Inspection border crossing early warning method and early warning system applied to power distribution network unmanned aerial vehicle

Abstract

The invention relates to the technical field of unmanned aerial vehicle inspection early warning, in particular to an inspection border crossing early warning method and early warning system applied to a power distribution network unmanned aerial vehicle, and the method comprises the steps: obtaining an inspection path of the unmanned aerial vehicle, measuring the position of the unmanned aerial vehicle and the electromagnetic field intensity of the position in real time, and measuring the signal-to-noise ratio of a communication link of the position of the unmanned aerial vehicle in real time; constructing a feature matrix at the current moment; obtaining the sensitivity of the unmanned aerial vehicle path to the electromagnetic field at the current moment; determining the dispersion degree of the path deviation, the first correlation between the electromagnetic field intensity and the path deviation and the second correlation between the electromagnetic field intensity and the signal-to-noise ratio at the current moment and the adjacent moment; determining a weighted fusion factor at the current moment; and in combination with the weighted fusion factor and the sensitivity, obtaining an electromagnetic influence coefficient at the current moment, and carrying out early warning on the inspection boundary crossing of the unmanned aerial vehicle. Therefore, the accuracy of unmanned aerial vehicle inspection border crossing early warning is improved.

Description

Technical Field

[0001] The present application relates to the field of UAV inspection and early warning technology, and specifically to an inspection and out-of-bounds early warning method and early warning system applied to distribution network UAVs. Background Art

[0002] The inspection cross-boundary warning technology of distribution network drones is an important breakthrough in the intelligent operation and maintenance of power systems. Its core significance lies in ensuring inspection safety, improving efficiency and reducing risks through precise boundary control. Although drones can break through spatial limitations, if they mistakenly enter live areas, cross safe distances or invade no-fly zones, they may cause equipment discharge, collision accidents, etc. The application of cross-boundary warning technology can reduce the risk of human operational errors and avoid equipment damage.

[0003] During drone inspections near high-voltage cables, the coupling of strong electromagnetic fields with multiple factors such as GPS signal errors and wind disturbances leads to complex sources of path deviations. Traditional methods rely on a single sensor or static models and are difficult to dynamically distinguish interference from different factors. Existing technologies are limited by a single data dimension and lack a dynamic weight allocation mechanism for multi-source interference. They are unable to effectively quantify the actual impact of electromagnetic interference in complex environments, resulting in poor adaptability in the event of sudden electromagnetic pulses or communication link fluctuations. It is difficult to ensure the stability and safety of drones in strong electromagnetic environments, resulting in low accuracy in drone inspection cross-border warnings. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide an inspection crossing warning method and warning system for distribution network drones. The technical solutions adopted are as follows: In a first aspect, an embodiment of the present application provides a patrol crossing warning method for a distribution network drone, the method comprising the following steps: Obtain the inspection path of the drone, measure the location of the drone and the electromagnetic field strength at the location in real time, and measure the signal-to-noise ratio of the communication link at the location of the drone in real time; Based on the inspection path and the measured position of the drone, determine the path deviation of the drone at each moment; construct a feature matrix at the current moment using the electromagnetic field strength and the rate of change of the electromagnetic field strength at the current moment and at each moment before, the path deviation and the rate of change of the path deviation at the current moment and at each moment before; Through the feature matrix at the current moment, taking the inspection path as the target value and combining the integrated learning algorithm, the prediction deviation, mean square error, and determination coefficient at the current moment are obtained, and the sensitivity of the drone path to the electromagnetic field at the current moment is obtained; Determine the discrete degree of the path deviation at the current moment and its adjacent moments, the first correlation between the electromagnetic field strength and the path deviation, and the second correlation between the electromagnetic field strength and the signal-to-noise ratio; After blurring and deblurring the discrete degree, the first correlation, and the second correlation, the output results are used as the first weight, the second weight, and the third weight, and the weighted fusion factor at the current moment is determined by combining the path deviation, the electromagnetic field strength, and the signal-to-noise ratio at the current moment; The weighted fusion factor is combined with the sensitivity to obtain the electromagnetic influence coefficient at the current moment, and an early warning is issued for the inspection crossing of the UAV.

[0005] In one embodiment, the path deviation is the shortest distance between the measurement position of the drone at each moment and the inspection path.

[0006] In one embodiment, the construction of the feature matrix at the current moment includes: For the current moment and each moment before it, the mean of the electromagnetic field strength at each moment and its adjacent moments is calculated, the electromagnetic field change gradient at each moment is calculated, and the path deviation change rate at each moment is calculated. The electromagnetic field strength, the mean, the electromagnetic field change gradient, the path deviation, and the path deviation change rate at all moments are respectively formed into column vectors to obtain a characteristic matrix.

