Low-altitude aircraft chip safety monitoring system

Abstract

The invention relates to the technical field of low-altitude aircraft chip safety monitoring, and discloses a low-altitude aircraft chip safety monitoring system which comprises a state acquisition module, a risk assessment module, an anomaly detection module, a strategy generation module, an instruction execution module, an effect verification module, a threshold setting module, a data output module, a feedback optimization module and a system regulation module. The state acquisition module analyzes the operation parameters and the environment interference to obtain a state value; the risk assessment module extracts and screens abnormal protection feature combinations; the anomaly detection module combines a sample variation matching feature with an evaluation combination; the strategy generation module maps a feature strategy based on real-time data and sets a rule; the instruction execution module captures feature data and analyzes a trend; the effect verification module adjusts rules according to errors; the threshold setting module determines a monitoring threshold interval; the data output module generates standardized data; the system realizes comprehensive monitoring of the chip state through cooperation of multiple modules, and the operation safety and stability are improved.

Description

Technical Field

[0001] This invention relates to the field of low-altitude aircraft chip security monitoring technology, specifically a low-altitude aircraft chip security monitoring system. Background Technology

[0002] With the widespread application of low-altitude aircraft in numerous fields such as logistics transportation, environmental monitoring, and urban management, the safe operation of their core chips is of paramount importance. As the "brain" of low-altitude aircraft, any malfunction or safety issue with the chip could lead to loss of control, mission failure, or even safety accidents, causing significant economic losses and safety hazards.

[0003] Currently, low-altitude aircraft chips face complex operating environments, such as drastic temperature changes, unstable voltage fluctuations, and interference during communication. These factors can all affect the normal operation of the chips. Existing chip safety monitoring systems have many shortcomings when dealing with these complex situations. For example, in terms of status acquisition, traditional systems may not be able to comprehensively and accurately collect chip operating parameters and environmental interference values, making it difficult to analyze the stability of parameters and differences in interference, thus failing to detect potential problems in chip operation in a timely manner.

[0004] In the risk assessment phase, existing systems may be unable to effectively extract the characteristic combinations of chip anomalies and protection, making it difficult to select the optimal combination of anomalies and protection, resulting in low accuracy and reliability of risk assessment. Anomaly detection modules may also have shortcomings, failing to effectively extract the sample changes of the current detection model based on the risk assessment parameter set, and unable to deeply analyze the relationship between anomalies and protection, leading to low efficiency and accuracy in anomaly detection.

[0005] In terms of strategy generation, traditional systems may not be able to map features to strategies by combining real-time runtime data, and the generated strategies may not be able to accurately cope with various situations during chip operation. During instruction execution, the capture and analysis of feature data at chip execution points are not precise enough, making it difficult to effectively classify, organize, and adjust strategy features based on the corresponding feature change trends generated by the strategy.

[0006] In the effect verification and threshold setting stages, the existing system may not be able to accurately analyze error values ​​through real-time features and execution data, thus failing to adjust the generation rules and resulting in unreasonable monitoring threshold settings. Regarding data output and feedback optimization, there are also issues such as insufficiently standardized data formatting, in-depth analysis of feedback information, and untimely adjustments to model training parameters and parameter generation rules, affecting the overall system optimization and performance improvement.

[0007] Furthermore, existing low-altitude aircraft chip safety monitoring systems suffer from weak inter-module collaboration, failing to form a complete and efficient monitoring system and thus failing to meet the high requirements for low-altitude aircraft chip safety monitoring. Therefore, there is an urgent need for a system capable of comprehensive, accurate, and efficient safety monitoring of low-altitude aircraft chips to address the numerous problems in existing technologies and ensure the safe and stable operation of low-altitude aircraft. Summary of the Invention

[0008] The purpose of this invention is to provide a chip safety monitoring system for low-altitude aircraft to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a low-altitude aircraft chip safety monitoring system, the system comprising: The status acquisition module analyzes the parameter stability and interference differences based on the chip operating parameter acquisition range and environmental interference values ​​of the low-altitude aircraft, calculates the perception differences between states, integrates them into a status acquisition parameter set, and obtains the status acquisition status values. The risk assessment module, based on the status values ​​collected, extracts the feature combination of chip anomalies and protection, selects the optimal anomaly and protection combination, and obtains the risk assessment parameter set. The anomaly detection module, based on the risk assessment parameter set, extracts the sample change of the current detection model, analyzes the relationship between anomalies and protection, matches sample features with assessment combinations, and obtains the anomaly detection parameter set. The strategy generation module extracts the feature values ​​of the current monitoring strategy based on the anomaly detection parameter set, combines the real-time running data to map the features to the strategy, sets generation rules, and applies the rules to the matching of state features to obtain the strategy generation threshold. The instruction execution module generates a threshold based on the strategy, captures the feature data of the chip execution point, performs feature inference on the feature changes of the execution point in combination with the operating environment, analyzes the corresponding feature change trend generated by the strategy, classifies and organizes the feature change trend according to the inference result, and performs strategy feature adjustment and analysis on the classified data in combination with the feature change information to obtain the instruction execution analysis value. The effect verification module, based on the instruction execution analysis value, performs error value analysis of features and strategies through real-time features and execution data, and combines the error value to adjust the generation rules to obtain a monitoring effect verification scheme; The threshold setting module, based on the monitoring effect verification scheme, extracts the verified strategy parameters, analyzes the parameter fluctuation range, sets the monitoring threshold interval and applies it to the instruction execution analysis to obtain the monitoring threshold parameter set. The data output module integrates the monitoring results based on the monitoring threshold parameter set, performs data formatting according to the chip security type, and generates standardized output data. The feedback optimization module, based on the standardized output data, collects system operation feedback information, analyzes and monitors the sources of error, adjusts model training parameters and parameter generation rules, and obtains a set of optimized system parameters. The system control module generates a safety monitoring and control scheme for low-altitude aircraft chips based on the system optimization parameter set.

[0010] Preferably, the status acquisition status values ​​include a temperature parameter set, a voltage parameter set, and a communication parameter set; the risk assessment parameter set includes anomaly characteristic parameters, protection characteristic parameters, and combined screening parameters; the anomaly detection parameter set includes sample change parameters, feature matching parameters, and training adjustment parameters; the strategy generation threshold includes feature change parameters, status matching parameters, and rule setting parameters; the instruction execution analysis values ​​include feature trend analysis parameters and strategy feature relationship parameters; the monitoring effect verification scheme includes error analysis parameters and rule adjustment parameters; the monitoring threshold parameter set includes parameter fluctuation parameters and interval setting parameters; the standardized output data includes type integration parameters and format processing parameters; the system optimization parameter set includes feedback analysis parameters and model adjustment parameters; and the low-altitude aircraft chip safety monitoring and control scheme includes configuration synchronization parameters and process coordination parameters.

