Power equipment defect detection method based on unmanned aerial vehicle inspection

Abstract

The application discloses a power equipment defect detection method based on unmanned aerial vehicle inspection, and belongs to the field of power equipment detection, and aims to solve the problems of fragmented inspection process, poor operation parameter adaptability and parameter adjustment depending on artificial experience; the method comprises the following steps: collecting historical basic parameters and real-time state parameters of the inspection area, calculating the risk value of the inspection area and matching the risk level; collecting historical inspection data, extracting core evaluation parameters to calculate defect detection rate and defect misjudgment rate, screening qualified operation collection parameter groups and grouping according to the risk level, removing double abnormal values, and delimiting a preliminary parameter range; obtaining a core function set of the inspection area, screening adaptive adjustment parameters, calculating the dynamic influence weight and the reference deviation value, and combining the preliminary parameter range to correct the exclusive operation execution collection parameters; the application constructs a full-link collaborative mechanism, realizes multi-dimensional parameter adaptation and quantitative adjustment, improves the inspection accuracy, adaptability and stability, and reduces resource waste.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power equipment detection, and particularly relates to a power equipment defect detection method based on unmanned aerial vehicle inspection. BACKGROUND

[0002] With the wide application of unmanned aerial vehicle technology in the field of power equipment inspection, the traditional inspection scheme cannot meet the differentiated and accurate inspection requirements of the power grid, and the existing technology has obvious shortcomings, mainly in the following three core problems:

[0003] The existing scheme lacks a "risk assessment-parameter adaptation-working condition correction" full-link cooperation mechanism, risk assessment is carried out in isolation, and is not associated with subsequent operation parameter setting, and the parameters are not dynamically adjusted in combination with the real-time working conditions of the region after being determined, resulting in fragmented inspection process and inability to achieve targeted inspection.

[0004] The operation parameter adaptation dimension is single, only a few working conditions such as environmental temperature are simply referred to, neither the basic parameter range is matched according to the regional risk level classification, nor the parameters are adjusted in association with the core needs such as flight safety and data quality, and the specificity is insufficient, and the high-risk area is prone to missed inspection.

[0005] Parameter adjustment optimization lacks quantitative support, key parameters are not selected through the correlation strength of parameters and core needs and the influence amplitude of parameter changes on inspection effect, and the linkage relationship among multiple parameters is not clear, relying on manual experience adjustment, which is difficult to adapt to working condition differences and the inspection effect fluctuates greatly, therefore we propose a power equipment defect detection method based on unmanned aerial vehicle inspection. SUMMARY

[0006] The purpose of the present application is to provide a power equipment defect detection method based on unmanned aerial vehicle inspection to solve the problems raised in the background art.

[0007] To achieve the above purpose, the present application provides the following technical scheme: a power equipment defect detection method based on unmanned aerial vehicle inspection, comprising the following steps:

[0008] Step 1: Collect historical basic parameters and real-time state parameters in units of independent inspection areas, calculate the risk value of the inspection area, and match the risk level of the inspection area;

[0009] Step 2: Collect historical inspection data of the inspection area, extract core evaluation parameters corresponding to each operation collection parameter group of the unmanned aerial vehicle, calculate the defect detection rate and defect misjudgment rate, and accordingly select the qualified operation collection parameter group and group it according to the risk level, analyze and determine the preliminary range of each operation collection parameter for the qualified operation collection parameter group set of different risk levels;

[0010] Step three: obtaining the core function set of the inspection area; extracting the candidate adaptive adjustment parameters of the operation collection parameters, calculating the comprehensive function correlation influence value, and screening the adaptive adjustment parameters; calculating the dynamic influence weight of the adaptive adjustment parameters, and dividing them into same-direction or reverse linkage adjustment parameters; calculating the reference deviation value and combining the preliminary reasonable range of the operation collection parameters to correct them to obtain the operation execution collection parameters, and arranging them into the unmanned aerial vehicle operation execution collection parameter set of the inspection area to control the unmanned aerial vehicle to carry out the inspection.

[0011] Preferably, in step one, the specific process of calculating the inspection area risk value and matching the risk level of the inspection area is as follows:

[0012] All areas to be inspected by the unmanned aerial vehicle are obtained, and two types of core parameters are collected for each independent inspection area;

[0013] The first type is the historical basic parameter, which includes the defect occurrence frequency within a preset period, the equipment operation period, and the component material adaptation coefficient;

[0014] The second type is the real-time state parameter, which includes the current regional environment humidity, the line load coefficient, and the environmental temperature;

[0015] A risk level quantification model is constructed based on the historical basic parameters and real-time state parameters. The defect occurrence frequency within a preset period, the equipment operation period, the result of 1 minus the component material adaptation coefficient, the current regional environment humidity, the line load coefficient, and the environmental temperature are multiplied by their respective preset weight coefficients, and then all the product results are summed to obtain the inspection area risk value;

[0016] A plurality of regional risk levels are preset, and a risk value interval is configured for each regional risk level;

[0017] The calculated inspection area risk value is compared with the risk value intervals corresponding to all regional risk levels to determine the risk value interval to which the inspection area risk value belongs, and then the risk level corresponding to the inspection area is output.

[0018] Preferably, in step one, the process of obtaining the material adaptation coefficient is as follows:

[0019] The material type of the main equipment component in the current inspection area, the typical operating environment type of the current inspection area, and the historical failure probability of the material of the main equipment component in the same environment as the typical operating environment type of the current inspection area are collected;

[0020] A preset material weight coefficient is set, and a material-environment adaptation correction function is set. When the material type of the main equipment component and the typical operating environment type of the current inspection area match each other, the material-environment adaptation correction function takes a first preset positive value;

[0021] The material-environment adaptation correction function takes a second preset positive value when the material type of the main equipment component does not match the typical operating environment type of the current inspection area; wherein the first preset positive value is less than the second preset positive value;

[0022] The material adaptation coefficient is calculated by subtracting the product of the material preset weight coefficient, the historical failure probability and the value of the material-environment adaptation correction function from 1.

