A debugging method and system for an intelligent converged terminal
By dynamically adjusting the fault judgment threshold and the length of the pre-trigger traceability time window, the problems of high false alarm rate and resource waste in the debugging of intelligent fusion terminals are solved, and adaptive fault detection and evidence collection are realized, improving the accuracy and resource utilization efficiency of the debugging system.
Patent Information
- Application Number
- CN202611216727.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-12
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-08-12
AI Technical Summary
In existing intelligent fusion terminal debugging technologies, the fault judgment threshold is a fixed value, which does not take into account the impact of fluctuations in the load of operating resources such as memory. This leads to data jitter under high load conditions, which can easily cause false alarms. At the same time, the data capture window with a fixed duration is used after a fault is triggered, which cannot adapt to the differences between fast-changing and slow-changing faults, thus limiting the accuracy and resource utilization efficiency of the debugging system.
By acquiring the normal operating status feature vector and original feature vector of the intelligent fusion terminal, the real-time feature vector deviation value is generated. Combined with the historical deviation value and operating resource parameters, the fault judgment threshold and the length of the pre-trigger traceability time window are dynamically adjusted to generate fault capture instructions, extract fault evidence packages, and realize the full-process adaptive closed loop.
It effectively reduces false alarm rates and storage resource waste, improves debugging efficiency and fault tracing capabilities, ensures the integrity of early warning information for slow-changing faults, and saves storage resources for fast-changing faults.
Smart Images

Figure CN122733592B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the technical field of debugging intelligent converged terminals, and in particular to a debugging method and system for intelligent converged terminals. Background Technology
[0002] Intelligent converged terminals refer to embedded devices deployed at the edge of scenarios such as the Internet of Things for power, integrating multi-source data acquisition, communication protocol conversion and local edge computing functions, and used to realize device status perception, data aggregation and collaborative control.
[0003] In related technologies, the existing debugging technology of intelligent fusion terminals usually continuously collects operating status feature data at equal interval sampling rates. The fault judgment threshold is mostly a fixed fault judgment threshold preset by the system. When any feature value exceeds the fixed threshold, a fault alarm is triggered, and status data for a fixed period of time before and after the trigger time is captured to generate a fault record.
[0004] Regarding the aforementioned technologies, during the debugging process of existing intelligent fusion terminals, the fault judgment threshold is a fixed value preset by the system, and the impact of fluctuations in the load of operating resources such as memory on the data acquisition quality is not considered. This leads to data jitter under high load conditions, which can easily cause false alarms. At the same time, after a fault is triggered, a fixed-duration pre-triggered data capture window is used. For fast-changing faults and slow-changing faults with significantly different evolution rates, the data backtracking granularity is indiscriminate. This results in the loss of key precursor information for slow-changing faults or the waste of storage resources for fast-changing faults. Consequently, the accuracy and resource utilization efficiency of the debugging system are limited, and there is still room for improvement. Summary of the Invention
[0005] To reduce false alarm rates and storage resource waste, this application provides a debugging method and system for intelligent converged terminals.
[0006] Firstly, this application provides a debugging method for an intelligent converged terminal, employing the following technical solution: A debugging method for an intelligent converged terminal, comprising: Obtain the preset normal operation status feature vector and original feature vector of the intelligent fusion terminal; The normal operating state feature vector and the original feature vector are analyzed to generate real-time feature vector deviation values; Obtain historical normal deviation values and operating resource parameters; The historical normal deviation values, operating resource parameters, and preset benchmark sensitivity coefficients are analyzed to generate dynamic deviation value thresholds. Analyze the real-time feature vector deviation value and the dynamic deviation value threshold to generate fault capture instructions; The fault capture command and real-time feature vector deviation values are analyzed to generate the pre-trigger traceability time window length; Based on the fault capture command and the length of the pre-triggered traceability time window, the deviation values of the original feature vector and the real-time feature vector are extracted to generate a simultaneous fault evidence package.
[0007] Optionally, the step of analyzing the normal operating state feature vector and the original feature vector to generate real-time feature vector deviation values includes: The normal operating state feature vector is substituted into the preset standardized feature vector calculation formula to generate a standardized normal operating state feature vector. Calculate the arithmetic mean of the standardized normal operating state feature vectors to generate a standardized mean vector; The standardized mean vector and the standardized normal operation feature vector are substituted into the preset standard deviation vector calculation formula to generate the standard deviation feature vector. The original feature vector, standardized mean vector, and standard deviation feature vector are analyzed to generate a standardized current feature vector; The standardized current feature vector and the standardized mean vector are substituted into the preset vector deviation value calculation formula to generate the real-time feature vector deviation value.
[0008] Optionally, the step of analyzing historical normal deviation values, operating resource parameters, and preset benchmark sensitivity coefficients to generate dynamic deviation value thresholds includes: The operating resource parameters and baseline sensitivity coefficients are analyzed to generate corrected sensitivity coefficients; Calculate the arithmetic mean of the historical normal deviations to generate the mean of the historical normal deviations; The historical normal deviation value and the mean of the historical normal deviation value are substituted into the preset standard deviation calculation formula to generate the standard deviation of the historical normal deviation value. Calculate the product of the standard deviation of the historical normal deviation value and the corrected sensitivity coefficient to generate a dynamic threshold safety margin; Calculate the sum of the dynamic threshold safety margin and the average of historical normal deviation values to generate the dynamic deviation threshold.
[0009] Optionally, the steps of analyzing runtime resource parameters and baseline sensitivity coefficients to generate corrected sensitivity coefficients include: Find the memory usage and processor usage in the runtime resource parameters; The memory usage and processor usage are substituted into the preset resource pressure index calculation formula to generate the runtime resource pressure index. The product of the baseline sensitivity coefficient and the operating resource pressure index and the preset resource pressure influence parameters is calculated to generate the sensitivity coefficient correction amount. The sum of the sensitivity coefficient correction and the reference sensitivity coefficient is calculated to generate the corrected sensitivity coefficient.
