An AI-based security protection method, device, equipment and storage medium

Abstract

This invention discloses an AI-based security and prevention method, device, equipment, and storage medium, belonging to the field of security and prevention technology. The method includes: constructing an environmental state model based on environmental vibration and temperature, and outputting environmental state coefficients; constructing a thermal imaging confidence model and a Hall sensor confidence model based on multiple confidence data from a thermal imaging camera and a Hall sensor, respectively; constructing a dynamic weight allocation model based on two types of confidence under the environmental state coefficients, and outputting the weights of each sensor in the security state model; finally, determining the charging safety status based on the security state coefficients and executing corresponding measures. This invention significantly improves the accuracy and adaptability of the security system through multi-sensor fusion and dynamic weight adjustment.

Description

Technical Field

[0001] This invention belongs to the field of security and prevention technology, and in particular relates to an AI-based security and prevention method, device, equipment and storage medium. Background Technology

[0002] With the widespread use of electric vehicle charging facilities, safety issues during the charging process have become increasingly prominent, necessitating intelligent security and prevention measures. The in-depth application of artificial intelligence technology in the security field has provided new possibilities for real-time monitoring and early warning, making an AI-based security and prevention method a research hotspot.

[0003] Current charging safety monitoring systems mostly rely on a single sensor (such as a temperature or current sensor) for threshold judgment, or employ a fixed-weight fusion strategy in multi-sensor systems. These methods are prone to false alarms or missed alarms in complex and changing environments due to sensor performance degradation or environmental interference. They lack a comprehensive consideration of sensor confidence and environmental factors, making it difficult to achieve accurate and adaptive safety status assessment.

[0004] In existing technologies, there is no systematic approach that incorporates multiple factors such as environmental vibration, temperature, dust, humidity, voltage fluctuations, magnetic field interference, and sensor usage time into a unified model and dynamically adjusts sensor weights. Therefore, developing an AI-based security and prevention method that can integrate multi-source information and is environmentally adaptable has significant practical implications and application value. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an AI-based security and prevention method, device, equipment, and storage medium, which solves the aforementioned problems.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an AI-based security and prevention method, device, equipment, and storage medium, comprising the following steps:

[0007] An environmental state model is constructed based on the environmental vibration amplitude and environmental temperature, and environmental state coefficients are output.

[0008] A thermal imaging confidence model is constructed based on the confidence data of the thermal imaging camera (distance between the thermal imaging camera and the object under test (trolley), ambient dust concentration, ambient humidity, and usage time of the thermal imaging camera) to output the thermal imaging confidence.

[0009] Based on the Hall sensor confidence data (power supply voltage fluctuation value (absolute difference between actual power supply voltage and reference power supply voltage), the ambient magnetic field strength, the distance between the central axis of the measuring wire and the central axis of the Hall sensor hole, and the usage time of the Hall sensor), a Hall sensor confidence model is constructed and the Hall sensor confidence is output.

[0010] A dynamic weight allocation model is constructed based on the confidence scores of thermal imaging and Hall sensors under environmental state coefficients, and the respective weights of the thermal imaging camera detection results and Hall sensor detection results in the pre-constructed safety state model are output.

[0011] The charging safety status is determined based on the current safety status coefficient output by the safety status model.

[0012] Based on the above technical solutions, the present invention also provides the following optional technical solutions:

[0013] Further technical solution: The steps for determining the charging safety status based on the current safety status coefficient output by the safety status model are as follows:

[0014] The current security status coefficient Compare with the corresponding threshold:

[0015] like This indicates that the charging status is safe;

[0016] like Then the charging status will be continuously monitored;

[0017] like If so, an alarm message will be generated;

[0018] like If so, the power will be forcibly cut off.

[0019] A further technical solution: The security state model is represented as follows:

[0020]

[0021]

[0022]

[0023] in, This represents the current safety status coefficient. Indicates the weight of the detection results from the thermal imaging camera. Indicates the weight of the Hall sensor detection results. Indicates the temperature hazard factor. Indicates the current hazard factor. This indicates the temperature detected by the thermal imaging camera. This indicates the safe temperature threshold (below this value is considered safe). This indicates the critical temperature threshold (reaching or exceeding this value is considered extremely dangerous). This indicates the detection result (current value) of the Hall sensor. This indicates the normal charging current. Indicates the dangerous current threshold, the The The higher the value, the safer the charging state.

[0024] A further technical solution: The dynamic weight allocation model is expressed as follows:

[0025]

[0026]

[0027]

[0028]

[0029] in, Indicates the weight of the detection results from the thermal imaging camera. Indicates the weight of the Hall sensor detection results. Indicates the current confidence level of the thermal imaging. This indicates the current confidence level of the Hall sensor. This represents the current environmental state coefficient. This represents the sensitivity coefficient of the environment to the thermal imaging camera. This represents the sensitivity coefficient of the environment to the Hall sensor.