[0023] Preferably, in step one, the process of obtaining the line load coefficient is as follows:

[0024] The real-time load power of the power supply line corresponding to the current inspection area, the rated load power of the line, and the current running time of the line from the start of this inspection to the parameter collection time are obtained.

[0025] The real-time load power, the rated load power and the current running time are normalized and de-dimensioned to eliminate the dimensional differences between different parameters.

[0026] A preset running time correction coefficient is multiplied by 1 plus the logarithm of the current running time to quantify the cumulative effect of long-time operation on the line load risk.

[0027] The ratio of the normalized real-time load power to the rated load power is multiplied by 1 plus the cumulative effect quantification result to obtain the preliminary calculation value of the line load coefficient.

[0028] The smaller value between the preliminary calculation value and 1.0 is selected as the line load coefficient of the power supply line.

[0029] Preferably, in step two, the specific process of collecting historical inspection data of the inspection area and extracting core evaluation parameters corresponding to each set of unmanned aerial vehicle operation collection parameters to calculate the defect detection rate and the defect misjudgment rate is as follows:

[0030] For each inspection area, the historical inspection data of the inspection area is collected, including the risk level of the inspection area at each inspection, the unmanned aerial vehicle operation collection parameter set and the defect detection result data of the inspection.

[0031] For each set of operation collection parameters, the core evaluation parameters are extracted from the corresponding defect detection result data, including the total number of real defects, the number of effectively detected defects and the defect misjudgment rate.

[0032] The defect detection rate of each set of operation collection parameters is obtained by dividing the number of effectively detected defects by the total number of real defects.

[0033] Preferably, in step two, the specific process of selecting qualified operation collection parameter sets and grouping them according to risk levels is as follows

[0034] filtering out the job collection parameter groups that simultaneously satisfy:

[0035] a defect misjudgment rate less than a corresponding preset defect misjudgment rate threshold value and a defect detection rate greater than or equal to a corresponding preset defect detection rate threshold value, and recording the job collection parameter groups as qualified job collection parameter groups;

[0036] grouping all the qualified job collection parameter groups according to risk levels corresponding to the qualified job collection parameter groups, to obtain a qualified job collection parameter group set corresponding to each risk level.

[0037] Preferably, the specific process of analyzing and determining the preliminary range of each job collection parameter for the qualified job collection parameter group set of different risk levels is as follows:

[0038] extracting sample data of the corresponding qualified job collection parameter group for each job collection parameter, and arranging the sample data to form a sample sequence of the job collection parameter; first, removing extreme samples in the sample sequence by the quartile range method, and then removing edge extreme values by the median correlation interval secondary screening method;

[0039] extracting the minimum value and the maximum value of the effective sample after secondary screening, taking the minimum value and the maximum value as the upper and lower limits of the interval respectively, to delimit the initial reasonable interval of the corresponding job collection parameter under the risk level;

[0040] arranging the initial reasonable intervals of all job collection parameters under each risk level to clarify the corresponding relationship between the risk level and the preliminary range of the job collection parameter, and forming a risk-parameter preliminary range linkage rule;

[0041] substituting the risk level of the current inspection area into the linkage rule for matching, and outputting the preliminary range of each job collection parameter corresponding to the area.

[0042] Preferably, in step three, the specific process of extracting the candidate adaptive adjustment parameters of the job collection parameter and calculating the comprehensive function correlation influence value thereof, and screening the adaptive adjustment parameters is as follows:

[0043] constructing a core appeal-core function correspondence table based on the inspection core appeals of each inspection area, substituting all core appeals corresponding to the current inspection area into the table for matching, and arranging and outputting the core functions to obtain the inspection area core function set of the current inspection area;

[0044] obtaining each job collection parameter corresponding to the current inspection area, extracting parameters that may affect the correlation effect between the job collection parameter and each core function in the inspection area core function set of the current inspection area as candidate adaptive adjustment parameters of the job collection parameter;

[0045] extracting a historical value sequence of each candidate adaptive adjustment parameter, a historical value sequence of a quantitative index of each core function in the core function set of the current inspection area, and performing normalization and de-dimensioning on both types of sequences; combining a preset weight of each core function, an absolute value of a correlation coefficient between the normalized historical value sequence of the candidate adaptive adjustment parameter and the normalized historical value sequence of the quantitative index of the core function, and a maximum difference sum of the normalized mean values of the core function quantitative index corresponding to each group after the candidate adaptive adjustment parameter historical value sequence is divided into three equal parts, to obtain a comprehensive function-related influence value of the candidate adaptive adjustment parameter;

[0046] filtering the candidate adaptive adjustment parameters with a comprehensive function-related influence value not lower than the threshold value, and marking them as adaptive adjustment parameters.

[0047] Preferably, in step three, the specific process of calculating the adaptive adjustment parameter dynamic influence weight and dividing it into same-direction or reverse linkage adjustment parameters and calculating the reference deviation value is as follows:

[0048] For each job collection parameter, all adaptive adjustment parameters of the job collection parameter are sorted to form an adaptive adjustment parameter set of the job collection parameter.

[0049] The comprehensive function-related influence values of all adaptive adjustment parameters in the adaptive adjustment parameter set are normalized to obtain the dynamic influence weight of each adaptive adjustment parameter on the job collection parameter.

[0050] The Pearson correlation coefficient of each adaptive adjustment parameter and the corresponding job collection parameter is calculated; if the correlation coefficient is positive, the adaptive adjustment parameter is marked as a same-direction linkage adjustment parameter; if the correlation coefficient is negative, the adaptive adjustment parameter is marked as a reverse linkage adjustment parameter.

[0051] The window arithmetic mean value of each adaptive adjustment parameter in the preset time window of the inspection area is calculated; the window arithmetic mean value of each adaptive adjustment parameter is subtracted from the corresponding preset reference value to obtain the reference deviation value of the adaptive adjustment parameter.

[0052] Preferably, in step three, the job execution collection parameter is analyzed and sorted into an inspection area exclusive unmanned aerial vehicle job execution collection parameter set to control the unmanned aerial vehicle to carry out inspection.