[0011] Preferably, the status acquisition module includes: The data acquisition submodule, based on the acquisition range of the low-altitude aircraft chip operating parameters and environmental interference values, performs regional acquisition of stability values ​​and interference values, locates invalid state data, removes abnormal interference data, and arranges the extracted stability values ​​and interference values ​​in regional order to generate a state parameter interference dataset. The difference analysis submodule analyzes the stability and interference between states based on the state parameter interference dataset, calculates the state parameter change ratio, sorts the difference values ​​between states by weight, marks areas with excessive fluctuation differences, and obtains state acquisition difference data. The distributed integration submodule, based on the state acquisition difference data, calls the state perception difference value for multi-dimensional aggregation, screens the differences in state acquisition values, classifies them according to the size of the values, and arranges the state perception values ​​in an orderly manner to generate the state acquisition status value.

[0012] Preferably, the risk assessment module includes: The parameter extraction submodule identifies the abnormal characteristics of each chip region based on the state acquisition state value, records the abnormal and protection data of the chip, normalizes the recorded data, sorts the normalized data into abnormal and protection categories, and generates a feature parameter dataset. The feature optimization submodule analyzes the abnormal and protection values ​​in the feature parameter dataset, filters parameter combinations with high matching degree with the collection status, adjusts the parameter combinations and generates parameter combination optimization results by feature matching and recording the matching results. The parameter selection submodule retrieves the optimization results of the parameter combination, determines the optimal combination of anomaly and protection, adjusts the parameter generation parameters, inputs the generation configuration, verifies the stability of the parameter set, and generates a risk assessment parameter set.

[0013] Preferably, the anomaly detection module includes: The sample analysis submodule, based on the risk assessment parameter set, collects sample data through the training library, including temperature anomalies, voltage fluctuations, and communication interruptions. It performs time series analysis on the data, removes abnormal samples, partitions the remaining data, and obtains sample analysis data. The parameter matching submodule analyzes the impact of sample variables on anomaly extraction and protection extraction using the sample analysis data, calculates the degree of influence of changes in each sample feature on parameter adjustment, determines the optimal matching parameter settings based on the impact score, iteratively adjusts parameters to capture the optimal combination, and obtains the parameter matching results. The training integration submodule selects anomaly and protection combinations that match the current sample conditions from the parameter docking results, conducts parameter training experiments, optimizes parameter settings through multiple training and verifications, determines and solidifies parameters as model standards, and generates anomaly detection parameter sets.

[0014] Preferably, the strategy generation module includes: The feature extraction submodule, based on the anomaly detection parameter set, locates the strategy feature monitoring points, extracts the feature change values ​​in the monitoring area, continuously records the feature increase / decrease rates, extracts multiple key change nodes corresponding to the change rates, sorts the node values ​​in order, and obtains the current feature change feature value. The state matching submodule analyzes the node change value and real-time running data based on the current feature change feature value, performs calibration according to a predetermined matching criterion, calls the matching criterion to perform feature redistribution within the state interval, and obtains the feature state matching structure. The rule allocation submodule, based on the feature state matching structure, adopts a dynamic adjustment method for policy rules to calculate the distribution of features among changing nodes, sets upper and lower limits for node rules, applies rules to regional feature values, allocates them, and obtains the policy generation threshold.

[0015] Preferably, the method for dynamically adjusting the strategy rules includes obtaining the allocation rule value of the matching feature change node, and setting the lower limit of the node rule through feature weight, state weight and dynamic adjustment weight to control the minimum matching.

[0016] Preferably, the instruction execution module includes: The feature data capture submodule generates a threshold based on the strategy, applies runtime environment algorithms to capture execution point feature data, removes outliers and corrects errors, stores the data in a hierarchical manner according to intervals, performs feature processing, and generates a feature data dataset. The strategy analysis submodule, based on the feature-based dataset, divides intervals according to strategy categories, extracts trend and fluctuation features, and generates a set of strategy categories and feature changes. The classification inference submodule, based on the strategy category and feature change feature set, adjusts feature parameters and calibrates trend data, extracts classification intervals, and performs numerical prediction to obtain instruction execution analysis values.

[0017] Preferably, the runtime environment algorithm includes calculating feature eigenvalues, generating a eigendata dataset, and comprehensively calculating the eigenvalues ​​after eigenvalues ​​by combining data point weights, original feature values ​​of execution points, environmental values, and category coefficients.

[0018] Preferably, the feedback optimization module includes: The data analysis submodule extracts real-time features and execution data based on the standardized output data, analyzes the real-time monitoring values ​​and execution values, and correlates the monitoring difference with the current feature information to generate feature monitoring error values. The parameter tuning submodule sets the training tuning parameters of the model based on the feature monitoring error value. It sets the adjustment range for policy categories with large errors, fine-tunes for categories with low errors, and filters and integrates matching parameter sets by comparing the tuning effects to generate the model tuning parameter set. The optimization control submodule adjusts the parameter set based on the model, applies the adjusted parameters in each training stage, implements model optimization item by item, monitors features and monitoring simultaneously, gradually adjusts the optimization order of the model for each strategy category, and generates a system optimization parameter set.

[0019] Compared with the prior art, the beneficial effects of the present invention are: In terms of status acquisition, the status acquisition module analyzes the stability of parameters and the difference in interference based on the range of operating parameters and environmental interference values, calculates the perceived differences between states and integrates them into a parameter set. It can comprehensively and accurately obtain the chip's operating status. Key parameters such as temperature, voltage and communication can be accurately acquired, laying a solid foundation for subsequent monitoring work.

[0020] The risk assessment module extracts combinations of anomalies and protection features based on the acquired status values ​​and selects the optimal combination. This accurately identifies potential chip risks, improves the accuracy of risk assessment, and allows staff to be aware of potential chip problems in advance. The anomaly detection module extracts sample changes based on the risk assessment parameter set, analyzes the relationship between anomalies and protection, and matches features with assessment combinations. This enables timely and accurate detection of chip anomalies, preventing further deterioration.

[0021] The strategy generation module combines anomaly detection parameter sets and real-time operational data to map features to strategies and set generation rules. The generated strategy generation thresholds can formulate appropriate strategies based on the chip's operating status, making the monitoring strategy more targeted and effective. The instruction execution module captures feature data of the chip's execution points, performs feature inference and trend analysis in conjunction with the operating environment, and adjusts and analyzes the strategy features of the categorized data to ensure that the strategy can be effectively executed and that the chip operates normally as expected.

[0022] The effectiveness verification module analyzes error values ​​using real-time features and execution data, adjusts generation rules accordingly, and continuously optimizes the monitoring process to improve the reliability of the monitoring system. The threshold setting module extracts verified strategy parameters, analyzes fluctuation ranges, and sets monitoring threshold intervals, providing clearer standards and making monitoring more scientific and rational.