[0010] Optionally, the step of analyzing real-time feature vector deviation values and dynamic deviation value thresholds to generate fault capture instructions includes: The real-time feature vector deviation value is compared with the dynamic deviation value threshold to generate the correspondence between the positions of the out-of-limit deviation values and the out-of-limit feature vector deviation values. The deviation values of the feature vector exceeding the limit are statistically counted according to the correspondence between the positions of the deviation values exceeding the limit, so as to generate the correspondence between the positions of consecutive exceeding the limit and the number of consecutive exceeding the limit deviation values. The corresponding relationship between the number of consecutive over-limits, the number of consecutive over-limit deviations, and the number of preset capture trigger deviations are analyzed to generate fault capture commands.
[0011] Optionally, the step of analyzing the correspondence between the locations of consecutive out-of-limit quantities, the number of consecutive out-of-limit deviation values, and the number of preset capture trigger deviation values to generate a fault capture command includes: Determine whether the number of consecutive out-of-limit deviation values is not less than the number of capture trigger deviation values; If it is less than, then continue to obtain the number of consecutive out-of-limit deviation values and perform loop judgment; If it is not less than, then find the position of the consecutive over-limit deviation value according to the correspondence between the number of consecutive over-limit deviation values and the position of the consecutive over-limit quantity; Get the capture trigger time; The system summarizes the locations of consecutive out-of-limit deviation values, the capture trigger time, and the preset capture trigger command to generate a fault capture command, and continues to obtain the next number of consecutive out-of-limit deviation values for cyclic judgment.
[0012] Optionally, the step of analyzing the fault capture command and real-time feature vector deviation values to generate the pre-trigger traceability time window length includes: The fault capture command and the real-time feature vector deviation value are substituted into the preset formula for calculating the average rate of change before triggering to generate the average rate of change of the deviation value. Calculate the quotient of the average rate of change of the deviation value and the preset average rate of change of the benchmark to generate the normalized rate of change of the deviation value. The normalized deviation rate of change is substituted into the preset formula for calculating the pre-trigger traceability time window length to generate the pre-trigger traceability time window length.
[0013] Secondly, this application provides a debugging system for an intelligent converged terminal, which adopts the following technical solution: A debugging system for an intelligent converged terminal, comprising: The acquisition module is used to acquire the normal operating status feature vector, the original feature vector, the historical normal deviation value, and the operating resource parameters; A memory for storing a program for a debugging method of an intelligent fusion terminal as described in any of the above items; The processor and the program in the memory can be loaded and executed by the processor to implement a debugging method for an intelligent fusion terminal as described in any of the above.
[0014] In summary, this application includes at least one of the following beneficial technical effects: 1. By analyzing the feature vectors of normal operation and the original feature vectors, the real-time feature vector deviation value is obtained. By analyzing the historical normal deviation value, operating resource parameters and benchmark sensitivity coefficient, the dynamic deviation value threshold is obtained. By analyzing the real-time feature vector deviation value and the dynamic deviation value threshold, the fault capture command is obtained. By analyzing the fault capture command and the real-time feature vector deviation value, the pre-trigger traceability time window length is obtained. Then, based on the fault capture command and the pre-trigger traceability time window length, the original feature vector and the real-time feature vector deviation value are extracted to obtain the fault evidence package. This realizes the full-process adaptive closed loop from anomaly detection, fault confirmation to evidence capture, effectively improving the debugging efficiency and fault tracing capability of the intelligent fusion terminal. 2. By finding the memory utilization and processor utilization in the running resource parameters, and substituting them into the resource pressure index calculation formula, the running resource pressure index is obtained. The baseline sensitivity coefficient is calculated by multiplying the running resource pressure index and the resource pressure influencing parameters to obtain the sensitivity coefficient correction amount. The sensitivity coefficient correction amount is calculated by summing it with the baseline sensitivity coefficient to obtain the corrected sensitivity coefficient. This allows the fault judgment threshold to be dynamically adjusted according to the terminal's own load status. When high load causes data acquisition jitter, the threshold is automatically widened to reduce false alarms, and the detection sensitivity is maintained during low load recovery to avoid missed alarms. 3. By substituting the fault capture command and the real-time feature vector deviation value into the formula for calculating the average rate of change before triggering, the average rate of change of the deviation value is obtained. The quotient of the average rate of change of the deviation value and the baseline average rate of change is calculated to obtain the normalized rate of change of the deviation value. The normalized rate of change of the deviation value is then substituted into the formula for calculating the length of the pre-trigger traceability time window to obtain the length of the pre-trigger traceability time window. The traceability time of the evidence data is then adaptively determined according to the speed of the fault's evolution. A short window is used for sudden and rapidly changing faults to save storage resources, while a long window is used for slowly evolving faults to fully capture the fault precursors. Attached Figure Description
[0015] Figure 1This is a flowchart of a debugging method for an intelligent fusion terminal according to an embodiment of this application.
[0016] Figure 2 This is a flowchart of the steps in this application embodiment to analyze the normal operating state feature vector and the original feature vector to generate a real-time feature vector deviation value.
[0017] Figure 3 This is a flowchart of the steps in this application embodiment to analyze historical normal deviation values, operating resource parameters, and preset benchmark sensitivity coefficients to generate dynamic deviation value thresholds.
[0018] Figure 4 This is a flowchart of the steps in this application embodiment to analyze operating resource parameters and baseline sensitivity coefficients to generate corrected sensitivity coefficients.