[0053] The reference deviation values of each adaptive adjustment parameter are normalized to determine the interval middle value and the reasonable range span of the preliminary reasonable range of the job collection parameter.

[0054] The job execution collection parameter is calculated in combination with the dynamic influence weight of each adaptive adjustment parameter on the job collection parameter, specifically as follows:

[0055] For each adaptive adjustment parameter in the co-directional linkage adjustment parameter subset, multiply its dynamic influence weight, the normalized benchmark deviation value and the span, and sum to obtain a co-directional adjustment sum;

[0056] For each adaptive adjustment parameter in the reverse linkage adjustment parameter subset, multiply its dynamic influence weight, the normalized benchmark deviation value and the span, and sum to obtain a reverse adjustment sum;

[0057] The final job execution collection parameter corresponding to the job collection parameter is obtained by adding the co-directional adjustment sum to the interval middle value in the preliminary reasonable range of the job collection parameter and then subtracting the reverse adjustment sum;

[0058] After completing the calculation of the job execution collection parameters corresponding to all job collection parameters in the current inspection area, all results are sorted to form a set of exclusive UAV execution job collection parameters for the current inspection area;

[0059] The UAV inspection control system loads the set of UAV execution job collection parameters and controls the UAV to carry out inspection work in the current inspection area according to the preset inspection path and work logic.

[0060] Compared with the prior art, the beneficial effects of the present application are:

[0061] (1) The power equipment defect detection method based on UAV inspection, through step one, risk values are quantified and classified by collecting historical basic parameters and real-time state parameters, step two, standard job parameter groups are screened according to risk levels and preliminary ranges are delimited, step three, dynamic weights and benchmark deviation correction parameters are calculated in combination with core functions to screen adjustment parameters, forming a complete technical closed loop; this mechanism enables risk assessment to directly guide parameter adaptation, and working condition correction to optimize the final job parameters, avoiding the problems of risk and parameter disconnection, and parameter and working condition inconsistency in traditional processes, significantly improving the inspection pertinence, accurately investing in high-risk areas, optimizing the efficiency in low-risk areas, and reducing resource waste and detection omissions.

[0062] (2) The power equipment defect detection method based on UAV inspection, through step two, standard job parameter groups are grouped according to risk levels, and the preliminary ranges of parameters of different risk levels are delimited by double outlier elimination; step three, key adjustment parameters are screened through the core demand-core function correspondence table; this design not only adapts the risk of the parameter adaptation area, but also directly guarantees the core demand, solving the drawbacks of traditional parameters that only refer to a single working condition and cannot consider both risk and demand, ensuring that the inspection effect is optimal in different areas and under different demands.

[0063] (3) The power equipment defect detection method based on unmanned aerial vehicle inspection, in step three, the key adjustment parameters are screened by the comprehensive function correlation influence value, the dynamic weight is obtained by normalizing the comprehensive influence value, the same direction / opposite direction linkage type is divided by using the Pearson correlation coefficient, and the final execution parameter is calculated by combining the reference deviation correction and the parameter preliminary range; the quantitative system replaces the traditional manual experience adjustment, so that the parameter screening has basis, the priority is clear, and the adjustment direction is controllable, the effect fluctuation caused by experience dependence is effectively avoided, and it is ensured that the unmanned aerial vehicle can load the exclusive optimal parameters for inspection under different working conditions and different regions, and the adaptability and inspection effect stability of the scheme are greatly improved. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 The flowchart of the present application. DETAILED DESCRIPTION

[0065] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0066] Embodiment one;

[0067] Please refer to Figure 1 The present application provides a power equipment defect detection method based on unmanned aerial vehicle inspection, comprising:

[0068] Step one: collect historical basic parameters and real-time state parameters in units of independent inspection areas, calculate the risk value of the inspection area, and match the risk level of the inspection area, and the specific process is as follows:

[0069] All areas that need to be inspected by the unmanned aerial vehicle are obtained, and two types of core parameters are collected in units of each independent inspection area:

[0070] The historical basic parameters include: defect occurrence frequency F in the preset period, equipment operation period Y, and component material adaptation coefficient M ((M∈[0, 1]), the higher the adaptability, the larger the value);

[0071] The real-time state parameters include: current area environment humidity H, line load coefficient L ((L∈[0, 1]), full load is 1), and environment temperature CT;

[0072] Based on the above parameters, the risk level quantization model is constructed after the normalization and dimensionless processing, the risk value R of the inspection area is calculated, and the calculation formula is as follows:

[0073] ;

[0074] Wherein, a1, a2, a3, a4, a5, a6 are preset weight coefficients, satisfying a1+a2+a3+a4+a5+a6=1;

[0075] A plurality of regional risk levels are set, and each risk level corresponds to a risk value interval, the inspection area risk value is matched with the risk value interval corresponding to all regional risk levels, and the corresponding risk level is output, wherein the higher the risk level, the greater the upper limit value and the lower limit value corresponding to the risk value interval;

[0076] Further, the component material adaptation coefficient and the line load coefficient are obtained by:

[0077] The material adaptation coefficient is obtained by:

[0078] The material type of the main equipment component in the region is collected (such as ceramic / glass / composite for insulator material), the typical operating environment type of the region (such as high humidity along the coast / dry inland / industrial area dusty), the historical failure probability of the material component under the same environment (extracted from the power equipment material reliability database), after normalization and dimensionless processing, and using the formula:

[0079] ;

[0080] Wherein, is a material preset weight coefficient; is a material-environment adaptation correction function, when the material matches the environment , the function takes value ; and when the material does not match the environment , the function takes value ; wherein, and are both positive values preset, and the correction coefficient in the matching state is less than the correction coefficient in the non-matching state, for correcting the adaptation difference of different materials in a specific environment.