[0023] The data output module integrates monitoring results and formats them according to security type, generating standardized output data to facilitate data management, analysis, and use, providing strong support for subsequent decision-making. The feedback optimization module collects system operation feedback information, analyzes error sources, and adjusts model training parameters and generation rules, enabling the system to continuously self-optimize and adapt to different operating environments and needs.

[0024] The system control module generates control schemes based on optimized parameter sets, realizing coordinated control of the entire system. This enables all modules to work together to form a complete and efficient monitoring system, comprehensively ensuring the safe operation of low-altitude aircraft chips, reducing chip failure risks, improving the safety and stability of low-altitude aircraft, and reducing accidents and losses caused by chip problems. This has significant economic and safety benefits. Attached Figure Description

[0025] Figure 1 This is a schematic diagram illustrating the working principle of the low-altitude aircraft chip safety monitoring system described in this invention. Figure 2 This is a design diagram of the status acquisition module; Figure 3 Design diagram for the risk assessment module; Figure 4 This is a design diagram of the anomaly detection module; Figure 5Design diagram for the feedback optimization module. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Please see Figures 1-5 This invention relates to a chip-based safety monitoring system for low-altitude aircraft. The system includes: a status acquisition module, a risk assessment module, an anomaly detection module, a strategy generation module, an instruction execution module, an effect verification module, a threshold setting module, a data output module, a feedback optimization module, and a system control module. Specific implementation methods are as follows: The status acquisition module analyzes the parameter stability and interference differences based on the chip operating parameter acquisition range and environmental interference values ​​of the low-altitude aircraft, calculates the perception differences between states, integrates them into a status acquisition parameter set, and obtains the status acquisition status values. The risk assessment module, based on the status values ​​collected, extracts the feature combination of chip anomalies and protection, selects the optimal anomaly and protection combination, and obtains the risk assessment parameter set. The anomaly detection module, based on the risk assessment parameter set, extracts the sample change of the current detection model, analyzes the relationship between anomalies and protection, matches sample features with assessment combinations, and obtains the anomaly detection parameter set. The strategy generation module extracts the feature values ​​of the current monitoring strategy based on the anomaly detection parameter set, combines the real-time running data to map the features to the strategy, sets generation rules, and applies the rules to the matching of state features to obtain the strategy generation threshold. The instruction execution module generates a threshold based on the strategy, captures the feature data of the chip execution point, performs feature inference on the feature changes of the execution point in combination with the operating environment, analyzes the corresponding feature change trend generated by the strategy, classifies and organizes the feature change trend according to the inference result, and performs strategy feature adjustment and analysis on the classified data in combination with the feature change information to obtain the instruction execution analysis value. The effect verification module, based on the instruction execution analysis value, performs error value analysis of features and strategies through real-time features and execution data, and combines the error value to adjust the generation rules to obtain a monitoring effect verification scheme; The threshold setting module, based on the monitoring effect verification scheme, extracts the verified strategy parameters, analyzes the parameter fluctuation range, sets the monitoring threshold interval and applies it to the instruction execution analysis to obtain the monitoring threshold parameter set. The data output module integrates the monitoring results based on the monitoring threshold parameter set, performs data formatting according to the chip security type, and generates standardized output data. The feedback optimization module, based on the standardized output data, collects system operation feedback information, analyzes and monitors the sources of error, adjusts model training parameters and parameter generation rules, and obtains a set of optimized system parameters. The system control module generates a safety monitoring and control scheme for low-altitude aircraft chips based on the system optimization parameter set.

[0028] Example 1: In this embodiment, the specific structure and working process of the status acquisition module are as follows: The status acquisition module includes a data acquisition submodule, a difference analysis submodule, and a distribution integration submodule. Each submodule cooperates with each other to complete the acquisition and analysis of the operating status of the low-altitude aircraft chip in order to generate accurate status acquisition values.

[0029] When the data acquisition submodule is working, it first needs to define the acquisition range of the low-altitude aircraft chip's operating parameters. This range covers various key parameters that may be involved in the chip's normal operation, such as temperature, voltage, and communication-related parameters. Simultaneously, environmental interference values ​​must be considered, as various interference factors in the chip's operating environment, such as electromagnetic interference and temperature fluctuations, can affect the acquired parameters. Based on these acquisition ranges and environmental interference values, the data acquisition submodule begins regional acquisition of stability and interference values. During the acquisition process, the acquired data needs preliminary processing. First, invalid state data is identified. This invalid data may be caused by acquisition equipment failure, communication interruption, or other reasons, and cannot accurately reflect the chip's operating status. After identifying invalid data, it is removed to avoid interfering with subsequent analysis. Next, the remaining valid data is processed to remove abnormal interference data. This abnormal interference data may be caused by sudden strong interference, and its values ​​deviate significantly from the normal range. After removing invalid and abnormal interference data, the extracted stability and interference values ​​are arranged in regional order. This regional order can be based on the distribution areas of different functional modules within the chip, or on the physical location distribution of the data collection points. This ordered arrangement generates a state parameter interference dataset. This dataset contains the pre-processed stability and interference data, providing a foundation for subsequent analysis.

[0030] The difference analysis submodule operates based on the state parameter interference dataset generated by the data acquisition submodule. First, it conducts an in-depth analysis of the stability and interference between states by comparing data from different regions and time periods to understand the changes in stability and interference values. Then, it calculates the state parameter change ratio, which reflects the magnitude and rate of parameter change between different states. The difference values ​​between states are then sorted by weight, with the weights determined based on factors such as the importance of the parameters and their impact on chip operation. This weighted sorting highlights important differences for focused attention later. During the sorting process, areas with excessively large fluctuations are marked, as these areas may contain potential faults or anomalies requiring further analysis. Through these operations, the difference analysis submodule acquires state acquisition difference data, reflecting the differences in chip operating states across various aspects, providing a basis for subsequent integration and processing.

[0031] The distributed integration submodule performs multi-dimensional aggregation processing on the status acquisition difference data obtained by the difference analysis submodule. First, it retrieves the status-aware difference values, which may originate from different sensing dimensions, such as time and space. Aggregating these multi-dimensional difference values ​​provides a more comprehensive understanding of the overall differences in the chip's operating status. During the aggregation process, it screens for differences in the status acquisition values, checking and analyzing the differences across various dimensions to identify any problems or anomalies. Then, it categorizes the differences by magnitude, dividing them into different intervals or categories for targeted processing. Simultaneously, the status-aware values ​​are arranged in an ordered manner, such as from smallest to largest, largest to smallest, or other reasonable sequences, making the data clearer and more organized. Through these operations, the distributed integration submodule ultimately generates the status acquisition status values. These status acquisition status values ​​include multiple key parameter sets, such as temperature, voltage, and communication parameters. These parameter sets comprehensively and accurately reflect the chip's operating status, providing reliable data support for subsequent risk assessment and anomaly detection modules.