[0007] In one embodiment, the determination of the sensitivity includes: The ratio of the determination coefficient to the mean square error at the current moment is calculated, and the difference between the prediction deviation and the path deviation at the current moment is calculated. The sensitivity is positively correlated with the ratio and negatively correlated with the difference.

[0008] In one embodiment, the difference is calculated as follows: Calculate the sum of the path deviation at the current moment and a preset value greater than 0, calculate the ratio of the predicted deviation at the current moment to the sum, record it as the first ratio, and the difference is the absolute value of the difference between the first ratio and the value 1.

[0009] In one embodiment, the blurring process includes: Calculating the discrete degree, the first correlation, and the second correlation at each moment, obtaining the maximum value of the discrete degree at the current moment and all previous moments, dividing the numerical interval from 0 to the maximum value into three fuzzy sets of low, medium, and high, and obtaining the fuzzy set to which the discrete degree at the current moment belongs according to the interval in which the discrete degree at the current moment is located; For the first correlation and the second correlation at the current moment, fuzzy processing is performed using a fuzzy processing process that is the same as the discrete degree at the current moment.

[0010] In one embodiment, the output result as the first weight, the second weight, and the third weight includes: The output results of the fuzzy sets of the discrete degree, the first correlation, and the second correlation at the current moment obtained by using the centroid method are used as the first weight, the second weight, and the third weight respectively.

[0011] In one embodiment, the determination of the weighted fusion factor includes: Calculate the product of the normalized value of the path deviation at the current moment and the first weight, calculate the product of the normalized value of the electromagnetic field strength at the current moment and the second weight, calculate the product of the normalized value of the signal-to-noise ratio at the current moment and the third weight, and the weighted fusion factor is the cumulative sum of the product, the multiplication result, and the product.

[0012] In one embodiment, the obtaining of the electromagnetic influence coefficient at the current moment and issuing an early warning for the inspection crossing of the drone include: The electromagnetic influence coefficient is the product of the sensitivity and the weighted fusion factor, recorded as the first product. When the normalized value of the first product is greater than a preset threshold, it is determined that the drone has a risk of crossing the inspection boundary. Otherwise, it is determined that the drone does not have a risk of crossing the inspection boundary.

[0013] In a second aspect, an embodiment of the present application also provides an inspection and crossing-boundary warning system for distribution network drones, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of any one of the above-mentioned methods when executing the computer program.

[0014] This application has at least the following beneficial effects: The present application obtains the inspection path of the drone, measures the position of the drone and the electromagnetic field strength at its location in real time, and measures the signal-to-noise ratio of the communication link at the drone's location in real time; improves the robustness of the final drone inspection cross-border judgment by collecting multi-source data; determines the path deviation of the drone at each moment based on the inspection path and the measured position of the drone; constructs a feature matrix at the current moment using the electromagnetic field strength at the current moment and the previous moments and the rate of change of the electromagnetic field strength, the path deviation at the current moment and the previous moments and the rate of change of the path deviation; obtains the prediction deviation, mean square error, and determination coefficient at the current moment through the feature matrix at the current moment, taking the inspection path as the target value, and combining the integrated learning algorithm to obtain the sensitivity of the drone path to the electromagnetic field at the current moment; improves the sensitivity of the drone path deviation recognition and enhances the accuracy of the drone inspection cross-border assessment by establishing a quantitative model of the impact of electromagnetic interference on the drone's flight trajectory; determines the current moment and its adjacent moments The discrete degree of the path deviation, the first correlation between the electromagnetic field strength and the path deviation, and the second correlation between the electromagnetic field strength and the signal-to-noise ratio; the discrete degree, the first correlation, and the second correlation are fuzzy processed and then re-output, and the output results are used as the first weight, the second weight, and the third weight. The uncertainty is processed by fuzzy logic, which solves the decision noise caused by environmental parameter fluctuations, reduces the impact of noise, and enhances the adaptability and reliability of feature extraction; combined with the path deviation, the electromagnetic field strength, and the signal-to-noise ratio at the current moment, the weighted fusion factor at the current moment is determined; the weighted fusion factor reflects the degree of influence of multiple factors on the path deviation of the drone. Combined with the weighted fusion factor and the sensitivity, the electromagnetic influence coefficient at the current moment is obtained, which effectively quantifies the impact of electromagnetic interference on the deviation of the drone inspection path, warns of the drone's inspection crossing the boundary, improves the stability and safety of the drone in a strong electromagnetic environment, and enhances the accuracy of the drone inspection crossing the boundary warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 A flowchart of the steps of a patrol crossing warning method for a distribution network drone provided in one embodiment of the present application; Figure 2 Flowchart for determining the electromagnetic influence coefficient. DETAILED DESCRIPTION