[0030] Further technical solution: The steps to construct a Hall sensor confidence model and output the Hall sensor confidence score based on Hall sensor confidence data (supply voltage fluctuation value (absolute difference between actual supply voltage and reference supply voltage), ambient magnetic field strength, distance between the central axis of the measuring wire and the central axis of the Hall sensor aperture, and Hall sensor usage time) are as follows:

[0031] The voltage fluctuation value (the absolute difference between the actual supply voltage and the reference supply voltage), the ambient magnetic field strength, the distance between the central axis of the measuring wire and the central axis of the Hall sensor hole, and the usage time of the Hall sensor are processed by maximum-minimum normalization to obtain the voltage fluctuation index, magnetic field strength index, distance deviation index, and Hall sensor duration index.

[0032] The current voltage fluctuation index, current magnetic field strength index, distance deviation index, and Hall sensor duration index are imported into the Hall sensor confidence model to output the current Hall sensor confidence level. The Hall sensor confidence model is expressed as follows:

[0033]

[0034] in, This indicates the current confidence level of the Hall sensor. This indicates the current voltage fluctuation index. This indicates the current magnetic field strength index. This indicates the current distance deviation index. This indicates the current duration index of the Hall sensor. This represents the voltage fluctuation attenuation coefficient. This represents the magnetic field strength attenuation coefficient. This represents the distance deviation attenuation coefficient. The duration attenuation coefficient is represented by the following. Furthermore, the larger the value, the more accurate the detection result of the Hall sensor.

[0035] Further technical solution: The steps for constructing a thermal imaging confidence model and outputting the thermal imaging confidence score based on thermal imaging camera confidence data (detection distance (distance between the thermal imaging camera and the object being measured (trolleybus), environmental dust concentration, environmental humidity, and thermal imaging camera usage time) are as follows:

[0036] The camera duration index is obtained by performing maximum-min normalization on the thermal imaging camera usage time.

[0037] The detection distance, ambient dust concentration, and ambient humidity are respectively imported into the formula. The distance index, dust concentration index, and humidity index were obtained. This indicates the corresponding detection distance, ambient dust concentration, and ambient humidity. This indicates the corresponding reference distance, reference ambient dust concentration, and reference humidity.

[0038] The current camera duration index, current distance index, current dust concentration index, and current humidity index are imported into the thermal imaging confidence model to output the current thermal imaging confidence score. The thermal imaging confidence model is expressed as follows:

[0039]

[0040] in, Indicates the current confidence level of the thermal imaging. Indicates the current distance index. This indicates the current dust concentration index. This indicates the current humidity index. This indicates the current camera duration index. Represents the weight coefficient and The The higher the value, the more accurate the detection results of the thermal imaging camera.

[0041] Further technical solution: The steps for constructing an environmental state model based on environmental vibration amplitude and environmental temperature and outputting environmental state coefficients are as follows:

[0042] The environmental vibration index is obtained by comparing the environmental vibration amplitude with the vibration amplitude reference value.

[0043] The ambient temperature index is obtained by comparing the absolute difference between the ambient temperature and the standard ambient temperature with the allowable deviation from the standard ambient temperature.

[0044] The current environmental vibration index and current environmental temperature index are imported into the environmental state model to output the current environmental state coefficients. The environmental state model is represented as follows:

[0045]

[0046] in, This represents the current environmental state coefficient. Indicates the vibration attenuation coefficient. This indicates the current environmental vibration index. Indicates the ambient temperature attenuation coefficient. The current ambient temperature index is indicated by the following: The higher the value, the better the environmental condition.

[0047] An AI-based security device employs the aforementioned AI-based security method.

[0048] An AI-based security device includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the program, it implements the aforementioned AI-based security method.

[0049] A storage medium storing executable instructions that, when executed by a processor, implement the aforementioned AI-based security and prevention method.

[0050] Beneficial effects

[0051] This invention provides an AI-based security and prevention method, device, equipment, and storage medium, which has the following advantages compared with the prior art:

[0052] 1. This invention integrates data from multiple sources, such as thermal imaging cameras and Hall sensors, and dynamically adjusts sensor weights based on environmental conditions. This effectively avoids false alarms or missed alarms caused by the failure of a single sensor or environmental interference, and significantly improves the monitoring accuracy and robustness of the system in complex environments.

[0053] 2. By introducing a sensor confidence model, this invention comprehensively considers actual factors such as sensor usage time, power supply fluctuations, magnetic field interference, dust and humidity, to achieve intelligent perception and compensation for sensor performance degradation and external environment, ensuring long-term stable operation of the system.

[0054] 3. This invention implements a multi-level safety response mechanism based on a safety status coefficient, including measures such as continuous monitoring, alarms, and forced power outages. The system can automatically take corresponding actions according to the real-time risk level, thereby improving the refinement and intelligence of safety management.

[0055] 4. This invention does not rely on a specific hardware platform and can be flexibly adapted to different sensor types and environmental scenarios through software definition. It is applicable to multiple fields such as electric vehicle (electric car and electric bicycle) charging safety, industrial equipment monitoring, and warehouse security, and has broad application value. Attached Figure Description

[0056] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0058] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0059] Please see Figure 1 The present invention provides an AI-based security and prevention method, comprising the following steps:

[0060] An environmental state model is constructed based on the environmental vibration amplitude and environmental temperature, and environmental state coefficients are output.