[0032] Throughout the entire status acquisition module's operation, the various sub-modules work closely together. The data acquisition sub-module provides the foundational data for subsequent analysis, the difference analysis sub-module delves into the discrepancies within the data, and the distribution integration sub-module integrates and categorizes these discrepancies, ultimately generating accurate status acquisition values. Each step undergoes rigorous data processing and analysis to ensure that the acquired information accurately and effectively reflects the operational status of the low-altitude aircraft chip, laying a solid foundation for the normal operation of the entire chip safety monitoring system. This comprehensive and meticulous status acquisition process enables the timely detection of potential anomalies during chip operation, providing strong support for subsequent risk assessment and fault handling, and ensuring the safe and stable operation of the low-altitude aircraft chip.

[0033] Example 2: In this embodiment, the risk assessment module consists of a parameter extraction submodule, a feature optimization submodule, and a parameter selection submodule. Each submodule uses hierarchical data processing logic to accurately extract and optimize the screening of chip anomalies and protection features, and finally generates a risk assessment parameter set for subsequent detection.

[0034] The parameter extraction submodule begins by receiving state values ​​from the status acquisition system. These values ​​include multi-dimensional data such as temperature, voltage, and communication parameters. This submodule identifies abnormal features in each chip region. The region division can be based on standards such as chip physical layout, functional modules, or signal paths; for example, the processor core area, storage module area, and communication interface area can be treated as independent regions. During identification, anomaly patterns are matched using a preset feature library, such as temperature parameters continuously exceeding threshold ranges, voltage fluctuations exceeding safe ranges, and abnormally high communication data packet error rates. The corresponding anomaly occurrence time, location, and specific parameter fluctuation values ​​are recorded. For recording protection features, the submodule covers the chip's built-in protection mechanism response data, such as frequency reduction strategies triggered by over-temperature protection, power switching logic during voltage anomalies, and reconnection protocols during communication interruptions. These data correspond to the anomaly features, collectively forming the original recorded data.

[0035] The parameter extraction submodule normalizes the recorded data, taking into account the dimensional differences of different parameters. Taking temperature (°C) and voltage (V) parameters as examples, linear transformations or standardization algorithms are used to map them to a unified numerical range, eliminating the influence of dimensions on subsequent analysis. After normalization, the data is categorized and organized according to anomaly and protection features. For example, all data related to temperature anomalies are grouped into the "Temperature Anomaly Feature Set," and the protection strategies corresponding to over-temperature protection are grouped into the "Temperature Protection Feature Set," and so on, generating a feature parameter dataset. This dataset is stored in a structured table format, containing fields such as timestamp, region identifier, anomaly type, protection measures, original parameter values, and normalized values.

[0036] The feature optimization submodule analyzes the feature parameter dataset, with the core objective of selecting parameter combinations that highly match the current data acquisition status. First, it performs statistical analysis on the anomalies and protection values ​​in the dataset, such as calculating the frequency of occurrence of various anomalies, the response time distribution of protection measures, and the correlation coefficient of parameter fluctuations. Taking voltage fluctuation anomalies as an example, analysis reveals that when the voltage is below threshold 1, power backup switching is triggered in 80% of cases, while when the voltage is below threshold 2, switching is triggered only in 50% of cases, accompanied by processor frequency reduction. This correlation analysis reveals the potential link between anomalies and protection measures.

[0037] When selecting parameter combinations with high matching scores, a multi-dimensional matching algorithm is employed. This algorithm comprehensively considers factors such as the severity of abnormal features, the effectiveness of protective measures, and the consistency of parameter changes over time. For example, for communication interruption anomalies, abnormal features such as "interruption duration," "reconnection success rate," and "communication rate recovery time" are combined with protective features such as "link detection cycle," "number of reconnection attempts," and "bit error rate threshold." By setting matching score rules (such as weighting percentages and threshold ranges), each parameter combination is scored, and the scoring results are recorded to form a preliminary list of optimized parameter combinations. For combinations with low scores, iterative optimization is performed by adjusting the parameter range or replacing relevant features. For example, the communication interruption time threshold is adjusted from 50ms to 30ms, and the matching score is recalculated until multiple sets of highly matched optimized parameter combinations are generated.

[0038] The parameter selection submodule determines the optimal anomaly and protection combination from the parameter combination optimization results output by the feature optimization submodule. This submodule first establishes a retrieval mechanism that allows for rapid searching based on keywords such as anomaly type, region location, and parameter importance level; for example, it prioritizes searching for combinations related to temperature anomalies in the processor core area. During the retrieval process, the stability of each parameter combination is verified through historical data backtesting and simulation testing. Historical data backtesting involves applying the parameter combination to chip operating data over a past period to check its accuracy in anomaly identification and the matching degree of protection measures; simulation testing involves injecting specific anomalies into a virtual environment and observing the response effect of the parameter combination.

[0039] When verifying the stability of the parameter set, the focus is on the robustness of the parameter combination under different operating conditions, such as the difference in performance between the chip running under full load and light load. By adjusting the parameter generation parameters (such as upper and lower thresholds, weighting coefficients, etc.), different generation configurations are input for comparative testing. Finally, a set of parameter combinations that can maintain high matching degree and stability under various operating conditions is selected. After solidification, this combination generates a risk assessment parameter set, which includes anomaly feature parameters (such as the judgment threshold and feature vector of various types of anomalies), protection feature parameters (such as the triggering conditions and execution logic of protection strategies), and combination screening parameters (such as matching degree scoring rules and parameter priorities). These parameters provide accurate evaluation basis for the subsequent anomaly detection module, ensuring that the system can accurately identify chip operation risks and activate corresponding protection mechanisms.

[0040] Example 3: In this embodiment, the specific details of the anomaly detection module are as follows: The anomaly detection module includes a sample analysis submodule, a parameter matching submodule, and a training integration submodule. Each submodule achieves accurate identification and parameter optimization of chip anomaly patterns through systematic processing of sample data, and finally generates an anomaly detection parameter set for real-time detection.

[0041] The sample analysis submodule begins by receiving the risk assessment parameter set, which contains structured data such as anomaly characteristic parameters and protection characteristic parameters. The submodule collects sample data through a training library interface. The training library stores various anomaly samples recorded during historical operation, including typical anomalies such as temperature anomalies (e.g., sustained high temperatures, sudden temperature rises and falls), voltage fluctuations (e.g., voltage exceeding the rated range, instantaneous spikes), and communication interruptions (e.g., data packet loss, link connection failure). For each type of sample, the timestamp of the anomaly, the chip region, the accompanying parameter fluctuation value, and the corresponding protective measure response data are recorded synchronously, forming a complete sample record.