[0017] In order to further explain the technical means and effects adopted by this application to achieve the predetermined invention purpose, the following, in combination with the accompanying drawings and preferred embodiments, describes in detail the specific implementation methods, structures, features and effects of the inspection and crossing-border warning method and warning system for distribution network drones proposed in this application. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0018] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0019] The specific scheme of the inspection crossing-boundary warning method and warning system for UAVs in distribution networks provided by the present application is described in detail below with reference to the accompanying drawings.

[0020] See also Figure 1 , which shows a flowchart of a method for early warning of inspection crossing the boundary of a distribution network drone provided by an embodiment of the present application, the method comprising the following steps: S1, obtain the inspection path of the drone, measure the position of the drone and the electromagnetic field strength at the location in real time, and measure the signal-to-noise ratio of the communication link at the location of the drone in real time.

[0021] This embodiment uses high-precision GPS equipment and combines it with GIS software to conduct on-site mapping of the distribution network area, obtain the precise longitude and latitude coordinates of the boundary of the distribution network area, pre-set the inspection path of the drone within the boundary of the distribution area, install high-precision GPS sensors and electromagnetic field sensors on the drone, the high-precision GPS sensor is used to obtain the real-time GPS longitude and latitude of the drone, the electromagnetic field sensor is used to obtain the real-time electromagnetic field strength of the drone's location, and install a wireless network analyzer in the drone to collect the signal-to-noise ratio of the communication link in real time.

[0022] It should be noted that the GPS latitude and longitude of the drone, the electromagnetic field strength, and the signal-to-noise ratio of the communication link collected in this embodiment are all collected synchronously, and the collection frequency is set to 10 Hz. The implementer can set it according to the actual situation, and this embodiment does not limit it here.

[0023] S2, based on the inspection path and the measured position of the UAV, determine the path deviation of the UAV at each moment; use the electromagnetic field strength and the rate of change of the electromagnetic field strength at the current moment and the previous moments, the path deviation and the rate of change of the path deviation at the current moment and the previous moments to construct a feature matrix at the current moment.

[0024] This embodiment calculates the shortest distance between the GPS longitude and latitude of the drone at each moment and the inspection path of the pre-set drone as the path deviation of the drone at each moment, normalizes the path deviation at the current moment and all previous moments, and forms a deviation sequence in chronological order, normalizes the electromagnetic field intensity at the current moment and all previous moments, and forms an electromagnetic field sequence in chronological order, and normalizes the signal-to-noise ratio at the current moment and all previous moments, and forms a signal-to-noise ratio sequence in chronological order. Among them, the purpose of normalization is to eliminate the dimensional influence between data. The normalization methods in this embodiment all use Sigmoid functions. The implementer can choose other existing feasible normalization methods at will, and this embodiment does not limit it here.

[0025] In the complex electromagnetic environment near high-voltage cables, the deviation of the drone inspection path may be caused by a variety of factors, including electromagnetic interference, GPS signal error, wind disturbance, etc. In order to accurately distinguish the influence of electromagnetic field interference and other factors on the deviation, this embodiment will establish a correlation model between the electromagnetic field and the path deviation through historical data. The historical data of the electromagnetic field sequence and the deviation sequence can reflect the dynamic relationship between the change of electromagnetic field intensity and the path deviation, thereby helping to quantify the influence of electromagnetic interference on the current path deviation, so as to improve the accuracy of the cross-border warning and ensure the stability and reliability of the drone in a strong electromagnetic field environment.

[0026] This embodiment uses the deviation sequence and electromagnetic field sequence at the current moment as historical data for modeling, uses the first 40% of the data of the deviation sequence and the electromagnetic field sequence as the training set, and the last 60% of the data as the validation set. Then, for the current moment, extract the feature matrix that can reflect the influence of the electromagnetic field on the path deviation. In this embodiment, the extracted feature matrix includes: the electromagnetic field intensity at the current moment and the moments before it, the mean of the electromagnetic field intensity at each moment and its adjacent moments, the gradient of the electromagnetic field change between each moment and the moment before it, the path deviation at each moment, and the path deviation change rate between each moment and the moment before it. Thus, this embodiment obtains the current moment size of The feature matrix is ​​shown in Table 1, where N represents the number of the current moment and all previous moments.