[0019] Figure 5 This is a flowchart of the steps in this application embodiment to analyze the real-time feature vector deviation value and dynamic deviation value threshold to generate a fault capture command.
[0020] Figure 6 This is a flowchart of the steps in this application embodiment to analyze the corresponding relationship between the number of consecutive over-limit positions, the number of consecutive over-limit deviation values, and the number of preset capture trigger deviation values to generate a fault capture command.
[0021] Figure 7 This is a flowchart of the steps in this application embodiment to analyze the fault capture command and the real-time feature vector deviation value to generate the length of the pre-trigger traceability time window. Detailed Implementation
[0022] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figures 1 to 7 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.
[0023] This application discloses a debugging method for an intelligent fusion terminal. This method primarily addresses the issues of reducing false alarm rates and wasting storage resources. Specifically, it discloses an intelligent fusion terminal and a processing terminal. The processing terminal is communicatively connected to the intelligent fusion terminal to achieve data interaction and control. After sending the normal operating state feature vector and the original feature vector of the intelligent fusion terminal to the processing terminal, the processing terminal analyzes the normal operating state feature vector and the original feature vector to obtain a real-time feature vector deviation value. Based on a determined dynamic deviation threshold and the real-time feature vector deviation value, the processing terminal filters the original feature vector to obtain a fault capture command. Then, based on the determined fault capture command and the pre-triggered traceability time window length, it extracts the deviation values of the original feature vector and the real-time feature vector to obtain a simultaneous fault evidence packet. This achieves a fully adaptive closed-loop process from anomaly detection and fault confirmation to evidence capture, effectively reducing the false alarm rate caused by high-load data jitter. It also reduces the recording redundancy of fast-changing faults while ensuring the integrity of slow-changing fault precursors, thereby improving the overall debugging efficiency and fault tracing capability of the intelligent fusion terminal under resource-constrained conditions.
[0024] Reference Figure 1 This application discloses a debugging method for an intelligent converged terminal, including the following steps: Step S100: Obtain the preset normal operation status feature vector and original feature vector of the intelligent fusion terminal.
[0025] Among them, the normal operation status feature vector refers to the set of multi-dimensional feature data collected by the intelligent fusion terminal during a period when there are no faults and the working status is stable, including normal communication packet loss rate, normal CPU utilization rate, normal signal strength, etc. The processing terminal obtains the above data and assigns a unique number to each feature data, and then summarizes them to obtain the normal operation status feature vector.
[0026] Normal communication packet loss rate refers to the proportion of data packets sent but not acknowledged by the intelligent fusion terminal during the sampling period when the intelligent fusion terminal is in a fault-free and stable working state, out of the total number of data packets sent. It is used to reflect the reliability of the communication channel. The processing terminal reads the cumulative number of data packets sent and the cumulative number of data packets lost from the protocol stack statistics interface of the intelligent fusion terminal's communication module, and then divides the cumulative number of lost data packets by the cumulative number of sent data packets to obtain the normal communication packet loss rate.
[0027] Normal CPU utilization rate refers to the proportion of time the CPU is in a non-idle state within the sampling period when the intelligent fusion terminal is in a fault-free and stable working state. It is used to reflect the processor load of the intelligent fusion terminal. The processing terminal reads CPU time statistics from the performance interface to obtain the ratio of idle time to total time within the sampling period. Then, subtracting this ratio from 1 gives the normal CPU utilization rate.
[0028] Normal signal strength refers to the power level of the wireless signal received by the intelligent fusion terminal when it is fault-free and in a stable working state. The unit is dBm. It can reflect the wireless coverage quality of the environment in which the intelligent fusion terminal is located. The processing terminal can obtain the normal signal strength by reading the standard interface of the wireless communication module, such as the RSSI reading interface of the Wi-Fi chip.
[0029] The original feature vector refers to the set of multi-dimensional feature data collected by the intelligent fusion terminal during operation according to the sampling period, including the original communication packet loss rate, the original CPU utilization rate, the original signal strength, etc. The processing terminal obtains the above data and assigns a unique number to each feature data, and then summarizes them to obtain the original feature vector.
[0030] The methods for obtaining the raw communication packet loss rate, raw CPU utilization, and raw signal strength in this step are the same as those for obtaining the normal communication packet loss rate, normal CPU utilization, and normal signal strength in this step, and will not be repeated here.
[0031] A smart converged terminal is an embedded device deployed at the edge of the Internet of Things (IoT). It integrates multiple functional modules such as communication, sensing, and computing, and can complete the collection of multi-source heterogeneous data, protocol conversion, edge computing, and collaborative interaction with the cloud locally.
[0032] Step S101: Analyze the normal operating state feature vector and the original feature vector to generate the real-time feature vector deviation value.
[0033] The real-time feature vector deviation value refers to the comprehensive distance value of the original feature vector after standardization within the most recent period, representing the overall deviation of the vector from the normal baseline. The processing terminal analyzes the feature vector in normal operation and the original feature vector to obtain the real-time feature vector deviation value. For specific methods, refer to... Figure 2 This process allows all eigenvalues to be compared on a uniform scale, thus resolving the difficulty in comparison caused by differences in the dimensions and orders of magnitude of different eigenvalues.
[0034] Step S102: Obtain historical normal deviation values and operating resource parameters.
[0035] Among them, the historical normal deviation value refers to the comprehensive distance value of the overall deviation of the intelligent fusion terminal status data from the normal baseline calculated during the period when the intelligent fusion terminal is fault-free and in a stable working state. The processing terminal can obtain the historical normal deviation value by finding the feature vector deviation value of the intelligent fusion terminal in the previous period when it is fault-free and in a stable working state in the historical debugging database at the current time.