[0081] The line load coefficient is obtained by:

[0082] The real-time load power of the power supply line corresponding to the inspection area is obtained through the power dispatching system interface , the rated load power of the line , the current running time of the line (from the start of this inspection to the continuous running time at the collection time), after normalization and dimensionless processing, using the formula:

[0083] ;

[0084] wherein, is a runtime correction factor, for quantifying the cumulative effect of long-time operation on load risk, the min function ensures that L does not exceed 1.0 (upper limit of full load).

[0085] It should be noted that step one provides key technical support for risk pre-control of unmanned aerial vehicle inspection through systematic parameter design and quantitative logic. Its technical advantages and actual value are reflected in the following multi-dimensional aspects:

[0086] First, the integrity of the inspection area risk assessment is achieved; this step synchronously collects historical basic parameters and real-time state parameters for each independent inspection area - the former covers defect occurrence frequency within a preset period, equipment operation period, and other indicators reflecting the long-term risk characteristics of the equipment, and the latter includes current regional environmental humidity, line load coefficient, and other dynamic working condition data. This effectively avoids the one-sidedness caused by relying solely on a single dimension (such as only looking at historical failures or only looking at real-time state) in traditional assessment, ensuring that the assessment of regional risks covers the entire life cycle of the equipment and real-time working conditions.

[0087] Second, the precision of risk quantification is ensured; for the two core influencing factors of material adaptation coefficient and line load coefficient, the calculation logic fully combines actual application scenarios: the material adaptation coefficient is quantitatively corrected by the material type, environmental type, and historical failure probability through the material-environment adaptation correction function, accurately reflecting the adaptation risk of different materials in a specific environment; the line load coefficient integrates real-time load, rated load, and the cumulative effect of runtime, and through the quantitative load risk superposition effect, it ensures the reasonableness of the parameters through boundary control; significantly improving the accuracy of core indicators, laying a foundation for reliable quantification of subsequent risk values.

[0088] Third, the technical guidance for subsequent inspection operations is established; the risk value is calculated through the weighted model, and it is matched with the preset risk level interval to clearly divide the risk level of different areas - this classification result directly provides clear basis for step two "selecting qualified operation parameter set according to risk level" and step three "adapting operation execution parameters", forming a technical closed loop of "risk classification-parameter adaptation", avoiding the problem of parameter setting and regional risk disconnection in traditional inspection, and providing support for the development of differentiated inspection strategies.

[0089] Fourthly, the inspection resource allocation and detection effect are optimized; by quantifying the regional risk in advance, the directional allocation of inspection resources can be realized - the high-risk areas can be preferentially allocated more detailed inspection parameters and resource investment, and the low-risk areas can reasonably optimize the inspection efficiency and reduce redundant consumption; at the same time, the high-risk focus is locked in advance, which lays a foundation for subsequent improvement of defect detection rate and reduction of misjudgment rate, and ensures that the inspection operation takes into account both efficiency and accuracy.

[0090] Step two: collect historical inspection data of the inspection area, extract core evaluation parameters corresponding to each operation collection parameter group of the unmanned aerial vehicle, calculate the defect detection rate and the defect misjudgment rate; according to this, the qualified operation collection parameter groups are screened out and grouped according to the risk level, and for different risk level qualified operation collection parameter group sets, the preliminary range of each operation collection parameter is analyzed and determined, and the specific process is as follows:

[0091] For each inspection area, collect the historical inspection data of the inspection area, including: the risk level of the inspection area at each inspection, the unmanned aerial vehicle operation collection parameter group (the flight control parameter and data collection parameter of the unmanned aerial vehicle when performing inspection, including: flight speed, collection frame rate, etc.) and the defect detection result data of the inspection (including: the total number of real defects in the coverage range of this inspection, the number of real defects identified and verified by rechecking in this inspection, the number of missed defects found through artificial rechecking / equipment fault feedback within a preset period after this inspection, and the number of defects identified as defects but verified as misjudgments in this inspection);

[0092] Among them, the inspection defects include: line defects: conductor crack, conductor strand break, conductor sag anomaly, conductor icing / hanging foreign matter;

[0093] Fitting defects: fitting rust, fitting deformation (such as line clamp misplacement), fitting bolt loosening / loss;

[0094] Insulator defects: insulator surface damage, insulator pollution, insulator flashover trace;

[0095] Device defects: substation device (such as transformer, circuit breaker) shell rusting / deformation, device joint overheating (infrared detection type defect), missing / fuzzy device identification.

[0096] For each operation collection parameter group, the core evaluation parameters are extracted from the corresponding defect detection result data, including: the total number of real defects, the number of effective detected defects (the number of real defect detections minus the number of misjudgment defects), and the defect misjudgment rate (the number of misjudgment defects divided by the sum of the number of real defect detections and the number of misjudgment defects);

[0097] By dividing the number of effective detected defects corresponding to each operation collection parameter group by the total number of real defects, the defect detection rate of the operation collection parameter group is obtained;

[0098] The preset defect misjudgment rate threshold and the defect detection rate threshold are used to screen the job collection parameter groups that simultaneously satisfy:

[0099] The job collection parameter groups whose defect misjudgment rate is less than the corresponding preset defect misjudgment rate threshold and whose defect detection rate is greater than or equal to the corresponding preset defect detection rate threshold are recorded as qualified job collection parameter groups;

[0100] Based on the risk levels corresponding to the qualified job collection parameter groups, the qualified job collection parameter groups are grouped according to risk levels to obtain a set of qualified job collection parameter groups corresponding to each risk level;

[0101] For each set of qualified job collection parameter groups of a risk level, the preliminary range of each job collection parameter is obtained by the following steps, specifically:

[0102] For each job collection parameter in the set of qualified job collection parameter groups, the sample data corresponding to each qualified job collection parameter group is extracted and arranged to obtain a sample sequence of the job collection parameter;

[0103] The sample sequence of the job collection parameter is subjected to outlier rejection by the interquartile range method, specifically: the first quartile Q1 and the third quartile Q3 of the samples are calculated, and the outlier boundary is defined as [Q1-1.5IQR, Q3+1.5IQR] (IQR=Q3-Q1), and the extreme samples outside the boundary are removed;

[0104] For the sample sequence after the first use of the quartile method to remove outliers, the median correlation interval secondary screening method is used to determine the initial reasonable interval of the job collection parameter; the specific steps are as follows:

[0105] First, the median of the remaining sample sequence after the first outlier rejection is calculated ; then, in combination with the distribution concentration trend of the samples under this risk level, the median and the interquartile range are used to construct a secondary screening interval , where (IQR) is the interquartile range calculated when the outliers are first removed (i.e. IQR=Q3-Q1), and all edge extreme values outside this interval range are removed;

[0106] After secondary screening, the minimum value and the maximum value of the remaining valid samples are extracted; take as the lower limit of the interval, as the upper limit of the interval, and the initial reasonable interval of the corresponding job collection parameter under this risk level is determined .