[0061] A thermal imaging confidence model is constructed based on the confidence data of the thermal imaging camera (distance between the thermal imaging camera and the object under test (trolley), ambient dust concentration, ambient humidity, and usage time of the thermal imaging camera) to output the thermal imaging confidence.

[0062] Based on the Hall sensor confidence data (power supply voltage fluctuation value (absolute difference between actual power supply voltage and reference power supply voltage), the ambient magnetic field strength, the distance between the central axis of the measuring wire and the central axis of the Hall sensor hole, and the usage time of the Hall sensor), a Hall sensor confidence model is constructed and the Hall sensor confidence is output.

[0063] A dynamic weight allocation model is constructed based on the confidence scores of thermal imaging and Hall sensors under environmental state coefficients, and the respective weights of the thermal imaging camera detection results and Hall sensor detection results in the pre-constructed safety state model are output.

[0064] The charging safety status is determined based on the current safety status coefficient output by the safety status model.

[0065] Through the above technical solution, this application can accurately identify changes in sensor reliability when charging piles face complex environmental interference. For example, in strong magnetic field interference scenarios, it automatically reduces the decision weight of Hall sensors to avoid false power outages caused by current detection distortion. Simultaneously, by comprehensively assessing the impact of environmental factors such as vibration and temperature on the overall safety status, for example, by increasing the sensitivity of the safety status coefficient in continuous vibration environments, it provides timely warnings of connector loosening risks. This solution effectively solves the problems of false alarms and missed alarms caused by neglecting sensor performance degradation and environmental coupling effects in traditional methods, improving the accuracy and environmental adaptability of safety status assessment.

[0066] Preferably, the steps for constructing an environmental state model based on environmental vibration amplitude and environmental temperature and outputting environmental state coefficients are as follows:

[0067] The environmental vibration index is obtained by comparing the environmental vibration amplitude with the vibration amplitude reference value.

[0068] The ambient temperature index is obtained by comparing the absolute difference between the ambient temperature and the standard ambient temperature with the allowable deviation from the standard ambient temperature.

[0069] The current environmental vibration index and current environmental temperature index are imported into the environmental state model to output the current environmental state coefficients. The environmental state model is represented as follows:

[0070]

[0071] in, This represents the current environmental state coefficient. Indicates the vibration attenuation coefficient. This indicates the current environmental vibration index. Indicates the ambient temperature attenuation coefficient. The current ambient temperature index is indicated by the following: The higher the value, the better the environmental condition.

[0072] The processing of the ratio between environmental vibration amplitude and the vibration amplitude reference value involves dividing the measured vibration amplitude by a preset vibration benchmark value. Specifically, this can be achieved by using an accelerometer to collect vibration data and calculating the ratio to the reference value. This eliminates the influence of different vibration dimensions and quantifies the interference of mechanical vibration on equipment stability. The processing of the ratio between the absolute difference between ambient temperature and standard ambient temperature involves calculating the absolute value of the deviation between the measured temperature value and the standard value, and then dividing it by the maximum allowable temperature fluctuation range. This can be achieved by using a temperature sensor to acquire real-time data and performing normalization calculations. This distinguishes the impact of normal temperature fluctuations and abnormal temperature rises on sensor performance. The vibration attenuation coefficient and temperature attenuation coefficient are pre-set weighting parameters, which can be determined through empirical calibration or machine learning optimization. These are used to adjust the differences in the contributions of vibration and temperature in environmental condition assessment. The exponential function model refers to an environmental condition calculation method based on a natural exponential function. Specifically, an exponential attenuation function can be used to nonlinearly fuse vibration and temperature exponents to suppress the abrupt impact of extreme environmental parameters on the assessment results.

[0073] Specifically, the environmental vibration index reflects the degree of interference of mechanical vibration on sensor stability by comparing the measured vibration amplitude with a reference value. For example, when the vibration amplitude exceeds the reference value, the index increases, indicating that the vibration interference is aggravated. The environmental temperature index quantifies the impact of the thermal environment on sensor accuracy by calculating the degree of temperature deviation from the standard value. For example, when the temperature deviation is close to the maximum allowable value, the index approaches 1, indicating a significant temperature anomaly. These two indices capture environmental interference factors from mechanical and thermodynamic perspectives, respectively, and are then weighted and fused using an exponential function model. The vibration attenuation coefficient and temperature attenuation coefficient adjust their weights according to the sensitivity of the environmental factors to the system; for example, the temperature attenuation coefficient can be increased in high-temperature sensitive scenarios. The final output environmental state coefficient varies continuously between 0 and 1; a larger value indicates a more stable environmental state, providing a dynamic basis for subsequent sensor confidence adjustment.