[0036] Operating resource parameters refer to the set of hardware usage indicators related to computing and storage resources of the intelligent fusion terminal at the current moment, including memory usage and processor usage.
[0037] Memory utilization rate refers to the proportion of physical memory used by the intelligent fusion terminal to the total physical memory at the end of the most recent sampling period. It is used to reflect the tightness of the intelligent fusion terminal's storage resources. At the end of each sampling period, the processing terminal reads the total memory and available memory through the memory management interface, then divides the available memory by the total memory to obtain the available memory utilization rate, and then subtracts the available memory utilization rate from 1 to obtain the memory utilization rate.
[0038] Processor utilization rate refers to the proportion of time the CPU is in a non-idle state at the end of the most recent sampling period of the intelligent fusion terminal. It is used to reflect the busyness of the processor. The processing terminal reads the CPU time statistics at the beginning and end of each sampling period and calculates the ratio of non-idle time to total CPU time within the period to obtain the processor utilization rate.
[0039] The historical debugging database refers to a database used to store the data acquired and analyzed by the intelligent fusion terminal throughout the debugging process. After each intelligent fusion terminal debugging is completed, the processing terminal matches all the acquired and analyzed data in that cycle with the start time of the current debugging and summarizes them. All debugging parameters are then stored in chronological order to obtain the historical debugging database.
[0040] Step S103: Analyze the historical normal deviation values, operating resource parameters, and preset benchmark sensitivity coefficients to generate dynamic deviation value thresholds.
[0041] The dynamic deviation threshold refers to the actual deviation threshold used for fault determination in the current sampling period. The processing terminal analyzes historical normal deviation values, operating resource parameters, and baseline sensitivity coefficients to obtain the dynamic deviation threshold. For specific methods, refer to [reference needed]. Figure 3 The steps.
[0042] The reference sensitivity coefficient refers to the reference parameter that determines the degree of influence of the pre-set control fault judgment threshold. The larger the reference sensitivity coefficient, the more lenient the dynamic deviation threshold will be, which will result in fewer false alarms but a higher probability of missed alarms. The smaller the reference sensitivity coefficient, the more stringent the dynamic deviation threshold will be, which will result in more false alarms but a lower probability of missed alarms. In one embodiment, the reference sensitivity coefficient is 3.
[0043] Step S104: Analyze the real-time feature vector deviation value and the dynamic deviation value threshold to generate a fault capture command.
[0044] The fault capture command refers to a set of parameters that trigger subsequent data capture actions after a continuous exceedance judgment confirms a fault. The processing terminal analyzes the real-time feature vector deviation value and the dynamic deviation value threshold to obtain the fault capture command. For specific methods, refer to... Figure 5 The steps.
[0045] Step S105: Analyze the fault capture command and the real-time feature vector deviation value to generate the pre-trigger traceability time window length.
[0046] The pre-trigger tracing time window length refers to the duration of data tracing backward from the capture trigger moment. The processing terminal analyzes the fault capture command and the real-time feature vector deviation value to obtain the pre-trigger tracing time window length. For specific methods, refer to [reference needed]. Figure 7 The steps.
[0047] Step S106: Extract the deviation values of the original feature vector and the real-time feature vector according to the fault capture command and the length of the pre-trigger traceability time window to generate a simultaneous fault evidence package.
[0048] The simultaneous stamped fault evidence package refers to a structured data block formed by packaging data such as the original feature vector and the real-time feature vector deviation value, which are extracted forward according to the length of the pre-trigger tracing time window before and after the trigger time. The simultaneous stamped fault evidence package can provide a diagnostic basis for subsequent fault diagnosis. The processing terminal extracts the data of the original feature vector and the real-time feature vector deviation value forward according to the length of the pre-trigger tracing time window, and summarizes it with the data such as the extraction position and trigger time in the fault capture instruction to obtain the simultaneous stamped fault evidence package.
[0049] Reference Figure 2 The steps for analyzing the feature vectors of normal operation and the original feature vectors to generate real-time feature vector deviation values include: Step S200: Substitute the normal operation state feature vector into the preset standardized feature vector calculation formula to generate a standardized normal operation state feature vector.
[0050] The standardized normal operation state feature vector refers to the feature vector sequence after standardizing the normal operation state feature vector. The processing terminal substitutes the normal operation state feature vector into the standardized feature vector calculation formula to obtain the standardized normal operation state feature vector.
[0051] The formula for calculating the standardized feature vector is: .
[0052] In the formula, This refers to the sampling time. Standardized normal operating state feature vector at that time It refers to the first The normal operating state feature vector of the dimensional feature. It refers to the first Historical normal mean of the dimensional feature It refers to the first The historical normal standard deviation of the feature vector is obtained by the processing terminal from the historical debugging database based on the current sampling time. When the intelligent fusion terminal is in the initial power-on stage, both the historical normal mean and the historical normal standard deviation are preset benchmark values. If the normal communication packet loss rate is included in the feature vector of the normal operating state, then... =1), Normal CPU utilization ( =2) and normal signal strength ( =3) constitutes, then =0.005, =0.002, =45%, =12%, =-75dBm, =8dBm.
[0053] Step S201: Calculate the arithmetic mean of the standardized normal operation state feature vector to generate a standardized mean vector.
[0054] The standardized mean vector refers to the arithmetic mean of each feature dimension in the standardized normal operating state feature vector over a sampling period. This means that the processing terminal sums the feature values of each dimension in the standardized normal operation state feature vector, counts the total number of sampling points in the standardized normal operation state feature vector, divides the sum of each dimension by the total number of sampling points, and marks and summarizes the resulting mean value according to the sampling point number to obtain the standardized mean vector.
[0055] Step S202: Substitute the standardized mean vector and the standardized normal operation state feature vector into the preset standard deviation vector calculation formula to generate the standard deviation feature vector.
