the preset risk threshold TRS to determine the risk status of the IoT device.

[0141] The risk assessment value RS is obtained by calculating the remaining useful life (RUL) and the time to failure (TTF).

[0142] The risk assessment value RS is obtained using the following formula:

[0143]

[0144] In the formula, RSd represents the risk assessment value of IoT device d, RULd represents the remaining lifetime of IoT device d, RULmax represents the maximum remaining lifetime of IoT device d in a healthy state, TTFd represents the potential failure time of IoT device d, and TTFdmax represents the maximum failure time of IoT device d in a healthy state.

[0145] The risk status of the IoT device is obtained through matching in the following way:

[0146] When the risk assessment value RS < the risk threshold TRS, it indicates that the risk status of the IoT device is normal and the IoT device is in a healthy state.

[0147] When the risk assessment value RS ≥ the risk threshold TRS, it indicates that the risk status of the IoT device is abnormal and the IoT device is in an unhealthy state.

[0148] Based on the health index HI, remaining lifetime RUL, and time to failure TTF of the IoT device, the failure probability Pfa of the IoT device is calculated.

[0149] The failure probability Pfa is obtained by weighting the health index HI, remaining lifetime RUL, and time to failure TTF (the shorter the remaining lifetime RUL of an IoT device, the shorter the potential time to failure TTF, the lower the health index HI, and the higher the failure probability; conversely, the longer the remaining lifetime RUL, the longer the potential time to failure TTF, the higher the health index HI, and the lower the failure probability).

[0150]

[0151] In the formula, Pfad represents the failure probability of IoT device d, exp represents the exponential function, θ1, θ2 and θ3 are the preset weight values ​​of health index HI, remaining lifetime RUL and time to failure TTF respectively, and θ1+θ2+θ3=1;

[0152] The maintenance recommendation generation unit generates recommendations for preventative maintenance and replacement of IoT devices based on the risk assessment value RS and failure probability Pfa of the IoT devices.

[0153] When an IoT device is in an abnormal state, maintenance and replacement adjustment suggestions are generated and the administrator is notified via SMS and email.

[0154] The adjustment recommendations are obtained through matching in the following ways:

[0155] When the risk assessment value RS≥0.8 or the failure probability Pfa≥0.7, it indicates that the IoT device is in the first risk state, the first adjustment suggestion is initiated, and preventive maintenance suggestions are generated.

[0156] When the risk assessment value RS≥0.9 or the failure probability Pfa≥0.8, it indicates that the IoT device is in the second risk state, and the second adjustment recommendation is initiated, which is to replace the IoT device.

[0157] When the risk assessment value RS < 0.8 or the failure probability Pfa < 0.7, it indicates that the IoT device is in the third risk state. Activation indicates the third adjustment recommendation. The IoT device is operating normally and does not require maintenance.

[0158] The decision support and optimization module includes a maintenance strategy decision-making unit and a resource allocation and optimization unit;

[0159] The maintenance strategy decision-making unit provides maintenance strategy decisions based on the obtained adjustment suggestions, specifically including:

[0160] When IoT devices are in a state of first risk, they require planned maintenance, and a maintenance plan should be generated for each IoT device.

[0161] When an IoT device is in a second-risk state, IoT device maintenance and replacement are required.

[0162] The resource allocation and optimization unit allocates appropriate maintenance resources, including maintenance personnel, tools, and spare parts, to different IoT devices based on the first and second risk states determined by the maintenance strategy. It also schedules maintenance windows and maintenance cycles for IoT devices based on the risk assessment value RS and the failure probability Pfa, and rationally allocates maintenance personnel and resources.

[0163] In this embodiment, a comprehensive risk assessment value, RS, is calculated by combining the Remaining Life (RUL) and Time To Failure (TTF) of an IoT device. By comparing this value with a preset risk threshold, TRS, the system can intelligently determine whether the device is in a risky state. Compared to traditional single-risk assessment methods, this multi-dimensional assessment approach more comprehensively and accurately reflects the health status of the device, helping to identify potential faults in a timely manner and prevent major device failures. This improvement significantly enhances the predictive capabilities and timeliness of fault response in IoT device management systems.