[0042] After acquiring sample data, the submodule performs time-series analysis. Taking temperature anomaly samples as an example, the temperature parameters of chips in the same region are arranged in chronological order, and temperature change curves are plotted. A sliding window algorithm is used to detect outliers in the curves (such as values ​​exceeding three standard deviations). For voltage fluctuation samples, the frequency and amplitude of voltage fluctuations over time are analyzed to identify non-periodic abnormal fluctuation patterns. During the analysis, abnormal samples caused by uncontrollable factors such as hardware failures and abnormal acquisition equipment are removed to ensure that the remaining samples accurately reflect typical abnormal conditions during chip operation.

[0043] After removing invalid samples, the remaining data is partitioned. The partitioning criteria can be based on anomaly type, chip region, or parameter fluctuation range. For example, temperature anomaly samples can be partitioned according to different functional areas of the chip (processor area, storage area), and each partition can be further subdivided according to the severity of the temperature anomaly (e.g., slight overheating, severe overheating). Finally, sample analysis data containing time series characteristics, regional characteristics, and anomaly level characteristics are obtained. This data is stored in a multidimensional table format to facilitate subsequent parameter matching processing.

[0044] The parameter matching submodule, based on sample analysis data, delves into the impact of sample variables on anomaly detection and protection detection. Taking voltage fluctuation samples as an example, sample variables include the start time, amplitude, and duration of voltage fluctuations. Anomaly detection focuses on accurately identifying voltage anomalies using these variables, while protection detection focuses on triggering appropriate protection strategies based on variable characteristics. The submodule calculates the degree of impact of changes in each sample feature on parameter adjustments. For example, when the voltage fluctuation amplitude increases by 10%, the corresponding anomaly detection threshold needs adjustment, or the response time of the protection strategy needs to be shortened.

[0045] The impact level is calculated using a feature importance assessment method. The weight of a feature is determined by comparing the difference in anomaly identification accuracy when a feature is included versus when it is removed. The optimal parameter settings are determined based on the impact score. The scoring rules comprehensively consider factors such as identification accuracy, the effectiveness of protection strategies, and parameter stability. For example, for communication interruption anomalies, parameters such as the "interruption detection time threshold" and "reconnection attempt interval" are combined and adjusted. A score is calculated after each adjustment. By iteratively adjusting parameters (e.g., adjusting the detection time threshold in 5ms increments and the reconnection interval in 100ms increments), the parameter combination that yields the highest score is captured, and the parameter matching result is obtained. This result includes the optimal parameter combination scheme for different anomaly types.

[0046] The training ensemble submodule selects anomaly and protection combinations that match the current sample conditions from the parameter matching results output by the parameter matching submodule. Selection criteria include the anomaly type, regional characteristics, and severity of the sample. For example, for a severe temperature anomaly sample in the processor area, the parameter combination specific to that region and anomaly level is selected from the parameter matching results. After selection, training experiments are conducted on the parameter combinations. The training process uses cross-validation, dividing the sample data into training and testing sets. Parameters are optimized on the training set, and the optimization effect is verified on the testing set.

[0047] Through multiple training and validation iterations, parameter settings were optimized. For example, during the initial training, it was found that the temperature anomaly detection threshold was set too low, resulting in a high false alarm rate. Therefore, the threshold was increased by 5°C, and training was repeated to verify whether the false alarm rate decreased while the false alarm rate remained within an acceptable range. During training, the performance of parameter combinations on different sample subsets was monitored to ensure their generalization ability and avoid overfitting. After multiple rounds of training and validation, the parameters were determined and solidified as the model standard, generating an anomaly detection parameter set.

[0048] This parameter set includes sample variation parameters (such as the range of sample feature changes for different anomaly types), feature matching parameters (such as the mapping rules between anomaly features and protection features), and training adjustment parameters (such as the number of training iterations and the learning rate). These parameters provide a standardized detection model for the anomaly detection module, enabling the system to accurately identify anomalies and match corresponding protection strategies based on real-time acquired chip operating data, ensuring the safe operation of the low-altitude aircraft chip under various operating conditions. Throughout the process, each submodule ensures the accuracy and adaptability of the anomaly detection model through refined processing of sample data and iterative optimization of parameters, providing a reliable basis for subsequent strategy generation and instruction execution.

[0049] Example 4: In this embodiment, the specific workflow of the strategy generation module is as follows: The strategy generation module includes a feature extraction submodule, a state matching submodule, and a rule allocation submodule. Each submodule achieves accurate setting of chip monitoring strategy through in-depth analysis of anomaly detection parameters and dynamic rule generation, and finally forms a strategy generation threshold that can be applied to real-time monitoring.

[0050] The feature extraction submodule operates based on an anomaly detection parameter set, which includes data such as sample change parameters and feature matching parameters. The submodule first locates policy feature monitoring points. These monitoring points are selected based on key operating parameter nodes of the chip. For example, in temperature parameters, monitoring points such as processor core temperature and storage module temperature are selected; in voltage parameters, monitoring points such as power input voltage and chip core voltage are selected. Taking a certain type of low-altitude aircraft chip as an example, its processor core temperature monitoring point is located at the thermistor inside the chip package, and the voltage monitoring point is set at the output of the power management module.

[0051] After locating the monitoring point, the submodule extracts the characteristic change values ​​in the monitoring area. Taking a temperature monitoring point as an example, the temperature value of that point is collected in real time and compared with historical data to calculate the rate of temperature change. The feature acceleration / deceleration rate is continuously recorded, for example, the temperature change value is recorded every 100 milliseconds, forming a time-series rate data. During the recording process, multiple key change nodes corresponding to the rate of change are extracted. These nodes include inflection points where the rate suddenly increases, and turning points where the rate changes from positive to negative. Taking a scenario of rapid temperature increase as an example, when the temperature rises from 50℃ to 65℃ within 200 milliseconds, the rate of change reaches 0.075℃ / ms, exceeding the normal operating rate of 0.02℃ / ms. At this point, the point is marked as a key change node. The node values ​​are sorted in chronological order to obtain the current characteristic change feature values. These feature values ​​are stored in array form, containing information such as node time, rate of change, and parameter type.

[0052] The state matching submodule analyzes node changes and real-time operational data based on the current characteristic change value. Taking a voltage monitoring point as an example, when the voltage drops from 3.3V to 2.8V within 100 milliseconds (a rate of change of -0.005V / ms), and real-time operational data shows a sudden increase in processor load from 30% to 80%, the submodule needs to analyze the correlation between these two data points to determine whether the voltage drop is caused by the sudden load increase. After the analysis, calibration is performed according to predetermined matching criteria. These criteria are based on the chip's design specifications and historical operational data. For example, it is stipulated that when the voltage drop rate exceeds 0.003V / ms and the load increase exceeds 50%, the voltage change is considered to be caused by normal load fluctuations.