[0056] Here, the standard deviation eigenvector refers to the standard deviation of each feature dimension in the standardized normal operation state eigenvector over a sampling period. This indicates that the standard deviation vector is used to reflect the dispersion of each feature data. The processing terminal substitutes the standardized mean vector and the standardized normal operation state feature vector into the standard deviation vector calculation formula to obtain the standard deviation feature vector.
[0057] The formula for calculating the standard deviation vector is: .
[0058] In the formula, It refers to the first At the sampling time, the first The standard deviation eigenvector of a dimensional feature. This refers to the number of sampling points; for example, 1000. It refers to the first At the sampling time, the first The standardized normal operation state feature vector of the dimensional features. It refers to the first At the sampling time, the first The standardized mean vector of the dimensional features.
[0059] Step S203: Analyze the original feature vector, the standardized mean vector, and the standard deviation feature vector to generate a standardized current feature vector.
[0060] Standardizing the current feature vector refers to the sequence of feature vectors obtained by standardizing the original feature vectors. This means that the processing terminal substitutes the original feature vector, the standardized mean vector, and the standard deviation feature vector into the standardized feature vector calculation formula to obtain the standardized current feature vector. The standardized feature vector calculation formula in this step is the same as the standardized feature vector calculation formula in step S200 above, and will not be repeated here.
[0061] Step S204: Substitute the standardized current feature vector and the standardized mean vector into the preset vector deviation value calculation formula to generate the real-time feature vector deviation value.
[0062] In this step, the real-time feature vector deviation value is the same as the real-time feature vector deviation value in step S101 above. The processing terminal substitutes the standardized current feature vector and the standardized mean vector into the vector deviation value calculation formula to calculate the real-time feature vector deviation value.
[0063] The formula for calculating the vector deviation value is: .
[0064] In the formula, It means The real-time feature vector deviation value at time t. It means The standardized current feature vector at time step, It means The standardized mean vector at time t. It means The inverse of the covariance matrix at time t.
[0065] Reference Figure 3 The steps for generating dynamic deviation thresholds by analyzing historical normal deviation values, operating resource parameters, and preset benchmark sensitivity coefficients include: Step S300: Analyze the operating resource parameters and baseline sensitivity coefficients to generate corrected sensitivity coefficients.
[0066] The corrected sensitivity coefficient refers to the parameter used to control the degree of influence of the fault determination threshold after adjusting the alignment threshold sensitivity coefficient according to resource pressure. The processing terminal can obtain the corrected sensitivity coefficient by analyzing the operating resource parameters and the baseline sensitivity coefficient. The specific method is described in [reference needed]. Figure 4 The steps.
[0067] Step S301: Calculate the arithmetic mean of the historical normal deviations to generate the mean of the historical normal deviations.
[0068] The historical normal deviation mean refers to the average value of all deviation values in the historical normal deviation values. The processing terminal calculates the sum of all deviation values in the historical normal deviation values and counts the number of all deviation values in the historical normal deviation values. The historical normal deviation mean is obtained by dividing the calculated sum of data by the number of data.
[0069] Step S302: Substitute the historical normal deviation value and the mean of the historical normal deviation value into the preset standard deviation calculation formula to generate the standard deviation of the historical normal deviation value.
[0070] The standard deviation of historical normal deviation values refers to the data used to quantify the dispersion of data in historical normal deviation values. The processing terminal substitutes the historical normal deviation values and the mean of historical normal deviation values into the standard deviation calculation formula to obtain the standard deviation of historical normal deviation values.
[0071] The formula for calculating the standard deviation of the deviation value is: .
[0072] In the formula, This refers to the standard deviation of historical normal values. This refers to the number of sampling points; for example, 1000. It refers to the first Historical normal deviation values at each sampling time It refers to the first The average of the historical normal deviation values at each sampling time.
[0073] Step S303: Calculate the product of the standard deviation of the historical normal deviation value and the corrected sensitivity coefficient to generate the dynamic threshold safety margin.
[0074] The dynamic threshold safety margin refers to a buffer zone that is proportional to the normal fluctuation range, which is superimposed on the standard deviation of the historical normal deviation value. The processing terminal multiplies the standard deviation of the historical normal deviation value by the correction sensitivity coefficient to obtain the dynamic threshold safety margin.
[0075] Step S304: Calculate the sum of the dynamic threshold safety margin and the average of the historical normal deviation values to generate the dynamic deviation threshold.
[0076] In this step, the dynamic deviation threshold is the same as the dynamic deviation threshold in step S103 above. The processing terminal adds the dynamic threshold safety margin to the average of historical normal deviation values to obtain the dynamic deviation threshold.
[0077] Reference Figure 4 The steps for analyzing operating resource parameters and baseline sensitivity coefficients to generate corrected sensitivity coefficients include: Step S400: Locate the memory usage and processor usage in the runtime resource parameters.
[0078] In this step, the memory usage rate and processor usage rate are the same as those in step S102 above. The processing terminal can obtain the memory usage rate and processor usage rate by searching in the running resource parameters.
[0079] Step S401: Substitute the memory usage rate and processor usage rate into the preset resource pressure index calculation formula to generate the runtime resource pressure index.
[0080] The runtime resource pressure index is a normalized index that is a weighted average of memory usage and processor usage. The processing terminal substitutes the memory usage and processor usage into the resource pressure index calculation formula to obtain the runtime resource pressure index.
[0081] The formula for calculating the resource pressure index is: .
[0082] In the formula, This refers to the operational resource pressure index. The weighting percentage of processor utilization. The weighting ratio for memory usage is as follows: = For example, =0.5 This refers to processor utilization. This refers to memory usage.