[0074] Compared to existing technologies, traditional methods typically monitor only a single environmental parameter or use fixed thresholds to determine environmental conditions, such as monitoring only whether the temperature exceeds a preset threshold. This proposed solution, however, integrates the effects of vibration and temperature to construct a dynamic attenuation model, enabling a more comprehensive reflection of the combined effects of environmental disturbances. Existing technologies do not consider the cumulative impact of vibration on sensor performance, while this solution quantifies the degree of disturbance caused by vibration intensity changing over time by processing the ratio of the vibration index to a reference value. Furthermore, existing environmental assessment models often employ linear superposition, while this solution's exponential function model more accurately simulates the nonlinear relationship between environmental disturbances and system stability.

[0075] Through the above technical solution, this application achieves dynamic quantitative assessment of the environmental conditions surrounding charging facilities, solving the environmental modeling error problem caused by neglecting the combined effects of vibration and temperature in traditional methods. By normalizing the vibration and temperature indices, interference from different physical dimensions is eliminated, enabling the environmental state coefficient to objectively reflect the degree of influence of complex environmental factors on sensor performance. The application of the exponential function model effectively suppresses the distortion of assessment results caused by extreme environmental parameter mutations, improving the continuity and reliability of environmental state assessment and providing accurate input parameters for subsequent sensor confidence adjustment.

[0076] Preferably, the steps for constructing a thermal imaging confidence model and outputting the thermal imaging confidence score based on thermal imaging camera confidence data (detection distance (distance between the thermal imaging camera and the object being measured (trolleybus)), environmental dust concentration, environmental humidity, and thermal imaging camera usage time) are as follows:

[0077] The camera duration index is obtained by performing maximum-min normalization on the thermal imaging camera usage time.

[0078] The detection distance, ambient dust concentration, and ambient humidity are respectively imported into the formula. The distance index, dust concentration index, and humidity index were obtained. This indicates the corresponding detection distance, ambient dust concentration, and ambient humidity. This indicates the corresponding reference distance, reference ambient dust concentration, and reference humidity.

[0079] The current camera duration index, current distance index, current dust concentration index, and current humidity index are imported into the thermal imaging confidence model to output the current thermal imaging confidence score. The thermal imaging confidence model is expressed as follows:

[0080]

[0081] in, Indicates the current confidence level of the thermal imaging. Indicates the current distance index. This indicates the current dust concentration index. This indicates the current humidity index. This indicates the current camera duration index. Represents the weight coefficient and The The higher the value, the more accurate the detection results of the thermal imaging camera.

[0082] Among them, the exponential function Reference values ​​in These refer to pre-set reference parameters, such as a reference distance of 5 meters, a reference dust concentration of 100 μg / m³, and a reference humidity of 80% RH, used to quantify the nonlinear impact of environmental parameters on image quality. Weighting coefficients The normalization constraint refers to satisfying Specifically, the relative importance of each factor can be determined by setting empirical values ​​or by using the analytic hierarchy process (AHP) to ensure that the contribution of each influencing factor is reasonably allocated under different environmental conditions.

[0083] Specifically, as the detection distance increases, the distance index... The ratio of distance to reference value decreases exponentially with increasing distance, reflecting the characteristic of decreased accuracy at long distances. Environmental dust concentration and humidity are converted into a dust concentration index via a corresponding exponential function. and humidity index When the actual value exceeds the reference value, the index value decreases rapidly to indicate increased environmental interference. Camera Duration Index The cumulative impact of sensor aging on confidence is represented by a linearly decreasing form. By weighted fusion of these four parameters, the constructed confidence model can dynamically reflect changes in the reliability of thermal imaging detection in complex environments. For example, when the camera usage time reaches a preset maximum threshold, This causes the contribution of the duration term to drop to zero, at which point the model mainly relies on environmental parameters to assess confidence.

[0084] Compared to existing technologies, traditional methods typically consider only a single environmental factor or employ fixed-weight fusion strategies, such as judging solely based on temperature thresholds or linearly superimposing factors like distance and humidity. This proposed solution, however, establishes a nonlinear mapping relationship through an exponential function, more accurately reflecting the attenuation effect of environmental parameters on image quality. Furthermore, it introduces usage time normalization to address the continuous decline in detection accuracy caused by sensor aging. In addition, the dynamic weight allocation mechanism overcomes the limitation of fixed weights in adapting to the coupling changes of multiple factors.

[0085] Through the above technical solution, this application can accurately assess the reliability of thermal imaging detection under complex operating conditions of charging facilities. When the detection distance exceeds the safe range, the distance index automatically decreases to reflect the spatial attenuation effect; when a sudden increase in dust concentration causes image blurring, the dust concentration index decreases, triggering a confidence warning; when the cumulative working time of the camera reaches a critical value, the duration index decreases significantly to indicate maintenance needs. This model, through multi-dimensional parameter coupling analysis, effectively suppresses misjudgments caused by environmental interference and device aging, providing an accurate detection data foundation for subsequent safety status assessment.