[0053] The matching criteria are invoked to perform feature redistribution within the state interval. For example, in the voltage anomaly state interval, the voltage drop feature caused by load fluctuations is redistributed to the normal operation state category, while the voltage drop feature caused by power module failure is retained in the anomaly state category. Through this redistribution, a feature state matching structure is obtained, which is presented in the form of a tree directory. The root node is the state category (normal / abnormal), and the child nodes are the matching status of each feature parameter. For example, the "voltage anomaly - power module failure" node includes sub-features such as voltage change rate, load rate change, and power module temperature.

[0054] The rule allocation submodule generates policy rules based on a feature-state matching structure and employs a dynamic adjustment method. Taking abnormal communication parameters as an example, when the communication data packet error rate rises from 0.1% to 1% within 500 milliseconds and the number of reconnections reaches 3, dynamic adjustment of policy rules is triggered. This method includes obtaining the allocation rule value of the matching feature change node. The determination of the rule value comprehensively considers feature weight, state weight, and dynamic adjustment weight. Feature weights are set according to the importance of the parameter; for example, the weight of the communication error rate is higher than the number of reconnections. State weights are set according to the urgency of the current operating state; for example, the state weight in flight mode is higher than that in standby mode. Dynamic adjustment weights are set according to the magnitude of changes in real-time data; for example, the dynamic adjustment weight increases when the error rate rises sharply.

[0055] To control minimum matching, node rule lower limits are set using feature weights, state weights, and dynamically adjusted weights. For example, a lower limit of 0.5% for the communication error rate rule is set; if the error rate does not reach this lower limit, the corresponding strategy is not triggered. The distribution of features across changing nodes is measured. For instance, analyzing the error rate distribution at each time point as the communication error rate increases from 0.1% to 1% helps determine the slope and peak point of the error rate increase. Upper and lower limits are set for node rules. For example, setting the upper limit for the communication error rate rule to 1.5% triggers the highest level of protection strategy when this limit is reached, and setting the lower limit to 0.5% triggers the basic protection strategy when this lower limit is reached.

[0056] Rules are applied to regional feature values ​​and assigned accordingly. For example, in the communication module region, error rate rules are applied to all communication interfaces in that region, and different policy levels are assigned based on the error rate of each interface. Taking interface A with an error rate of 1.2% and interface B with an error rate of 0.6% as an example, interface A triggers a medium-level protection policy (such as reducing the communication rate), and interface B triggers a basic protection policy (such as adding a check bit), ultimately obtaining the policy generation threshold. This threshold includes feature change parameters (such as the change rate threshold and amplitude threshold for each parameter), state matching parameters (such as the state category determination threshold and feature redistribution threshold), and rule setting parameters (such as weight coefficients and rule upper and lower limits). These thresholds provide clear policy execution basis for subsequent instruction execution modules, enabling the system to dynamically generate appropriate monitoring policies based on the chip's real-time operating status, ensuring the safe and stable operation of the low-altitude aircraft chip.

[0057] Throughout the strategy generation process, the submodules collaborate closely. The feature extraction submodule provides accurate feature change data for state matching, the state matching submodule ensures the correct correspondence between features and states through rule calibration and reallocation, and the rule allocation submodule generates flexible and scalable strategy rules based on a dynamic adjustment method. The final strategy generation threshold can adapt to various complex operating conditions during chip operation, achieving precise monitoring of chip safety. For example, when a low-altitude aircraft traverses a region with strong electromagnetic interference, the system dynamically adjusts the monitoring strategy for communication parameters through the strategy generation module, increasing the error checking frequency and reducing the communication rate, thereby ensuring the stability of the communication link and preventing chip communication abnormalities caused by electromagnetic interference from affecting the normal flight of the aircraft.

[0058] Example 5: In this embodiment, the specific workflow of the instruction execution module is as follows: The instruction execution module includes a feature data capture submodule, a strategy analysis submodule, and a classification inference submodule. Each submodule achieves dynamic monitoring and strategy adjustment of the chip's execution instructions through real-time response to the strategy generation threshold and feature analysis, and finally forms instruction execution analysis values ​​for feedback optimization.

[0059] The feature data capture submodule operates based on a strategy-generated threshold, which includes rule data such as feature change parameters and state matching parameters. The submodule applies a runtime environment algorithm to capture execution point feature data. Taking the power management module of a low-altitude aircraft chip as an example, the execution point feature data includes the power input voltage value, the chip core voltage value, and the power module temperature value. The runtime environment algorithm generates characteristic data through comprehensive calculation. The specific process is as follows: First, the weights of the data points are determined. The power input voltage, which directly affects the chip's operational stability, is weighted at 0.4, the core voltage at 0.3, and the temperature at 0.3. Then, the original feature values ​​of the execution points are obtained, such as a power input voltage of 3.2V, a core voltage of 1.8V, and a temperature of 60℃ at a certain moment. Next, environmental values ​​are considered, such as the current ambient temperature of the aircraft being 25℃ and the electromagnetic interference level being moderate. Finally, the characteristic values ​​are calculated by combining the category coefficient (the category coefficient for the power module is 1.2).

[0060] During data capture, outliers are removed and errors are corrected. For example, if the power input voltage is 1.5V at a certain moment, which is significantly lower than the normal range (3.0V-3.6V), it is identified as an outlier and removed. For minor errors caused by voltage fluctuations, a moving average filtering algorithm is used for correction. Data is stored in interval layers, with the interval division determined based on the importance and frequency of change of the parameters. For example, voltage parameters are divided into interval layers of 0.1V, and temperature parameters are divided into interval layers of 5℃. After storage, feature processing is performed to generate a feature dataset. This dataset is stored in the form of a time-series database and includes fields such as timestamps, original values, feature values, and outlier markers.

[0061] The strategy analysis submodule divides intervals based on a feature-based dataset, categorizing them by strategy type. Strategy categories are determined by chip functional modules and anomaly types, such as power management strategies, communication strategies, and processor strategies. Taking power management strategies as an example, it divides the intervals into overvoltage protection, undervoltage protection, and overtemperature protection intervals. It extracts trends and fluctuation characteristics. Taking voltage parameter changes within the overvoltage protection interval as an example, it analyzes the rate of increase, duration, and fluctuation amplitude of voltage rising from 3.6V to 3.8V, generating a strategy category and feature set. This feature set includes strategy category identifiers, interval ranges, trend curves, and fluctuation parameters.