[0107] The initial reasonable intervals of all operation collection parameters corresponding to each risk level are sorted, and the one-to-one correspondence between different risk levels and the preliminary ranges of each operation collection parameter is determined to form a risk-parameter preliminary range linkage rule;

[0108] Based on the risk level of the current inspection area, it is substituted into the corresponding risk-parameter preliminary range linkage rule for matching, and the corresponding preliminary range of each operation collection parameter is output.

[0109] It should be noted that this step does not use generalized historical inspection data collection methods, but extracts three types of strongly associated core data: "inspection area risk level - unmanned aerial vehicle operation collection parameter group - defect detection result". Among them, "operation collection parameter group" focuses on flight speed, collection frame rate and other operation parameters that directly affect inspection efficiency, and "defect detection result" includes verified true defect total number, effective detection number, false positive number and other verification data, ensuring that the collected data serve the core goal of "operation parameter effectiveness evaluation" and avoiding interference of redundant data on subsequent analysis, providing a high-quality data foundation for parameter screening;

[0110] By defining two quantitative indicators "defect detection rate = effective detection defect number / true defect total number" and "defect false positive rate = false positive defect number / (true detection defect number + false positive defect number)", the actual inspection effect of the operation collection parameter group is converted into comparable and selectable numerical standards. This design can accurately exclude invalid parameter groups with high detection quantity but false positive rate exceeding the threshold (such as a parameter group with many detected defects but false positive rate exceeding the threshold, which is not included in the standard range), ensuring that the "standard operation collection parameter group" meets the requirements of "detection accuracy" and "false positive controllability", and providing an effective sample set for subsequent parameter range determination;

[0111] For abnormal value interference caused by sensor failure, extreme weather and other special scenarios in historical sample data, this step designs a double mechanism of "fourth quantile distance method first screening + median related interval second screening":

[0112] The first screening eliminates obvious extreme values (such as abnormal collection frame rate caused by sensor failure) by calculating Q1, Q3 and IQR (Q3-Q1) and eliminating values outside the range [Q1-1.5IQR, Q3+1.5IQR];

[0113] The second screening constructs the interval [Pmed-2IQR, Pmed+2IQR] based on the median Pmed of the samples after the first screening and the original IQR, and further eliminates edge abnormal data (such as irregular low-speed flight data in heavy rain weather);

[0114] The mechanism can effectively exclude the interference of individual special scene data on the parameter range, ensure that the finally drawn parameter interval (such as flight speed 5-8 m / s, acquisition frame rate 15-20 fps) reflects the needs of most normal inspection scenes, and avoid "overly wide and inaccurate" or overly narrow parameter range;

[0115] By arranging the initial reasonable range of operation collection parameters corresponding to each risk level, a one-to-one correspondence rule of "risk level-operation parameter preliminary range" is constructed, which directly converts the abstract risk classification in step one into an operable UAV operation parameter basis: for example, the high-risk area (old line dense, frequent historical defects) corresponds to "narrow flight speed range (4-6 m / s) + high acquisition frame rate range (18-22 fps)" to ensure detection accuracy, and the low-risk area (new line) corresponds to "wide flight speed range (6-10 m / s) + moderate acquisition frame rate range (12-18 fps)" to improve inspection efficiency. This rule not only provides a clear basic framework for parameter optimization in step three, but also ensures that the inspection operation in different risk areas has accurate parameter guidance, avoiding the waste of resources or detection omission caused by traditional "one-size-fits-all" parameter setting.

[0116] Step three: Obtain the core function set of the inspection area; extract the candidate adaptive adjustment parameters of the operation collection parameters, calculate their comprehensive function correlation influence value, and screen the adaptive adjustment parameters; calculate the dynamic influence weight of the adaptive adjustment parameters, and divide them into same-direction or opposite-direction linkage adjustment parameters, calculate the reference deviation value and combine the preliminary reasonable range of the operation collection parameters for correction, obtain the operation execution collection parameters, and arrange them into the exclusive UAV operation execution collection parameter set of the inspection area to control the UAV to carry out the inspection. The specific process is as follows:

[0117] Based on the core demands of each inspection area, a core demand-core function correspondence table is constructed, in which each core demand is pre-set to correspond to a core function; for example, the following correspondence table:

[0118]

[0119] For each inspection area, obtain all core demands corresponding to the current inspection area, and substitute them into the core demand-core function correspondence table for matching, output the corresponding core functions, and arrange all core functions to obtain the core function set of the inspection area;

[0120] Obtain each operation collection parameter corresponding to the current inspection area, and for each operation collection parameter, extract all parameters that may affect the correlation effect between the operation collection parameter and each core function in the core function set from related data sources such as regional working conditions, equipment status, and historical collection, and record them as candidate adaptive adjustment parameters of the operation collection parameter;

[0121] For each candidate adaptive adjustment parameter, based on the historical inspection database of the current inspection area, the historical value sequence of the candidate adaptive adjustment parameter and the historical value sequence of the quantitative indicator corresponding to each core function in the core function set of the inspection area are extracted and normalized and de-dimensioned;

[0122] Using the formula:

[0123] ;

[0124] Wherein:

[0125] : is the comprehensive function correlation influence value of the candidate adaptive adjustment parameter (the closer the value is to 1, the more significant the overall influence of the candidate adaptive adjustment parameter on each core function of the current inspection area); ∈[0,1], the value is closer to 1, the more significant the overall influence of the candidate adaptive adjustment parameter on each core function of the current inspection area);

[0126] m: is the number of core functions in the core function set of the current inspection area;

[0127] : is the preset weight of the kth core function;

[0128] : is the Pearson correlation coefficient of the normalized candidate adaptive adjustment parameter history sequence and the kth core function quantitative indicator history sequence;

[0129] : is the normalized mean of the kth core function quantitative indicator corresponding to the grouping after the candidate adaptive adjustment parameter sequence is divided into three equal parts (i.e., after sorting all historical value sequences of the candidate adaptive adjustment parameter by value, the candidate adaptive adjustment parameter is divided into 3 interval groups, and the kth core function quantitative indicator corresponding to the grouping is normalized); (the performance of the core function corresponding to the different value interval of the candidate adaptive adjustment parameter is quantified, so as to quantify the influence amplitude of the change of the candidate adaptive adjustment parameter on the core function);

[0130] max(...): takes the maximum difference of the core function quantitative indicator mean between the groups (reflects the influence amplitude of the change of the candidate adaptive adjustment parameter on the core function);

[0131] A preset comprehensive function correlation influence threshold is used to screen out candidate adaptive adjustment parameters whose comprehensive function correlation influence value is greater than or equal to the corresponding preset threshold, and mark them as adaptive adjustment parameters;

[0132] All adaptive adjustment parameters of the job collection parameters are sorted to obtain the adaptive adjustment parameter set of the job collection parameters;

[0133] For each job acquisition parameter, the corresponding set of adaptive adjustment parameters is extracted, the comprehensive functional correlation influence value of all adaptive adjustment parameters in the set is processed according to the normalization principle in the same set, the influence intensity of each adaptive adjustment parameter is converted into a relative proportion through normalization, and the dynamic influence weight of each adaptive adjustment parameter on the job acquisition parameter is obtained.

[0134] The Pearson correlation coefficient of each adaptive adjustment parameter and the job acquisition parameter is calculated by the Pearson correlation coefficient method.

[0135] If the Pearson correlation coefficient of the adaptive adjustment parameter and the job acquisition parameter is positive, the adaptive adjustment parameter is marked as a same-direction linkage adjustment parameter.

[0136] If the Pearson correlation coefficient of the adaptive adjustment parameter and the job acquisition parameter is negative, the adaptive adjustment parameter is marked as an opposite-direction linkage adjustment parameter.

[0137] The window arithmetic mean of each adaptive adjustment parameter in the preset time window of the inspection area is calculated, and each adaptive adjustment parameter is preset to correspond to a reference value. The reference deviation value of each adaptive adjustment parameter is obtained by subtracting the window arithmetic mean of each adaptive adjustment parameter from the corresponding reference value.

[0138] After normalizing the reference deviation value of each adaptive adjustment parameter, the dynamic influence weight of each adaptive adjustment parameter on the job acquisition parameter and the preliminary reasonable range of the job acquisition parameter are combined, and the formula is used:

[0139]

[0140] Wherein:

[0141] : is the final job execution acquisition parameter corresponding to the job acquisition parameter;

[0142] : is the middle value of the preliminary reasonable range interval of the job acquisition parameter;

[0143] : is the same-direction linkage adjustment parameter subset, is the opposite-direction linkage adjustment parameter subset;

[0144] : is the dynamic influence weight of the i-th adaptive adjustment parameter in the same-direction linkage adjustment parameter subset;

[0145] : is the dynamic influence weight of the j-th adaptive adjustment parameter in the opposite-direction linkage adjustment parameter subset;

[0146] : Normalized baseline deviation value of the i-th parameter in the parameter subset for co-acting adjustment;

[0147] : Normalized baseline deviation value of the j-th parameter in the parameter subset for counter-acting adjustment;

[0148] : Preliminary reasonable range span of the work collection parameter, i.e. the maximum value minus the minimum value of the preliminary reasonable range of the parameter (used to control the magnitude of the deviation impact, to avoid the parameter exceeding the reasonable range);

[0149] Complete the work execution collection parameter calculation corresponding to all work collection parameters in the current inspection area, and systematize all results to form a set of unmanned aerial vehicle execution work collection parameters exclusive to the current inspection area;

[0150] The unmanned aerial vehicle inspection control system loads the execution work collection parameter set and controls the unmanned aerial vehicle to carry out inspection work on the current inspection area according to the preset inspection path and work logic.

[0151] It should be noted that this step first converts the abstract demands of the inspection area into concrete core functions through the "core demand-core function correspondence table", and forms a core function set by screening for the current area; this design solves the problem of "demands and parameters disconnection" in traditional inspection - no longer relying on experience to set parameters, but taking core functions as anchor points to ensure that subsequent parameter adjustment always revolves around the key demands of the area (such as prioritizing the association of "regional environment adaptation guarantee" and "data collection quality guarantee" functions in high-humidity coastal areas, and prioritizing the association of "defect detection accuracy guarantee" and "defect positioning accuracy guarantee" functions in old line areas), providing clear target orientation for parameter adaptation;

[0152] For each work collection parameter, after extracting candidate adaptation adjustment parameters from working conditions, equipment, and historical data, multi-dimensional quantitative screening is achieved through the "comprehensive function correlation influence value" formula: this formula integrates core function weights, parameter and function correlation strength, and parameter change impact on function, converting "whether the parameter affects the function" into comparable numerical values; this design avoids the one-sidedness of "only looking at a single correlation indicator" in traditional screening, ensuring that the "adaptation adjustment parameters" selected are key parameters that significantly affect core functions, focusing on key variables for subsequent parameter adjustment;