[0027]

[0028] In Table 1, It represents the eigenvalue of feature 5 at the acquisition time N. Among them, the electromagnetic field intensity is recorded as feature 1, the mean value of the electromagnetic field intensity is recorded as feature 2, the electromagnetic field change gradient is recorded as feature 3, the path deviation is recorded as feature 4, and the path deviation change rate is recorded as feature 5.

[0029] It should be noted that the mean value of the electromagnetic field strength calculated in this embodiment is the mean value of the electromagnetic field strength at each moment and the three adjacent moments before it. The implementer can set the number of adjacent moments and the division ratio of the training set and the verification set according to actual conditions, and this embodiment does not impose any restrictions on this.

[0030] S3, through the feature matrix at the current moment, taking the inspection path as the target value, combined with the integrated learning algorithm, obtains the prediction deviation, mean square error, and determination coefficient at the current moment, and obtains the sensitivity of the drone path to the electromagnetic field at the current moment.

[0031] This embodiment uses the feature matrix at the current moment as input, the preset inspection path of the drone as the target value, sets the number of trees to 100, and sets the maximum depth to 8 to avoid overfitting. The random forest regression model is used to train the model through the training set to predict the path deviation, that is, for the current moment t, the predicted deviation at that moment is output, and the mean square error of the prediction is calculated through the random forest regression model. and the coefficient of determination Among them, the random forest regression model is an existing well-known technology, and the implementer can choose other existing feasible regression models, such as the gradient boosting regression model, etc. The implementer can set the training parameters of the random forest regression model according to the actual situation, and this embodiment does not limit it here.

[0032] Based on the above analysis, the sensitivity of the UAV path to the electromagnetic field at the current time t is calculated. , the specific calculation formula is: ; In the formula, is the sensitivity of the UAV path to the electromagnetic field at the current time t, is the determination coefficient output by the random forest regression model at the current time t, is the mean square error output by the random forest regression model at the current time t, is the prediction deviation output by the random forest regression model at the current time t, is the path deviation at the current time t, To preset a value greater than 0 to avoid the denominator being 0, in this embodiment , the implementer can set it according to the actual situation, and this embodiment does not limit it here, and e is a natural constant. Recorded as the first ratio.

[0033] It should be understood that the coefficient of determination It reflects the predictive ability of electromagnetic field characteristics for the current path deviation. When it is closer to 1, the electromagnetic field change can explain most of the path deviation fluctuations, indicating that electromagnetic interference is the main factor. At this time, the stronger the random forest regression model's ability to capture electromagnetic interference, the greater the impact of electromagnetic interference, and the larger the A value; mean square error Measures the prediction accuracy of the random forest regression model in a local time window, mean square error The larger the value, the worse the prediction ability of the random forest regression model, the lower the model confidence, and the smaller the A value; In the equation, the numerator represents the model prediction deviation, and the denominator represents the actual deviation. The smaller the difference between the ratio of the two and the value 1, the higher the consistency between the model prediction deviation and the actual deviation, the greater the impact of electromagnetic interference on the drone path, and the larger the A value.

[0034] The sensitivity quantifies the intensity of the influence of the electromagnetic field on the path deviation at the current moment. The higher the value of the sensitivity, the more sensitive the deviation of the drone's driving path at the current moment is to electromagnetic interference.

[0035] In another embodiment, the sensitivity of the drone path to the electromagnetic field at the current time t The calculation method is: ; In the formula, is the sensitivity of the UAV path to the electromagnetic field at the current time t, is the determination coefficient output by the random forest regression model at the current time t, is the mean square error output by the random forest regression model at the current time t, is the prediction deviation output by the random forest regression model at the current time t, is the path deviation at the current time t, and e is a natural constant.

[0036] S4, determining the discrete degree of the path deviation at the current moment and its adjacent moments, the first correlation between the electromagnetic field strength and the path deviation, and the second correlation between the electromagnetic field strength and the signal-to-noise ratio.

[0037] During the drone inspection process, the impact of electromagnetic interference on path deviation is often a complex problem with multiple coupled factors. Although the interference intensity of the electromagnetic field can be preliminarily quantified through a single-dimensional analysis, in actual scenarios, path deviation may also be affected by the quality of the communication link, environmental noise, and the dynamic characteristics of the drone itself. Therefore, in order to more comprehensively evaluate the overall impact of electromagnetic interference on drone inspections, it is necessary to simultaneously consider the dynamic relationship between electromagnetic field strength, signal-to-noise ratio changes, and path deviation, so as to improve the accuracy of interference identification and ensure the stable operation of drones in complex electromagnetic environments.