[0164] The system calculates the probability of failure (Pfa) of IoT devices based on the Health Index (HI), Remaining Life (RUL), and Time To Failure (TTF), and integrates the impact of these key indicators through a weighted approach. This method makes the assessment of failure probability more flexible and accurate, allowing for dynamic adjustments based on the device's health status and operational data. Especially when the device's health is poor, the increased probability of failure can help managers take timely measures to prevent device failure. Compared to traditional methods that rely solely on static data monitoring, this method provides more accurate predictions for device maintenance and optimizes decision support.

[0165] Based on the risk assessment value (RS) and failure probability (Pfa) of IoT devices, this system classifies devices into multiple risk levels and automatically generates corresponding maintenance or replacement suggestions according to different risk states. If a device is in a high-risk state, the system will suggest preventative maintenance or replacement; if the device is in good health, no maintenance is required. This intelligent risk classification management not only improves the accuracy of fault prediction but also makes resource allocation and maintenance strategies more targeted, ensuring optimized utilization of maintenance resources and avoiding unnecessary maintenance costs.

[0166] Example 5

[0167] For IoT security management methods based on big data, please refer to... Figure 2 Specifically, it includes the following steps:

[0168] Step 1: The data acquisition module is responsible for collecting the operational data of IoT devices and forming a data operation set W;

[0169] Step 2: The data preprocessing module cleans and standardizes the data run set W to obtain the data standard set WB;

[0170] Step 3: The feature engineering module extracts features from the data standard set WB, including frequency variation Δf, temperature fluctuation ΔT, and power consumption variation ΔE, and forms a time series feature group F;

[0171] Step 4: The health prediction model training module is based on the time series feature group F. It uses a regression model to train the prediction model and obtain the prediction result Y, including the remaining lifetime (RUL) and time to failure (TTF) of the IoT device.

[0172] Step 5: The Fault Prediction and Risk Assessment module uses Remaining Life (RUL) and Time to Fail (TTF) to predict the status of IoT devices and assess fault risks, obtain the risk assessment value (RS) and failure probability (Pfa) of IoT devices, and generate adjustment suggestions.

[0173] Step Six: The Decision Support and Optimization module provides decision support based on the adjustment suggestions from IoT devices and adjusts resource allocation.

[0174] In this embodiment, the accuracy and real-time nature of data acquisition are crucial to the effectiveness of subsequent steps, especially the sensor precision and the frequency of device log updates. This module provides real-time feedback to the entire system, and the acquisition of real-time data ensures that the system can make decisions and optimize based on the latest status.

[0175] Z-score standardization helps eliminate potential differences in the units of measurement in the original data, allowing each feature to be processed under the same standard and preventing certain features from dominating the model due to their large values.

[0176] Feature selection and extraction are crucial to the accuracy and effectiveness of a model. Accurate temporal features can provide strong support for health prediction. Extracting temporal features takes into account the time dimension, capturing dynamic changes in equipment operating status and further improving prediction accuracy.

[0177] Training a regression model requires a large amount of historical data to ensure its accuracy and reliability. The model's performance depends on data quality, feature relevance, and the choice of regression method.

[0178] The accuracy of health predictions directly impacts subsequent risk assessments and decision support, particularly in failure prediction and maintenance decisions. Risk assessment and failure prediction provide a basis for equipment maintenance and resource allocation, helping to reduce system downtime and maintenance costs. Failure probability calculations help managers develop appropriate strategies based on risk levels, avoiding unforeseen losses due to equipment failures.