[0062] The classification inference submodule adjusts feature parameters and calibrates trend data based on strategy categories and feature change feature sets. Taking a voltage overvoltage scenario as an example, when the characteristic data shows that the voltage remains in the 3.7V-3.8V range (overvoltage protection range is above 3.6V), the overvoltage feature parameters are adjusted, updating the voltage rise rate from 0.05V / ms to 0.08V / ms, and the trend curve is calibrated to predict that the voltage may rise to 4.0V within the next 500 milliseconds. Classification intervals are extracted, and based on the adjusted parameters, the overvoltage risk level is divided into primary (3.6V-3.8V), intermediate (3.8V-4.0V), and advanced (above 4.0V). Numerical predictions are performed for each interval. For example, if the voltage in the primary interval maintains the current rate of increase, the intermediate interval may trigger frequency reduction protection, and the advanced interval may trigger power cut-off protection. Finally, the instruction execution analysis value is obtained, which includes feature trend analysis parameters (such as the rate of change and predicted values ​​of each parameter) and strategy feature relationship parameters (such as the correlation between strategy triggering conditions and feature changes).

[0063] The feedback optimization module operates based on standardized output data. The data analysis submodule extracts real-time features and execution data. Taking a power supply overvoltage scenario as an example, the real-time feature is the voltage characteristic value of 3.75V, and the execution data is the trigger time and frequency reduction amplitude of the frequency reduction strategy. The real-time monitoring value and execution value are analyzed, and the monitoring difference is calculated (e.g., if the overvoltage threshold is set to 3.6V, and the current value is 3.75V, the difference is 0.15V). The monitoring difference is then correlated with the current feature information (voltage rise rate 0.08V / ms, temperature 65℃) to generate a feature monitoring error value.

[0064] The parameter adjustment submodule sets the training adjustment parameters of the model based on the feature monitoring error value. If the monitoring error in the power supply overvoltage scenario is large (e.g., the voltage exceeds 3.6V multiple times but the frequency reduction strategy is not triggered in time), a larger adjustment range is set for this strategy category, such as lowering the overvoltage threshold from 3.6V to 3.5V and shortening the frequency reduction trigger delay time from 100ms to 50ms; fine-tuning is performed for low error categories (such as communication strategy categories), such as adjusting the communication error rate threshold from 0.5% to 0.45%. By comparing the feature monitoring error values ​​before and after the adjustment, matching parameter sets are selected and integrated to generate the model adjustment parameter set.

[0065] The optimization control submodule applies the adjusted parameters based on the model adjustment parameter set in each training stage. Taking power management strategy model training as an example, it first applies the overvoltage threshold adjustment parameter (3.5V) and the frequency reduction delay adjustment parameter (50ms) to test the accuracy of strategy triggering in a simulated overvoltage scenario. If the triggering time is shortened from the original 120ms to 60ms, close to the target value of 50ms, other parameters are further optimized; if there is still a large error, the frequency reduction amplitude parameter is further adjusted. Model optimization is implemented item by item, and features and monitoring are carried out simultaneously. For example, while adjusting the overvoltage threshold, the impact of temperature features on strategy triggering is monitored to avoid false triggering due to temperature rise. The optimization order of each strategy category model is gradually adjusted, prioritizing the optimization of strategy categories with large errors and then optimizing those with small errors, generating a system optimization parameter set. This parameter set includes feedback analysis parameters (such as error source classification and adjustment effect evaluation) and model adjustment parameters (such as thresholds and delay times for each strategy category).

[0066] The system control module generates a safety monitoring and control scheme for low-altitude aircraft chips based on the system optimization parameter set. Taking power overvoltage risk as an example, the control scheme includes configuring synchronization parameters (such as synchronizing the overvoltage threshold of 3.5V to all power monitoring points) and process coordination parameters (such as setting the execution order of frequency reduction strategy and power backup switching strategy during overvoltage). The effect verification module analyzes the error values ​​of features and strategies based on command execution analysis values, through real-time features and execution data. For example, it compares the difference between the adjusted overvoltage threshold of 3.5V and the real-time voltage characteristic value of 3.7V, and combines the error value to adjust the generation rules. For example, it adjusts the trigger condition of the overvoltage strategy from "voltage exceeds the threshold" to "voltage exceeds the threshold and the rise rate exceeds 0.05V / ms", thus obtaining the monitoring effect verification scheme.

[0067] The threshold setting module, based on the monitoring effect verification scheme, extracts the verified strategy parameters and analyzes the parameter fluctuation range. For example, fluctuations in the overvoltage threshold within the range of 3.5V ± 0.05V are considered normal. The monitoring threshold range is set to 3.45V-3.55V and applied to instruction execution analysis to obtain the monitoring threshold parameter set. The data output module, based on the monitoring threshold parameter set, integrates the monitoring results and formats the data according to chip safety type (power safety, communication safety, processor safety, etc.) to generate standardized output data, providing clear data support for subsequent system optimization and fault analysis.

[0068] Throughout the entire instruction execution module's operation, each submodule collaborates closely with other modules to achieve real-time monitoring and strategy optimization of the chip's operating status. For example, when a low-altitude aircraft encounters a sudden power voltage fluctuation, the system uses a feature data capture submodule to collect voltage feature data in real time, a strategy analysis submodule to quickly identify abnormal voltage ranges, a classification inference submodule to predict voltage change trends and generate execution analysis values, a feedback optimization module to adjust strategy parameters based on the execution results, and finally, a targeted control scheme is generated through the system control module to ensure the chip can still operate safely under voltage fluctuations and guarantee the overall stability of the low-altitude aircraft.

[0069] 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 process, method, article, or apparatus.

[0070] 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. A low altitude vehicle chip security monitoring system, characterized by, The system comprises: a state acquisition module, which acquires a range of low-altitude aircraft chip operating parameters and environmental interference values, analyzes parameter stability and interference differences, calculates inter-state perception differences, integrates state acquisition parameter sets, and obtains state acquisition state values; a risk assessment module, which extracts feature combinations of chip abnormalities and protection based on the state acquisition state values, screens optimal abnormal and protection combinations, and obtains a risk assessment parameter set; an abnormality detection module, which extracts current detection model sample changes based on the risk assessment parameter set, analyzes the relationship between abnormalities and protection, matches sample features and evaluation combinations, and obtains an abnormality detection parameter set; a strategy generation module, which extracts current monitoring strategy feature values based on the abnormality detection parameter set, combines real-time operating data, performs feature and strategy mapping, sets generation rules, and applies the rules to state feature matching, and obtains a strategy generation threshold value; an instruction execution module, which captures chip execution point feature data based on the strategy generation threshold value, combines operating environment feature changes of the execution point, analyzes corresponding feature change trends of the strategy generation, classifies and organizes the feature change trends according to the inference results, adjusts and analyzes the classified data based on the feature change information, and obtains an instruction execution analysis value; an effect verification module, which performs feature and strategy error value analysis based on the instruction execution analysis value through real-time features and execution data, adjusts generation rules based on the error values, and obtains a monitoring effect verification scheme; a threshold setting module, which extracts verification strategy parameters based on the monitoring effect verification scheme, analyzes parameter fluctuation ranges, sets monitoring threshold intervals, and applies them to instruction execution analysis, and obtains a monitoring threshold parameter set; a data output module, which integrates monitoring generation results based on the monitoring threshold parameter set, formats data according to chip security types, and generates standardized output data; a feedback optimization module, which collects system operation feedback information based on the standardized output data, analyzes monitoring error sources, adjusts model training parameters and parameter generation rules, and obtains a system optimization parameter set; a system control module, which generates a low-altitude aircraft chip security monitoring control scheme based on the system optimization parameter set.