[0083] Step S402: Calculate the product of the baseline sensitivity coefficient, the operating resource pressure index, and the preset resource pressure influence parameters to generate the sensitivity coefficient correction amount.
[0084] The sensitivity coefficient correction amount refers to the sensitivity coefficient correction amount that is applied to the baseline sensitivity coefficient based on resource pressure. The processing terminal multiplies the baseline sensitivity coefficient by the operating resource pressure index and the resource pressure influence parameter to obtain the sensitivity coefficient correction amount.
[0085] The resource pressure influence parameter refers to the influence coefficient that determines the degree of influence of resource pressure on the sensitivity coefficient. In one embodiment, the resource pressure influence parameter is 0.5.
[0086] Step S403: Calculate the sum of the sensitivity coefficient correction and the reference sensitivity coefficient to generate the corrected sensitivity coefficient.
[0087] In this step, the corrected sensitivity coefficient is the same as the corrected sensitivity coefficient in step S300 above. The processing terminal adds the sensitivity coefficient correction amount to the reference sensitivity coefficient to obtain the corrected sensitivity coefficient.
[0088] Reference Figure 5 The steps for analyzing real-time feature vector deviation values and dynamic deviation value thresholds to generate fault capture instructions include: Step S500: Compare the real-time feature vector deviation value with the dynamic deviation value threshold to generate the correspondence between the position of the out-of-limit deviation value and the out-of-limit feature vector deviation value.
[0089] The correspondence between the positions of the out-of-limit deviation values refers to the mapping relationship between the sampling point positions of each feature vector deviation value that exceeds the dynamic deviation value threshold and the corresponding feature vector deviation value in the real-time feature vector deviation values. The out-of-limit feature vector deviation value refers to the set of deviation values in the real-time feature vector deviation values that exceed the dynamic deviation value threshold. Starting from the starting position in the real-time feature vector deviation values, the processing terminal compares the feature vector deviation value at each sampling point with the dynamic deviation value threshold, extracts the deviation values that are greater than the dynamic deviation value threshold and their corresponding sampling point positions, and forms a mapping table one by one to obtain the correspondence between the positions of the out-of-limit deviation values. Then, all the deviation values are extracted from the correspondence between the positions of the out-of-limit deviation values and summarized to obtain the out-of-limit feature vector deviation value.
[0090] Step S501: Statistically count the deviation values of the feature vector exceeding the limit according to the correspondence of the position of the deviation value exceeding the limit, so as to generate the correspondence of the position of the consecutive exceeding the limit and the number of consecutive exceeding the limit deviation values.
[0091] The "correspondence relationship of consecutive over-limit quantity positions" refers to the mapping relationship between the sampling point positions of consecutive over-limits and the corresponding over-limit feature vector deviation values and the corresponding number of consecutive over-limit sampling points in the correspondence relationship of over-limit deviation value positions. The "number of consecutive over-limit deviation values" refers to the number of over-limit feature vector deviation values in each mapping table constituting consecutive over-limits in the correspondence relationship of over-limit deviation value positions. The processing terminal traverses the correspondence relationship of over-limit deviation value positions, matching the sampling point positions of at least one adjacent sampling point with the corresponding deviation value one by one, and then counts the number of adjacent sampling point positions to obtain the number of consecutive over-limit deviation values. Finally, the adjacent sampling point positions, the number of consecutive over-limit deviation values, and the corresponding deviation values are matched one by one to form a mapping table, thus obtaining the correspondence relationship of consecutive over-limit quantity positions. Since the sampling point positions corresponding to each deviation value in the correspondence relationship of over-limit deviation value positions cannot be determined, that is, the sampling point positions corresponding to each deviation value are not necessarily adjacent, there may be multiple sets of mapping tables for consecutive over-limit quantity positions in the correspondence relationship of consecutive over-limit quantity positions.
[0092] Step S502: Analyze the correspondence between the number of consecutive over-limit positions, the number of consecutive over-limit deviation values, and the number of preset capture trigger deviation values to generate a fault capture command.
[0093] The fault capture command in this step is the same as the fault capture command in step S104 above. The processing terminal analyzes the correspondence between the positions of the consecutive over-limit quantities, the number of consecutive over-limit deviation values, and the number of capture trigger deviation values to obtain the fault capture command. The specific method is as follows: Figure 6 The steps.
[0094] The number of captured trigger deviation values refers to the minimum number of consecutive out-of-limit occurrences required to confirm a fault. In one embodiment, the number of captured trigger deviation values is 5.
[0095] Reference Figure 6 The steps for generating a fault capture command include analyzing the correspondence between the number of consecutive out-of-limit locations, the number of consecutive out-of-limit deviation values, and the number of preset capture trigger deviation values. Step S600: Determine whether the number of consecutive out-of-limit deviation values is not less than the number of capture trigger deviation values.
[0096] The processing terminal determines whether the number of consecutive out-of-limit deviation values is not less than the number of capture trigger deviation values, thereby determining whether it is necessary to capture the group of consecutive out-of-limit deviation values and the original feature vector in the correspondence between the number of consecutive out-of-limit deviation values and the position of the consecutive out-of-limit deviation values.
[0097] Step S6001: If it is less than, continue to obtain the number of consecutive out-of-limit deviation values and perform cyclic judgment.
[0098] If the processing terminal determines that the number of consecutive out-of-limit deviation values is less than the number of capture trigger deviation values, it means that it is not necessary to capture the group of consecutive out-of-limit deviation values and the original feature vector in the correspondence of the consecutive out-of-limit number positions. Therefore, the processing terminal continues to obtain the next consecutive out-of-limit deviation value for loop judgment, and stops looping after traversing all the data in the correspondence of the consecutive out-of-limit number positions.