[0153] By normalizing the comprehensive function correlation influence value of the adaptive adjustment parameter, the dynamic influence weight of each parameter on the operation collection parameter is obtained, solving the problem of priority level when multiple adjustment parameters; at the same time, the same direction / opposite direction linkage adjustment parameter is divided by Pearson correlation coefficient, and the adjustment direction is clear (such as the same direction parameter value increases, the operation parameter needs to be increased synchronously; the opposite direction parameter value increases, the operation parameter needs to be decreased); the design replaces the traditional "judging adjustment direction and amplitude by experience", so that the parameter adjustment has both quantitative priority basis and clear direction guidance, avoiding parameter misalignment caused by adjustment confusion;

[0154] Based on the preliminary reasonable range of the operation parameter determined in step two, combined with the "normalized reference deviation value", dynamic influence weight and linkage direction of the adaptive adjustment parameter, the final operation execution parameter is calculated by Pexec formula: by controlling the parameter not to exceed the reasonable boundary through the preliminary range span, and by adapting the real-time working condition of the current area through the same direction addition and opposite direction subtraction correction logic (such as the current wind speed is higher than the reference value, and the "wind speed" adaptive adjustment parameter is opposite direction linkage, so the operation parameter (such as flight speed) needs to be corrected downward based on the intermediate value of the preliminary range); the final "patrol area exclusive unmanned aerial vehicle execution operation collection parameter set" is formed, realizing the upgrade of the parameter from "risk level general adaptation" to "regional working condition exclusive precise adaptation", ensuring that the unmanned aerial vehicle can execute the patrol in the optimal parameter in different areas (such as high wind speed area, high humidity area, old line area), which not only guarantees the realization of the core function (such as flight safety, detection accuracy), but also avoids the poor patrol effect caused by parameter redundancy or deficiency;

[0155] In summary, step three realizes the key technical support of the invention to realize "differentiated and precise unmanned aerial vehicle patrol" by "target orientation-quantitative screening-direction priority clear-working condition adaptation correction" technical closed loop, which converts the risk classification and preliminary parameters of the previous two steps into exclusive landing parameters that meet the actual needs of the area, significantly improving the adaptability and effectiveness of the patrol scheme.

[0156] Although embodiments of the present application 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 therein without departing from the principles and spirit of the application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection, characterized in that, Includes the following steps: Step 1: Collect historical basic parameters and real-time status parameters for each independent inspection area, calculate the risk value of the inspection area, and match the risk level of the inspection area. Step 2: Collect historical inspection data of the inspection area, extract the core evaluation parameters corresponding to each operation parameter group of the UAV, and calculate the defect detection rate and defect misjudgment rate; Based on this, the parameter sets for compliance operations were selected and grouped according to risk level. For the parameter sets for compliance operations at different risk levels, the preliminary range of each operation's parameter set was analyzed and determined. The specific process for analyzing and determining the preliminary range of each set of data collection parameters for compliance operations at different risk levels is as follows: For each task data collection parameter, sample data of the corresponding qualified task data collection parameter group is extracted and organized into a sample sequence of the task data collection parameter; first, extreme samples in the sample sequence are removed by interquartile range method, and then marginal extreme values ​​are removed by median correlation interval secondary screening method. Extract the minimum and maximum values ​​of the valid samples after secondary screening, and use them as the upper and lower limits of the interval to define the initial reasonable range of the operation collection parameters under this risk level. Organize the initial reasonable ranges of all operational data collection parameters under each risk level, clarify the correspondence between risk level and preliminary range of operational data collection parameters, and form a risk-parameter preliminary range linkage rule; Substitute the risk level of the current inspection area into the linkage rule for matching, and output the preliminary range of each operation collection parameter corresponding to the area; Step 3: Obtain the core function set of the inspection area; extract candidate adaptation adjustment parameters of the operation collection parameters, calculate their comprehensive functional correlation impact value, and screen the adaptation adjustment parameters; calculate the dynamic impact weight of the adaptation adjustment parameters and classify them into unidirectional or reverse linkage adjustment parameters, calculate the benchmark deviation value and make corrections in combination with the preliminary reasonable range of the operation collection parameters to obtain the operation execution collection parameters, organize them into a set of dedicated UAV operation collection parameters for the inspection area, and control the UAV to carry out inspections; The specific process of obtaining the core function set of the inspection area, extracting candidate adaptation adjustment parameters for operation data collection, calculating their comprehensive functional correlation impact value, and screening adaptation adjustment parameters is as follows: Based on the core inspection requirements of each inspection area, a core requirement-core function correspondence table is constructed. All core requirements corresponding to the current inspection area are substituted into the table for matching, and the core functions are sorted out to obtain the core function set of the current inspection area. Obtain each operation collection parameter corresponding to the current inspection area, extract the parameters that affect the association effect between the operation collection parameter and each core function in the core function set of the current inspection area, and use them as candidate adaptation adjustment parameters for the operation collection parameter; Extract the historical value sequence of each candidate adaptation adjustment parameter and the historical value sequence of the quantitative indicators of each core function in the core function set of the current inspection area. Normalize and remove the dimensions from both types of sequences. The comprehensive functional correlation impact value of the candidate adaptation adjustment parameter is obtained by summing the maximum difference between the absolute value of the correlation coefficient between the historical value sequence of the candidate adaptation adjustment parameter and the historical value sequence of the normalized candidate adaptation adjustment parameter and the normalized mean value of the core function quantitative indicator of each group after dividing the historical value sequence of the candidate adaptation adjustment parameter into three equal parts. Candidate adaptation adjustment parameters whose comprehensive function correlation impact value is not lower than the preset corresponding threshold are selected and marked as adaptation adjustment parameters; The specific process for calculating the dynamic impact weights of the adaptation adjustment parameters, classifying them into unidirectional or inverse linkage adjustment parameters, and calculating the benchmark deviation value is as follows: For each data collection parameter, compile all the adaptation and adjustment parameters for that data collection parameter to form a set of adaptation and adjustment parameters for that data collection parameter. The comprehensive functional correlation impact value of all adaptive adjustment parameters in the adaptive adjustment parameter set is normalized to obtain the dynamic impact weight of each adaptive adjustment parameter on the data collection parameter of the operation. Calculate the Pearson correlation coefficient between each adaptation adjustment parameter and the corresponding job acquisition parameter; If the correlation coefficient is positive, the adaptation adjustment parameter is marked as a same-direction linkage adjustment parameter; if the correlation coefficient is negative, the adaptation adjustment parameter is marked as a reverse linkage adjustment parameter. Calculate the window arithmetic mean of each adaptive adjustment parameter within the preset time window of the inspection area; subtract the window arithmetic mean of each adaptive adjustment parameter from the corresponding preset benchmark value to obtain the benchmark deviation value of the adaptive adjustment parameter; The specific process of obtaining the operation execution parameters, organizing them into a set of dedicated drone operation parameters for the inspection area, and controlling the drone to carry out the inspection is as follows: Normalize the baseline deviation values ​​of each adaptation adjustment parameter, and determine the intermediate value and reasonable range of the initial reasonable range of the operation acquisition parameters; The job execution acquisition parameters are calculated by combining the dynamic influence weights of each adaptation adjustment parameter on the job acquisition parameters, specifically as follows: For each adaptive adjustment parameter within the subset of parameters for same-direction linkage adjustment, its dynamic influence weight, normalized benchmark deviation value, and the above span are multiplied together to obtain the total same-direction adjustment; For each adaptive adjustment parameter within the subset of reverse linkage adjustment parameters, its dynamic influence weight, normalized benchmark deviation value, and the above span are multiplied together to obtain the total reverse adjustment. The final job execution acquisition parameters are obtained by adding the midpoint of the interval of the preliminary reasonable range of the job acquisition parameters to the sum of the same-direction adjustments, and then subtracting the sum of the reverse adjustments. After completing the calculation of the operation execution collection parameters corresponding to all operation collection parameters in the current inspection area, all results are sorted out to form a set of drone operation collection parameters specific to the current inspection area. The UAV inspection control system loads the set of parameters collected by the UAV for the operation and controls the UAV to carry out inspection operations in the current inspection area according to the preset inspection path and operation logic.