[0086] Preferably, the steps for constructing a Hall sensor confidence model and outputting the Hall sensor confidence level based on Hall sensor confidence data (supply voltage fluctuation value (absolute difference between actual supply voltage and reference supply voltage), ambient magnetic field strength, distance between the central axis of the measuring wire and the central axis of the Hall sensor aperture, and Hall sensor usage time) are as follows:

[0087] The voltage fluctuation value (the absolute difference between the actual supply voltage and the reference supply voltage), the ambient magnetic field strength, the distance between the central axis of the measuring wire and the central axis of the Hall sensor hole, and the usage time of the Hall sensor are processed by maximum-minimum normalization to obtain the voltage fluctuation index, magnetic field strength index, distance deviation index, and Hall sensor duration index.

[0088] The current voltage fluctuation index, current magnetic field strength index, distance deviation index, and Hall sensor duration index are imported into the Hall sensor confidence model to output the current Hall sensor confidence level. The Hall sensor confidence model is expressed as follows:

[0089]

[0090] in, This indicates the current confidence level of the Hall sensor. This indicates the current voltage fluctuation index. This indicates the current magnetic field strength index. This indicates the current distance deviation index. This indicates the current duration index of the Hall sensor. This represents the voltage fluctuation attenuation coefficient. This represents the magnetic field strength attenuation coefficient. This represents the distance deviation attenuation coefficient. The duration attenuation coefficient is represented by the following. Furthermore, the larger the value, the more accurate the detection result of the Hall sensor.

[0091] The power supply voltage fluctuation value refers to the absolute difference between the actual power supply voltage and the reference power supply voltage. This can be achieved by collecting voltage data using a voltage sensor and calculating the difference using a difference module, reflecting the impact of power supply stability on the sensor's measurement accuracy. The ambient magnetic field strength refers to the intensity of magnetic field interference around the Hall sensor's installation location. This can be measured using a magnetometer and converted into a numerical value, quantifying the degree of interference from the external magnetic field on current measurement. The distance between the central axis of the measuring wire and the central axis of the Hall sensor hole refers to the physical offset caused by installation position deviation. This can be obtained using a laser rangefinder or image recognition technology, characterizing the measurement error caused by mechanical displacement of the sensor. The Hall sensor usage time refers to the cumulative working time of the sensor, recorded by a timer module, used to assess performance degradation caused by sensor aging. The attenuation coefficients of each influencing factor can be determined through expert experience calibration or historical data fitting analysis.

[0092] Specifically, the system first converts power supply voltage fluctuations, ambient magnetic field strength, installation distance deviation, and usage time into dimensionless exponents through normalization. For example, the voltage fluctuation exponent can be calculated by comparing the current voltage difference with the historical maximum fluctuation value. Then, each exponent is substituted into an exponential decay model, and a confidence score is generated through weighted summation and exponential operations. For instance, when the voltage fluctuation exponent increases due to abnormal power supply, its corresponding decay term increases, leading to an exponential decrease in the confidence score, thus accurately reflecting the negative impact of power supply instability on sensor reliability. This model allows for flexible control of the contribution weights of different factors by adjusting the decay coefficient. For example, when ambient magnetic field interference is the primary risk, the magnetic field strength decay coefficient can be increased to strengthen the suppression effect of this factor on the confidence score. The final output confidence score reflects the sensor's reliability level under complex operating conditions in real time, providing data support for subsequent dynamic weight allocation.

[0093] Compared to existing technologies, traditional methods typically determine sensor status based on a single threshold, such as monitoring whether voltage fluctuations exceed a fixed limit, while ignoring multi-factor coupling effects and dynamic parameter changes. Existing technologies also lack quantitative assessment mechanisms for installation deviations and sensor aging; for example, they do not include measurement errors caused by mechanical displacement in confidence level calculations. This solution, through multi-dimensional parameter modeling and an exponential decay mechanism, achieves for the first time a synergistic analysis of power supply stability, magnetic field interference, installation accuracy, and aging degree. For instance, in a strong magnetic field environment, even if the voltage is stable, the confidence level will still significantly decrease due to the exponential increase in magnetic field strength, thus more comprehensively capturing potential risks.

[0094] Through the above technical solution, this application effectively solves the problem of confidence decline of Hall sensors caused by multiple interference factors in complex environments. For example, when the installation position shifts due to vibration of the charging pile, the weight of the sensor can be automatically reduced through the distance deviation index, avoiding the impact of current measurement errors caused by mechanical displacement on the overall safety assessment. Simultaneously, this solution can dynamically identify the aging degree of the sensor. For instance, when the usage time index reaches a critical value, its confidence weight is automatically reduced to prevent the risk of missed alarms caused by component performance degradation, thereby improving the accuracy and reliability of charging safety status assessment.

[0095] Preferably, the security state model is represented as follows:

[0096]

[0097]

[0098]

[0099] in, This represents the current safety status coefficient. Indicates the weight of the detection results from the thermal imaging camera. Indicates the weight of the Hall sensor detection results. Indicates the temperature hazard factor. Indicates the current hazard factor. This indicates the temperature detected by the thermal imaging camera. This indicates the safe temperature threshold (below this value is considered safe). This indicates the critical temperature threshold (reaching or exceeding this value is considered extremely dangerous). This indicates the detection result (current value) of the Hall sensor. This indicates the normal charging current. Indicates the dangerous current threshold, the The The higher the value, the safer the charging state.