[0099] The next consecutive out-of-limit deviation value number refers to the number of deviation values in the next consecutive out-of-limit deviation value sequence in the correspondence between consecutive out-of-limit quantity positions.
[0100] Step S6002: If it is not less than, then find the position of the continuous over-limit deviation value according to the correspondence between the number of continuous over-limit deviation values and the position of the continuous over-limit quantity.
[0101] If the processing terminal determines that the number of consecutive over-limit deviation values is not less than the number of capture trigger deviation values, it means that it is necessary to capture the group of consecutive over-limit deviation values and the original feature vector in the correspondence between the number of consecutive over-limit deviation values and the position of the consecutive over-limit deviation values. Therefore, the processing terminal needs to determine the position of the consecutive over-limit deviation values to provide data support for the subsequent determination of the fault capture command.
[0102] The location of continuous out-of-limit deviation values refers to the set of sampling points in the mapping table corresponding to the number of continuous out-of-limit deviation values in the correspondence between the number of continuous out-of-limit deviation values. The processing terminal can obtain the location of continuous out-of-limit deviation values by searching and summarizing in the correspondence between the number of continuous out-of-limit deviation values.
[0103] Step S60021: Obtain the capture trigger time.
[0104] The capture trigger time refers to the sampling timestamp when the number of consecutive over-limits first reaches the capture trigger deviation value. The processing terminal reads the current time of the system clock when it first determines that the number of consecutive over-limits has reached the capture trigger deviation value, and thus obtains the capture trigger time.
[0105] Step S60022: Summarize the locations of consecutive out-of-limit deviation values, the capture trigger time, and the preset capture trigger command to generate a fault capture command, and continue to obtain the next number of consecutive out-of-limit deviation values for cyclic judgment.
[0106] In this step, the fault capture instruction is the same as that in step S104 above. The processing terminal summarizes the location of the continuous over-limit deviation value, the capture trigger time and the preset capture trigger instruction to obtain the fault capture instruction, and continues to obtain the next number of continuous over-limit deviation values for loop judgment. After traversing all the data in the correspondence between the number of continuous over-limit deviation values and the location, the loop stops.
[0107] The number of the next consecutive out-of-limit deviation values in this step is the same as the number of the next consecutive out-of-limit deviation values in step S6001 above, and will not be repeated here.
[0108] Reference Figure 7 The steps for analyzing fault capture commands and real-time feature vector deviation values to generate the pre-trigger traceability time window length include: Step S700: Substitute the fault capture command and the real-time feature vector deviation value into the preset formula for calculating the average rate of change before triggering to generate the average rate of change of the deviation value.
[0109] The average rate of change of deviation value refers to the change in deviation value per unit time within the preset sampling points before the capture trigger for each dimension feature. The processing terminal substitutes the fault capture command and the real-time feature vector deviation value into the formula for calculating the average rate of change before the trigger to obtain the average rate of change of deviation value.
[0110] The formula for calculating the average rate of change before triggering is: .
[0111] In the formula, This refers to the average rate of change of the deviation value. This refers to the number of sampling points; for example, 30. This refers to the sampling period, taking 1 second as an example. It means The real-time feature vector deviation value at time t. It means The real-time feature vector deviation value at time t. This refers to the moment when the capture is triggered.
[0112] Step S701: Calculate the quotient of the average rate of change of the deviation value and the preset average rate of change of the benchmark to generate the normalized rate of change of the deviation value.
[0113] The normalized deviation rate of change refers to the rate of change of each dimension feature after normalizing the average rate of change of the deviation value of each dimension feature. The processing terminal divides the average rate of change of the deviation value by the benchmark average rate of change to obtain the normalized deviation rate of change.
[0114] The benchmark average rate of change refers to a reference value used to measure how fast the average rate of change of the deviation value is. In one embodiment, the benchmark average rate of change is 0.1 standard deviations per second.
[0115] Step S702: Substitute the rate of change of the normalized deviation value into the preset formula for calculating the length of the pre-trigger traceability time window to generate the length of the pre-trigger traceability time window.
[0116] In this step, the length of the pre-trigger traceability time window is the same as that in step S105 above. The processing terminal substitutes the rate of change of the normalized deviation value into the formula for calculating the length of the pre-trigger traceability time window to obtain the length of the pre-trigger traceability time window.
[0117] The formula for calculating the pre-trigger retrospective time window length is: .
[0118] In the formula, This refers to the length of the pre-trigger traceback time window. This refers to the minimum pre-trigger traceability time window length, taking 5 seconds as an example. This refers to the maximum pre-trigger traceability time window length, taking 300 seconds as an example. This refers to the rate of change of the normalized deviation value.
[0119] Based on the same inventive concept, embodiments of this application provide a debugging system for an intelligent fusion terminal, comprising: The acquisition module is used to acquire the normal operating status feature vector, the original feature vector, the historical normal deviation value, the operating resource parameters, the number of the next consecutive out-of-limit deviation values, and the capture trigger time; A memory used to store a program for debugging a smart fusion terminal; The processor and memory can load and execute programs, and implement a debugging method for intelligent fusion terminals.
[0120] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0121] This application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as a debugging method for an intelligent fusion terminal.
[0122] Computer storage media include, for example, USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media that can store program code.
[0123] Based on the same inventive concept, embodiments of this application provide a smart terminal, including a memory and a processor, wherein the memory stores a computer program that can be loaded and executed by the processor to provide a debugging method for a smart fusion terminal.
[0124] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0125] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.