2. The low altitude vehicle chip safety monitoring system of claim 1, wherein: The state acquisition state values include temperature parameter sets, voltage parameter sets, and communication parameter sets. The risk assessment parameter set includes abnormal feature parameters, protection feature parameters, and combination screening parameters. The abnormality detection parameter set includes sample change parameters, feature matching parameters, and training adjustment parameters. The strategy generation threshold value includes feature change parameters, state matching parameters, and rule setting parameters. The instruction execution analysis value includes feature trend analysis parameters and strategy feature relationship parameters. The monitoring effect verification scheme includes error analysis parameters and rule adjustment parameters. The monitoring threshold parameter set includes parameter fluctuation parameters and interval setting parameters. The standardized output data includes type integration parameters and format processing parameters. The system optimization parameter set includes feedback analysis parameters and model adjustment parameters. The low-altitude aircraft chip security monitoring control scheme includes configuration synchronization parameters and process coordination parameters.

3. The low altitude vehicle chip safety monitoring system of claim 1, wherein: The state acquisition module comprises: A data acquisition submodule acquires a stability value and an interference value based on a low-altitude aircraft chip operating parameter acquisition range and an environmental interference value, positions invalid state data, removes abnormal interference data, arranges the extracted stability value and interference value in order of region, and generates a state parameter interference data set; A difference analysis submodule analyzes the stability and interference between states based on the state parameter interference data set, calculates a state parameter change ratio, sorts the difference values between states by weight, marks a region with a large fluctuation difference, and acquires state acquisition difference data; A distribution integration submodule calls state perception difference values based on the state acquisition difference data, performs multi-dimensional summarization, screens state acquisition numerical value differences, classifies them by numerical value, and sequentially arranges state perception values to generate a state acquisition state value.

4. The low altitude vehicle chip safety monitoring system of claim 1, wherein: The risk assessment module comprises: A parameter extraction submodule identifies the abnormal features of each regional chip based on the state acquisition state value, records the chip's abnormality and protection data, normalizes the recorded data, classifies the normalized data by abnormality and protection, and generates a feature parameter data set; A feature optimization submodule analyzes the abnormality and protection values in the feature parameter data set, selects a parameter combination with high matching degree with the acquisition state, records the matching result through feature matching, adjusts the parameter combination, and generates a parameter combination optimization result; A parameter selection submodule retrieves the parameter combination optimization result, determines the optimal abnormality and protection combination with the highest matching degree, adjusts the parameters to generate parameters, inputs the generated configuration, verifies the stability of the parameter set, and generates a risk assessment parameter set.

5. The low altitude vehicle chip safety monitoring system of claim 1, wherein: The anomaly detection module comprises: A sample analysis submodule collects sample data including temperature abnormalities, voltage fluctuations, and communication interruptions from a training library based on the risk assessment parameter set, performs time series analysis on the data, removes abnormal samples, partitions the remaining data, and acquires sample analysis data; A parameter matching submodule analyzes the influence of sample variables on abnormality extraction and protection extraction based on the sample analysis data, calculates the influence degree of each sample feature change on parameter adjustment, determines the optimal matching parameter setting based on the influence score, adjusts the parameters to capture the optimal combination, and acquires a parameter docking result; A training integration submodule selects abnormality and protection combinations that match the current sample conditions from the parameter docking result, performs parameter training experiments, optimizes the parameter settings through multiple training and verification, determines and solidifies the parameters as model standards, and generates an anomaly detection parameter set.

6. The low altitude vehicle chip safety monitoring system of claim 1, wherein: The strategy generation module comprises: A feature extraction submodule locates strategy feature monitoring points based on the anomaly detection parameter set, extracts feature change values in the monitoring region, continuously records the feature increase / decrease rate, extracts multiple key change nodes corresponding to the change rate, sorts the node values in order, and acquires the current feature change feature value; The state matching sub-module analyzes the node change value and the real-time running data based on the current characteristic change characteristic value, and calibrates according to a predetermined matching criterion, calls the matching criterion to perform characteristic re-distribution in the state interval, and obtains a characteristic state matching structure. The rule distribution sub-module calculates the distribution of characteristics among the change nodes by using a strategy rule dynamic adjustment method based on the characteristic state matching structure, sets the upper limit and the lower limit of the node rule, applies the rule to the regional characteristic value, and distributes to obtain a strategy generation threshold.

7. The low altitude vehicle chip safety monitoring system of claim 6, wherein: The strategy rule dynamic adjustment method comprises obtaining the distribution rule value of the matching characteristic change node, setting the node rule lower limit by the characteristic weight, the state weight and the dynamic adjustment weight to control the minimum matching.

8. The low altitude vehicle chip safety monitoring system of claim 1, wherein: The instruction execution module comprises: The characteristic data capturing sub-module captures the execution point characteristic data by applying a running environment algorithm based on the strategy generation threshold, eliminates abnormal values and corrects errors, stores in layers according to intervals, performs characteristic processing, and generates a characteristic data dataset. The strategy analysis sub-module divides intervals according to strategy categories based on the characteristic data dataset, extracts change trends and fluctuation characteristics, generates strategy categories and characteristic change characteristic sets. The classification inference sub-module adjusts characteristic parameters and calibrates trend data based on the strategy categories and characteristic change characteristic sets, extracts classification intervals, and performs numerical prediction to obtain instruction execution analysis values.

9. The low altitude vehicle chip safety monitoring system of claim 8, wherein: The running environment algorithm comprises calculating characteristic characteristic values, generating a characteristic data dataset, and comprehensively calculating the characteristic values after characteristic processing by data point weights, execution point original characteristic values, environment values and category coefficients.

10. The low altitude vehicle chip safety monitoring system of claim 1, wherein: The feedback optimization module comprises: The data analysis sub-module extracts real-time characteristics and execution data based on the standardized output data, analyzes real-time monitoring values and execution values, corresponds the monitoring difference value to the current characteristic information, and generates a characteristic monitoring error value. The parameter adjustment sub-module sets the training adjustment parameters of the model based on the characteristic monitoring error value, sets the adjustment range for the strategy category with large error, fine-tunes the low error category, filters and integrates the matched parameter set by comparing the adjustment effect, and generates a model adjustment parameter set. The optimization control sub-module applies the adjustment parameters in each training link based on the model adjustment parameter set, implements model optimization item by item, synchronously monitors characteristics and monitoring, gradually adjusts the optimization order of each strategy category model, and generates a system optimization parameter set.

Images