[0100] The safe temperature threshold refers to a pre-set temperature critical point. When the detected temperature is below this value, it is considered safe. This can be achieved using the upper limit of the surface temperature of the charging equipment specified in industry standards, for example, 40℃. The critical temperature threshold is the temperature critical point that triggers emergency protection. This can be set to 80℃ based on the material's temperature resistance characteristics. The normal charging current refers to the rated operating current of the charging equipment under standard operating conditions. This can be set according to the nominal value of the battery management system. The dangerous current threshold is the critical current value that may cause circuit overload. This can be set to 150% of the rated current. Dynamic weight allocation refers to dynamically adjusting the contribution ratio of temperature and current detection results based on sensor confidence levels. This can be achieved through confidence ratio calculations.

[0101] Specifically, the temperature hazard coefficient maps the degree of temperature anomaly to a range of 0 to 1 by calculating the proportion of detected temperatures exceeding the safety threshold. When the detected temperature is below the safety threshold, the temperature hazard coefficient is set to zero to avoid false alarms; when the temperature exceeds the critical threshold, the hazard coefficient reaches its maximum value of 1 to trigger the highest alarm. The current hazard coefficient calculates the absolute deviation between the measured current and the normal value, and combines it with a hazard threshold for normalization, enabling the simultaneous identification of abnormal states such as excessively high or low current. The safety status coefficient integrates the hazard levels of temperature and current through a weighted summation method. The weighting coefficient is dynamically adjusted based on the sensor's real-time confidence level, giving sensors with higher reliability of detection data a greater weight in the evaluation. Normalization transforms physical quantities of different dimensions into a unified hazard indicator, while the dynamic weighting mechanism effectively suppresses errors caused by sensor performance degradation or environmental interference.

[0102] Compared to existing technologies, traditional methods use fixed thresholds to judge a single parameter, failing to quantify the degree of danger and ignoring changes in sensor reliability. This solution achieves quantitative assessment of the degree of temperature and current anomalies by constructing a normalized hazard coefficient; through a dynamic weight allocation mechanism, the assessment strategy can be automatically adjusted according to environmental conditions and sensor status; and through multi-source data fusion, the synergistic effect of two key indicators, temperature mutation and current anomaly, is integrated, significantly improving the assessment accuracy under complex operating conditions.

[0103] Through the above technical solutions, this application effectively solves the problem of misjudgment caused by sensor performance degradation, suppresses the influence of environmental interference through a dynamic weight adjustment mechanism, and achieves adaptive assessment of charging safety status. Quantitative calculation of temperature and current hazard coefficients provides an accurate basis for graded early warning, normalization processing ensures the effective fusion of multi-source heterogeneous data, and dynamic weight allocation enhances the system's adaptability to changes in sensor reliability.

[0104] Preferably, the dynamic weight allocation model is expressed as:

[0105]

[0106]

[0107]

[0108]

[0109] in, Indicates the weight of the detection results from the thermal imaging camera. Indicates the weight of the Hall sensor detection results. Indicates the current confidence level of the thermal imaging. This indicates the current confidence level of the Hall sensor. This represents the current environmental state coefficient. This represents the sensitivity coefficient of the environment to the thermal imaging camera. This represents the sensitivity coefficient of the environment to the Hall sensor.

[0110] Among these, the current thermal imaging confidence level refers to the detection reliability of a thermal imaging camera under specific conditions, reflecting the accuracy of thermal imaging detection. The current Hall sensor confidence level refers to the detection reliability of a Hall sensor under specific conditions, reflecting the accuracy of current detection. The environmental state coefficient characterizes the degree of interference from the environment on sensor operation. The sensitivity coefficient refers to the correction strength of different environmental factors on the sensor confidence level, which can be achieved using preset adjustment parameters. For example, the sensitivity coefficient of a thermal imaging camera can be set to 0.5, and the sensitivity coefficient of a Hall sensor can be set to 0.3, quantifying the differences in the impact of environmental changes on different sensors.

[0111] Specifically, the dynamic weight allocation model exponentially corrects the confidence scores of the two types of sensors using an environmental state coefficient. This ensures that the confidence scores of sensors more sensitive to environmental influences decrease more significantly when environmental conditions deteriorate. For example, when environmental vibration intensifies or the temperature deviates from the standard value, the environmental state coefficient decreases. If the thermal imaging camera is more sensitive to vibration, its corrected confidence score will decrease exponentially, leading to a smaller proportion in the weight calculation. By normalizing the corrected confidence scores, dynamic weight allocation is achieved, ensuring that sensor data less susceptible to interference is prioritized in complex environments, thereby improving the reliability of the fusion results.

[0112] Compared to existing technologies, traditional methods employ a fixed weight allocation strategy, failing to consider the dynamic impact of environmental changes on sensor performance. This makes them prone to misjudgments due to decreased sensor confidence levels caused by increased dust concentration or voltage fluctuations. In contrast, this proposed solution introduces environmental state coefficients and sensitivity coefficients to establish a quantitative correlation between sensor confidence and environmental factors. This allows the weight allocation to adaptively adjust with environmental conditions, effectively suppressing detection errors caused by environmental interference.