[0179] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A big data-based Internet of Things (IoT) security management system, characterized by: It includes a data acquisition module, a data preprocessing module, a feature engineering module, a health prediction model training module, a fault prediction and risk assessment module, and a decision support and optimization module; The data acquisition module is responsible for collecting the operational data of IoT devices and forming a data operation set W; The data preprocessing module cleans and standardizes the data runtime set W to obtain the data standard set WB. The feature engineering module extracts features from the data standard set WB, including frequency variation Δf, temperature fluctuation ΔT, and power consumption variation ΔE, and forms a time series feature group F. The health prediction model training module is based on the time series feature group F. It uses a regression model to train the prediction model and obtain the prediction result Y, which includes the remaining lifetime (RUL) and time to failure (TTF) of the IoT device. The fault prediction and risk assessment module uses the remaining lifetime (RUL) and time to failure (TTF) to predict the status of IoT devices and assess fault risks, obtain the risk assessment value (RS) and failure probability (Pfa) of IoT devices, and generate adjustment suggestions. The fault prediction and risk assessment module includes a risk assessment and status prediction unit and a maintenance suggestion generation unit; The risk assessment and status prediction unit calculates the risk assessment value RS of the IoT device based on the remaining lifetime RUL and time to failure TTF, and compares it with the preset risk threshold TRS to determine the risk status of the IoT device. The risk assessment value RS is obtained by calculating the remaining useful life (RUL) and the time to failure (TTF). The risk status of the IoT device is obtained through matching in the following way: When the risk assessment value RS < the risk threshold TRS, it indicates that the risk status of the IoT device is normal and the IoT device is in a healthy state. When the risk assessment value RS ≥ the risk threshold TRS, it indicates that the risk status of the IoT device is abnormal and the IoT device is in an unhealthy state. Based on the health index HI, remaining lifetime RUL, and time to failure TTF of the IoT device, the failure probability Pfa of the IoT device is calculated. The failure probability Pfa is obtained by weighting the health index HI, remaining lifetime RUL, and time to failure TTF. The decision support and optimization module provides decision support based on the adjustment suggestions of IoT devices and adjusts resource allocation.

2. The IoT security management and control system based on big data according to claim 1, characterized in that: The data acquisition module includes a sensor data acquisition unit and a device data acquisition unit; The sensor data acquisition unit collects the temperature T, battery voltage V, and power consumption E of the IoT device through sensors; the temperature T is acquired through a temperature sensor, the battery voltage V is acquired through a voltage sensor, and the power consumption E is acquired through a power sensor. The device data acquisition unit collects the usage frequency f and runtime L of the IoT device through the usage log of the working device; The data collected by the sensor data acquisition unit and the data collected by the device data acquisition unit are fitted to obtain the data running set W.

3. The IoT security management and control system based on big data according to claim 2, characterized in that: The data preprocessing module includes a data cleaning unit and a data standardization unit; The data cleaning unit cleans the data run set W, including processing missing values, outliers and duplicate data, and obtains the cleaned data set WC. Missing values ​​are handled by filling in missing values ​​using interpolation methods; Outliers are detected in the data using the IQR method, and then corrected, removed, and marked. Deduplication checks for duplicate records in the data run set W, including removing duplicate data with the same timestamp; The data standardization unit performs Z-score standardization on the cleaned data set WC, transforming data of different scales into the same standard range to obtain the data standard set WB.

4. The IoT security management and control system based on big data according to claim 1, characterized in that: The feature engineering module includes a temporal feature extraction unit and a health index construction unit; The timing feature extraction unit extracts the timing features of the device from the data standard set WB, including usage frequency change Δf, temperature fluctuation ΔT and power consumption change ΔE, forming a timing feature group F; The frequency change Δf is obtained by the difference between the frequency of use of the IoT device at time t and the frequency of use of the IoT device at time t-1. The temperature fluctuation ΔT is obtained by calculating the average temperature of the IoT devices; The power consumption change ΔE is obtained by the difference between the power consumption of the IoT device at time t and the power consumption of the IoT device at time t-1.