2. The method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 1, characterized in that: In step one, the specific process of calculating the risk value of the inspection area and matching the risk level of the inspection area is as follows: Acquire all areas to be inspected by the drone, and collect two types of core parameters for each independent inspection area; The first category is historical basic parameters, which include the frequency of defect occurrence within a preset period, the equipment operating years, and the component material compatibility coefficient. The second category is real-time status parameters, which include the current regional ambient humidity, line load factor, and ambient temperature. Based on the historical basic parameters and real-time status parameters, a risk level quantification model is constructed. The frequency of defect occurrence within a preset period, the equipment operating period, the result of subtracting the component material compatibility coefficient from 1, the current area humidity, the line load coefficient, and the ambient temperature are multiplied by their respective preset weight coefficients. Then, all the product results are summed to obtain the risk value of the inspection area. Several regional risk levels are preset, and a risk value range is configured for each regional risk level; The calculated risk value of the inspection area is compared with the risk value range corresponding to the risk level of all areas to determine the risk value range to which the risk value of the inspection area belongs, and then the risk level corresponding to the inspection area is output.

3. The method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 2, characterized in that: Step one, the process of obtaining the material compatibility factor: Collect the material type of the main equipment components in the current inspection area, the typical operating environment type of the current inspection area, and the historical failure probability of the material of the main equipment components in the same environment as the typical operating environment type of the current inspection area; The material has a preset weight coefficient, and a material-environment adaptation correction function is set. When the material type of the main equipment component matches the typical operating environment type of the current inspection area, the material-environment adaptation correction function takes the first preset positive value. When the material type of the main equipment component does not match the typical operating environment type of the current inspection area, the material-environment adaptation correction function takes a second preset positive value; wherein, the first preset positive value is less than the second preset positive value; The material adaptation coefficient is obtained by subtracting the product of the material preset weight coefficient, the historical failure probability, and the material-environment adaptation correction function value from 1.

4. The method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 3, characterized in that: Step one, the process of obtaining the line load factor: Obtain the real-time load power of the power supply line corresponding to the current inspection area, the rated load power of the line, and the current running time of the line from the start of this inspection to the time of parameter acquisition; The real-time load power, rated load power and current running time are normalized and dimensionless to eliminate the dimensional differences between different parameters. A preset runtime correction factor is used. This factor is multiplied by 1 plus the logarithm of the current runtime to quantify the cumulative effect of long-term operation on line load risk. The ratio of normalized real-time load power to rated load power is multiplied by 1 and then added to the above-mentioned cumulative effect quantification result to obtain the preliminary calculated value of the line load factor. The smaller value between the preliminary calculated value and 1.0 is selected as the line load factor for this power supply line.

5. The method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 4, characterized in that: In step two, the specific process of collecting historical inspection data of the inspection area, extracting the core evaluation parameters corresponding to each parameter group collected by the UAV, and calculating the defect detection rate and defect misjudgment rate is as follows: For each inspection area, historical inspection data is collected, including: the risk level of the inspection area at each inspection, the parameter set collected by the drone operation, and the defect detection results data of the inspection. For each set of operation parameters, core evaluation parameters are extracted from the corresponding defect detection results data, including: total number of actual defects, number of effectively detected defects, and defect false positive rate. The defect detection rate of each task acquisition parameter group is obtained by dividing the number of valid detected defects corresponding to each task acquisition parameter group by the total number of actual defects.

6. The method for detecting defects in power equipment based on unmanned aerial vehicle (UAV) inspection according to claim 5, characterized in that: Step two involves screening the parameter groups for compliant operations and grouping them according to risk level. Filter from all job parameter sets that simultaneously meet the following conditions: The operation data collection parameter group whose defect false judgment rate is less than the corresponding preset defect false judgment rate threshold and whose defect detection rate is greater than or equal to the corresponding preset defect detection rate threshold is recorded as the qualified operation data collection parameter group. Based on the risk level corresponding to each set of parameters for compliance operations, all sets of parameters for compliance operations are grouped according to risk level to obtain a set of parameters for compliance operations corresponding to each risk level.

Images