[0113] Through the above technical solution, this application solves the problem of unreliable fusion strategies caused by environmental interference and sensor performance degradation in charging safety monitoring. It achieves dynamic optimization of sensor weights, reduces the risk of false alarms and false negatives, and improves the accuracy and environmental adaptability of safety status assessment. For example, in high temperature and high humidity environments, the confidence correction value of the thermal imaging camera decays due to the decrease in the environmental state coefficient, and its weight decreases accordingly. The system automatically increases the weight ratio of Hall sensor data to avoid erroneous judgments caused by thermal imaging detection distortion.

[0114] Preferably, the step of determining the charging safety status based on the current safety status coefficient output by the safety status model is as follows:

[0115] The current security status coefficient Compare with the corresponding threshold:

[0116] like This indicates that the charging status is safe;

[0117] like Then the charging status will be continuously monitored;

[0118] like If so, an alarm message will be generated;

[0119] like If so, the power will be forcibly cut off.

[0120] The safety status coefficient is used to quantitatively characterize the real-time safety level of the charging system. Threshold interval division refers to a segmented judgment standard set according to actual safety requirements, which can be determined using engineering experience data combined with machine learning training, to establish response mechanisms corresponding to different risk levels. Continuous monitoring refers to high-frequency collection of charging parameters, which can be achieved by shortening the sensor sampling interval, to capture potential risk trends. Alarm information generation refers to triggering audible and visual alarm devices, which can be achieved by sending early warning signals to the monitoring center using an IoT communication module, to provide early warning of abnormal states. Forced power cut-off refers to disconnecting the charging circuit, which can be achieved by using a relay control circuit to disconnect the power supply, to prevent extreme dangerous situations from occurring.

[0121] Specifically, when the safety coefficient is between 0.8 and 1, the system maintains the original charging process without intervention, and the charging equipment is in its optimal operating state. When the coefficient drops to between 0.5 and 0.8, the system automatically increases the sensor data acquisition frequency to twice the normal level, for example, adjusting the temperature sampling interval from 30 seconds to 15 seconds, and simultaneously activates the data change rate analysis module. When the coefficient further drops to between 0.3 and 0.5, the system simultaneously activates the local buzzer alarm and the remote monitoring platform pop-up notification, and generates a diagnostic report containing abnormal parameter types through preset warning logic. When the coefficient exceeds the critical value of 0.3, the system completes the charging circuit disconnection operation within 50 milliseconds, and simultaneously records the equipment operation data for the 60 seconds before the power outage for accident tracing.

[0122] Compared to existing technologies, traditional methods only set a single power outage threshold and lack intermediate early warning mechanisms, such as directly cutting off the power when the temperature exceeds 85°C. This solution establishes a four-level response mechanism, adding two buffer zones between completely safe and extremely dangerous states. This avoids frequent false power outages caused by a single threshold and continuously monitors parameter change trends, providing operators with buffer time for risk management.

[0123] Through the above technical solution, this application effectively solves the problem of simplistic response measures in charging safety monitoring, achieving a smooth transition from normal to dangerous states. By setting a continuous monitoring interval, it is possible to identify whether parameter fluctuations are transient disturbances, reducing the probability of accidental power outages due to momentary anomalies. The alarm interval setting allows potential risks to be detected 10-15 minutes in advance, providing time for manual intervention. The forced power-off threshold is set in an extremely low safety range, ensuring that power-off operations are only performed when a substantial danger is confirmed, thus protecting equipment safety and improving the availability of the charging system.

[0124] An AI-based security device employs the aforementioned AI-based security method.

[0125] An AI-based security device includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the program, it implements the aforementioned AI-based security method.

[0126] A storage medium storing executable instructions that, when executed by a processor, implement the aforementioned AI-based security and prevention method.

[0127] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0128] 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 alterations 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. An AI-based security prevention method, characterized in that, Includes the following steps: An environmental state model is constructed based on the environmental vibration amplitude and environmental temperature, and environmental state coefficients are output. A thermal imaging confidence model is constructed based on the distance between the thermal imaging camera and the object being measured, the concentration of dust in the environment, the humidity of the environment, and the duration of use of the thermal imaging camera, and the thermal imaging confidence is output. Based on the fluctuation value of the power supply voltage of the Hall sensor, the magnetic field strength of the environment, the distance between the central axis of the measuring wire and the central axis of the Hall sensor hole, and the usage time of the Hall sensor, a Hall sensor confidence model is constructed and the Hall sensor confidence is output. A dynamic weight allocation model is constructed based on the confidence scores of thermal imaging and Hall sensors under environmental state coefficients, and the respective weights of the thermal imaging camera detection results and Hall sensor detection results in the pre-constructed safety state model are output. The charging safety status is determined based on the current safety status coefficient output by the safety status model. The security state model is represented as follows: in, This represents the current safety status coefficient. Indicates the weight of the detection results from the thermal imaging camera. Indicates the weight of the Hall sensor detection results. Indicates the temperature hazard factor. Indicates the current hazard factor. This indicates the detection results from the thermal imaging camera. Indicates the safe temperature threshold. This represents the critical temperature threshold. This indicates the detection result of the Hall sensor. This indicates the normal charging current. Indicates the dangerous current threshold, the The The higher the value, the safer the charging state.