5. The IoT security management and control system based on big data according to claim 4, characterized in that: The health index construction unit calculates and obtains a comprehensive health index HI based on the time-series feature group F, performs standardization processing, and uses the health index HI to determine the health status of IoT devices. The health index HI is obtained by weighted summation of usage frequency change Δf, temperature fluctuation ΔT, and power consumption change ΔE. The health status of the IoT device is obtained through the following methods: When 0.5 ≤ Health Index HI < 1, it indicates that the IoT device is in normal health condition; When 0 < Health Index HI < 0.5, it indicates that the health status of the IoT device is abnormal.

6. The IoT security management and control system based on big data according to claim 4, characterized in that: The health prediction model training module includes a feature selection unit and a training and prediction unit; The feature selection unit uses the Pearson correlation coefficient R to select features in the time-series feature group F that are related to remaining lifetime RUL and time to failure TTF. The training and prediction unit trains the device health prediction model using a regression model and predicts the remaining lifetime (RUL) and time to failure (TTF) of the IoT device. Specifically, the mean squared error (MSE) is used as the loss function to train the regression model; The remaining lifetime RUL is obtained by calculating the features in the time-series feature group F; The time-of-failure (TTF) is obtained by calculating the features in the time-series feature group F.

7. The IoT security management and control system based on big data according to claim 1, characterized in that: The maintenance recommendation generation unit generates recommendations for preventative maintenance and replacement of IoT devices based on the risk assessment value RS and failure probability Pfa of the IoT devices. When an IoT device is in an abnormal state, maintenance and replacement adjustment suggestions are generated and the administrator is notified via SMS and email. The adjustment recommendations are obtained through matching in the following ways: When the risk assessment value RS≥0.8 or the failure probability Pfa≥0.7, it indicates that the IoT device is in the first risk state, the first adjustment suggestion is initiated, and preventive maintenance suggestions are generated. When the risk assessment value RS≥0.9 or the failure probability Pfa≥0.8, it indicates that the IoT device is in the second risk state, and the second adjustment recommendation is initiated, which is to replace the IoT device. When the risk assessment value RS < 0.8 or the failure probability Pfa < 0.7, it indicates that the IoT device is in the third risk state. Activation indicates the third adjustment recommendation. The IoT device is operating normally and does not require maintenance.

8. The IoT security management and control system based on big data according to claim 7, characterized in that: The decision support and optimization module includes a maintenance strategy decision-making unit and a resource allocation and optimization unit; The maintenance strategy decision-making unit provides maintenance strategy decisions based on the obtained adjustment suggestions, specifically including: When IoT devices are in a state of first risk, they require planned maintenance, and a maintenance plan should be generated for each IoT device. When an IoT device is in a second-risk state, IoT device maintenance and replacement are required. The resource allocation and optimization unit allocates appropriate maintenance resources, including maintenance personnel, tools, and spare parts, to different IoT devices based on the first and second risk states determined by the maintenance strategy. It also schedules maintenance windows and maintenance cycles for IoT devices based on the risk assessment value RS and the failure probability Pfa, and rationally allocates maintenance personnel and resources.

9. A big data-based IoT security management and control method, applied to the big data-based IoT security management and control system of any one of claims 1 to 8, characterized in that: Includes the following steps: Step 1: The data acquisition module is responsible for collecting the operational data of IoT devices and forming a data operation set W; Step 2: The data preprocessing module cleans and standardizes the data run set W to obtain the data standard set WB; Step 3: The feature engineering module extracts features from the data standard set WB, including frequency variation Δf, temperature fluctuation ΔT, and power consumption variation ΔE, and forms a time series feature group F; Step 4: The health prediction model training module is based on the time series feature group F. It uses a regression model to train the prediction model and obtain the prediction result Y, including the remaining lifetime (RUL) and time to failure (TTF) of the IoT device. Step 5: The Fault Prediction and Risk Assessment module uses Remaining Life (RUL) and Time to Fail (TTF) to predict the status of IoT devices and assess fault risks, obtain the risk assessment value (RS) and failure probability (Pfa) of IoT devices, and generate adjustment suggestions. Step Six: The Decision Support and Optimization module provides decision support based on the adjustment suggestions from IoT devices and adjusts resource allocation.

Images