Claims
1. A debugging method for an intelligent converged terminal, characterized in that, include: Obtain the preset normal operation status feature vector and original feature vector of the intelligent fusion terminal; The normal operating state feature vector and the original feature vector are analyzed to generate real-time feature vector deviation values; Obtain historical normal deviation values and operating resource parameters; Analyze historical normal deviation values, operating resource parameters, and preset benchmark sensitivity coefficients to generate dynamic deviation value thresholds; Analyze the real-time feature vector deviation value and the dynamic deviation value threshold to generate fault capture instructions; The fault capture command and real-time feature vector deviation values are analyzed to generate the pre-trigger traceability time window length; Based on the fault capture command and the length of the pre-triggered tracing time window, the deviation values of the original feature vector and the real-time feature vector are extracted to generate a simultaneous fault evidence package; The steps for analyzing historical normal deviation values, operating resource parameters, and preset benchmark sensitivity coefficients to generate dynamic deviation thresholds include: The operating resource parameters and baseline sensitivity coefficients are analyzed to generate corrected sensitivity coefficients; Calculate the arithmetic mean of the historical normal deviations to generate the mean of the historical normal deviations; The historical normal deviation value and the mean of the historical normal deviation value are substituted into the preset standard deviation calculation formula to generate the standard deviation of the historical normal deviation value. Calculate the product of the standard deviation of the historical normal deviation value and the corrected sensitivity coefficient to generate the dynamic threshold safety margin; Calculate the sum of the dynamic threshold safety margin and the average of historical normal deviation values to generate the dynamic deviation threshold; The steps for analyzing operating resource parameters and baseline sensitivity coefficients to generate corrected sensitivity coefficients include: Find the memory usage and processor usage in the runtime resource parameters; The memory usage and processor usage are substituted into the preset resource pressure index calculation formula to generate the runtime resource pressure index. The product of the baseline sensitivity coefficient and the operating resource pressure index and the preset resource pressure influence parameters is calculated to generate the sensitivity coefficient correction amount. Calculate the sum of the sensitivity coefficient correction and the reference sensitivity coefficient to generate the corrected sensitivity coefficient; The steps for analyzing fault capture commands and real-time feature vector deviation values to generate the pre-trigger traceability time window length include: The fault capture command and the real-time feature vector deviation value are substituted into the preset formula for calculating the average rate of change before triggering to generate the average rate of change of the deviation value. Calculate the quotient of the average rate of change of the deviation value and the preset average rate of change of the benchmark to generate the normalized rate of change of the deviation value. The normalized deviation rate of change is substituted into the preset formula for calculating the pre-trigger traceability time window length to generate the pre-trigger traceability time window length. The formula for calculating the average rate of change before triggering is: ; In the formula, This refers to the average rate of change of the deviation value. This refers to the preset number of sampling points. This refers to the preset sampling period. It means The real-time feature vector deviation value at time t. It means The real-time feature vector deviation value at time t. This refers to the moment when the capture is triggered.
2. The debugging method for an intelligent fusion terminal according to claim 1, characterized in that, The steps for analyzing the normal operating state feature vector and the original feature vector to generate the real-time feature vector deviation value include: The normal operating state feature vector is substituted into the preset standardized feature vector calculation formula to generate a standardized normal operating state feature vector. Calculate the arithmetic mean of the standardized normal operating state feature vectors to generate a standardized mean vector; The standardized mean vector and the standardized normal operation feature vector are substituted into the preset standard deviation vector calculation formula to generate the standard deviation feature vector. The original feature vector, standardized mean vector, and standard deviation feature vector are analyzed to generate a standardized current feature vector; The standardized current feature vector and the standardized mean vector are substituted into the preset vector deviation value calculation formula to generate the real-time feature vector deviation value.
3. The debugging method for an intelligent fusion terminal according to claim 1, characterized in that, The steps for analyzing real-time feature vector deviation values and dynamic deviation value thresholds to generate fault capture instructions include: The real-time feature vector deviation value is compared with the dynamic deviation value threshold to generate the correspondence between the positions of the out-of-limit deviation values and the out-of-limit feature vector deviation values. The deviation values of the feature vector exceeding the limit are statistically counted according to the correspondence between the positions of the deviation values exceeding the limit, so as to generate the correspondence between the positions of consecutive exceeding the limit and the number of consecutive exceeding the limit deviation values. The corresponding relationship between the number of consecutive over-limits, the number of consecutive over-limit deviations, and the number of preset capture trigger deviations are analyzed to generate fault capture commands.
4. The debugging method for an intelligent fusion terminal according to claim 3, characterized in that, The steps for generating a fault capture command include analyzing the correspondence between the number of consecutive out-of-limit locations, the number of consecutive out-of-limit deviation values, and the number of preset capture trigger deviation values: Determine whether the number of consecutive out-of-limit deviation values is not less than the number of capture trigger deviation values; If it is less than, then continue to obtain the number of consecutive out-of-limit deviation values and perform loop judgment; If it is not less than, then find the position of the consecutive over-limit deviation value according to the correspondence between the number of consecutive over-limit deviation values and the position of the consecutive over-limit quantity; Get the capture trigger time; The system summarizes the locations of consecutive out-of-limit deviation values, the capture trigger time, and the preset capture trigger command to generate a fault capture command, and continues to obtain the next number of consecutive out-of-limit deviation values for cyclic judgment.
5. A debugging system for an intelligent fusion terminal, characterized in that, include: The acquisition module is used to acquire the normal operating status feature vector, the original feature vector, the historical normal deviation value, and the operating resource parameters; A memory for storing a program for debugging a smart fusion terminal as described in any one of claims 1 to 4; The processor and the program in the memory can be loaded and executed by the processor to implement the debugging method of the intelligent fusion terminal as described in any one of claims 1 to 4.
Citation Information
Patent Citations
Traceable dynamic window model-based power consumption data trend anomaly analysis method
CN111191676A
Self-adaptive backtracking window calculation obtaining method and device and storage medium
CN121579261A