2. The AI-based security and prevention method according to claim 1, characterized in that, The steps for determining the charging safety status based on the current safety status coefficient output by the safety status model are as follows: The current security status coefficient Compare with the corresponding threshold: like This indicates that the charging status is safe; like Then the charging status will be continuously monitored; like If so, an alarm message will be generated; like If so, the power will be forcibly cut off.

3. The AI-based security and prevention method according to claim 1, characterized in that, The dynamic weight allocation model is represented as follows: in, Indicates the weight of the detection results from the thermal imaging camera. Indicates the weight of the Hall sensor detection results. Indicates the current confidence level of the thermal imaging. This indicates the current confidence level of the Hall sensor. This represents the current environmental state coefficient. This represents the sensitivity coefficient of the environment to the thermal imaging camera. This represents the sensitivity coefficient of the environment to the Hall sensor.

4. The AI-based security and prevention method according to claim 3, characterized in that, The steps to construct a Hall sensor confidence model and output the Hall sensor confidence score based on the Hall sensor power supply voltage fluctuation value, the ambient magnetic field strength, the distance between the central axis of the measuring wire and the central axis of the Hall sensor aperture, and the Hall sensor usage time are as follows: The voltage fluctuation value, the magnetic field strength of the environment, the distance between the central axis of the measuring wire and the central axis of the Hall sensor hole, and the usage time of the Hall sensor are processed by maximum-minimum normalization to obtain the voltage fluctuation index, magnetic field strength index, distance deviation index, and Hall sensor duration index. The current voltage fluctuation index, current magnetic field strength index, distance deviation index, and Hall sensor duration index are imported into the Hall sensor confidence model to output the current Hall sensor confidence level. The Hall sensor confidence model is expressed as follows: in, This indicates the current confidence level of the Hall sensor. This indicates the current voltage fluctuation index. This indicates the current magnetic field strength index. This indicates the current distance deviation index. This indicates the current duration index of the Hall sensor. This represents the voltage fluctuation attenuation coefficient. This represents the magnetic field strength attenuation coefficient. This represents the distance deviation attenuation coefficient. The duration attenuation coefficient is represented by the following. Furthermore, the larger the value, the more accurate the detection result of the Hall sensor.

5. The AI-based security and prevention method according to claim 3, characterized in that, The steps for constructing a thermal imaging confidence model and outputting the thermal imaging confidence score based on the distance between the thermal imaging camera and the object being measured, the ambient dust concentration, the ambient humidity, and the duration of use of the thermal imaging camera are as follows: The camera duration index is obtained by performing maximum-min normalization on the thermal imaging camera usage time. The detection distance, ambient dust concentration, and ambient humidity are respectively imported into the formula. The distance index, dust concentration index, and humidity index were obtained. This indicates the corresponding detection distance, ambient dust concentration, and ambient humidity. This indicates the corresponding reference distance, reference ambient dust concentration, and reference humidity. The current camera duration index, current distance index, current dust concentration index, and current humidity index are imported into the thermal imaging confidence model to output the current thermal imaging confidence score. The thermal imaging confidence model is expressed as follows: in, Indicates the current confidence level of the thermal imaging. Indicates the current distance index. This indicates the current dust concentration index. This indicates the current humidity index. This indicates the current camera duration index. Represents the weight coefficient and The The higher the value, the more accurate the detection results of the thermal imaging camera.

6. The AI-based security and prevention method according to claim 3, characterized in that, The steps for constructing an environmental state model and outputting environmental state coefficients based on environmental vibration amplitude and environmental temperature are as follows: The environmental vibration index is obtained by comparing the environmental vibration amplitude with the vibration amplitude reference value. The ambient temperature index is obtained by comparing the absolute difference between the ambient temperature and the standard ambient temperature with the allowable deviation from the standard ambient temperature. The current environmental vibration index and current environmental temperature index are imported into the environmental state model to output the current environmental state coefficients. The environmental state model is represented as follows: in, This represents the current environmental state coefficient. Indicates the vibration attenuation coefficient. This indicates the current environmental vibration index. Indicates the ambient temperature attenuation coefficient. The current ambient temperature index is indicated by the following: The higher the value, the better the environmental condition.

7. An AI-based security and prevention device, characterized in that, The AI-based security and prevention method described in any one of claims 1-6 is adopted.

8. An AI-based security and prevention device, characterized in that, It includes a memory and a processor, the memory storing a computer program that can run on the processor, and the processor executing the program to implement the AI-based security and prevention method according to any one of claims 1-6.

9. A storage medium, characterized in that, It stores executable instructions for inducing the processor to execute, thereby implementing the AI-based security and prevention method as described in any one of claims 1-6.

Images