[0038] This embodiment presets a time window at each moment, and one time window includes 100 moments. The length of the time window can be set by the implementer. That is, in this embodiment, for any moment, 99 moments adjacent to the any moment are selected to form the time window of the any moment together with the any moment.

[0039] For the current moment, the discrete degree of the path deviation at all moments in the time window of the current moment is calculated, the correlation between the electromagnetic field strength at all moments in the time window of the current moment and the path deviation is calculated, which is recorded as the first correlation, and the correlation between the electromagnetic field strength at all moments in the time window of the current moment and the signal-to-noise ratio is calculated, which is recorded as the second correlation.

[0040] It should be noted that the degree of dispersion in this embodiment is calculated using the variance, and the implementer can choose other existing feasible calculation methods, such as standard deviation, coefficient of variation, etc. The correlation in this embodiment is calculated using covariance, and the implementer can choose other existing feasible calculation methods, such as Pearson correlation coefficient, cosine similarity, etc.

[0041] S5, after blurring and deblurring the discrete degree, the first correlation and the second correlation respectively, output the results as the first weight, the second weight and the third weight, and combine the path deviation, the electromagnetic field strength and the signal-to-noise ratio at the current moment to determine the weighted fusion factor at the current moment.

[0042] Obtain the discrete degree of all moments before the current moment, and count the maximum value of all the discrete degrees, record the maximum value as x, and divide the discrete degree of all moments before the current moment into three fuzzy sets: low, medium, and high. The value range of the low fuzzy set is [0, 0.4x], the coverage range of the medium fuzzy set is [0.3x, 0.7x], and the coverage range of the high fuzzy set is set to [0.6x, x]. For each fuzzy set, count the mode within the range, and use the left boundary, mode, and right boundary of each fuzzy set as the left vertex, peak point, and right vertex of the triangular membership function of the fuzzy set, respectively. Use the triangular membership function to calculate the membership of the discrete degree of the current moment in each fuzzy level, and realize the fuzzy mapping of the discrete degree of the current moment. The setting of the value range of the three fuzzy sets of low, medium, and high can be set by the implementer according to the actual situation.

[0043] The first correlation and the second correlation at the current moment are fuzzified by using the same fuzzy processing method as the discrete degree. Thus, the fuzzy sets of the discrete degree, the first correlation and the second correlation at the current moment can be obtained respectively.

[0044] The output result of the centroid method is used as the first weight for the fuzzy set of the discrete degree at the current moment, the output result of the centroid method is used as the second weight for the fuzzy set of the first correlation at the current moment, and the output result of the centroid method is used as the third weight for the fuzzy set of the second correlation at the current moment. The centroid method is a well-known technology, and the specific process is not described in detail.

[0045] It should be understood that the values ​​of path deviation, electromagnetic field strength, and the signal-to-noise ratio may fluctuate due to sensor noise or instantaneous interference, resulting in deviations in the physical meaning represented by the calculated discrete degree and correlation. By fuzzifying and defuzzifying them, even if affected by interference, as long as the membership is in the same fuzzy set, the physical meaning and information expression they represent will remain consistent, reducing the impact of other factors and interference.

[0046] Based on the above analysis, this embodiment calculates the weighted fusion factor at the current moment. The specific calculation method is: In the formula, is the weighted fusion factor at the current time t, is the first weight, is the path deviation at the current time t, Norm() is the normalization function, is the second weight, is the electromagnetic field strength at the current time t, is the third weight, is the signal-to-noise ratio at the current time t.

[0047] It should be understood that the weighted fusion factor is used to quantify the combined impact of the electromagnetic field strength and the signal-to-noise ratio on the path deviation within the time window at the current moment. , , They respectively represent the combined contribution weights of path deviation, electromagnetic field strength and signal-to-noise ratio, reflecting the relative importance of each factor on the path deviation within the current time window. The larger the value, the greater the influence of the corresponding factor in the current time window on the path deviation.

[0048] S6, combining the weighted fusion factor and the sensitivity to obtain the electromagnetic influence coefficient at the current moment, and issuing an early warning for the inspection crossing of the UAV.

[0049] In this embodiment, the product of the weighted fusion factor and the sensitivity at the current moment is recorded as the first product, which is used as the electromagnetic influence coefficient at the current moment. The electromagnetic influence coefficient describes the overall impact of electromagnetic interference on the deviation of the UAV inspection path in a complex electromagnetic environment, identifies the degree of correlation between electromagnetic interference and path deviation, and distinguishes the impact of electromagnetic interference and other external factors. The higher the electromagnetic influence coefficient value, the worse the stability of the UAV in a strong electromagnetic environment, and the more likely it is that there will be a risk of inspection crossing the boundary. The electromagnetic influence coefficient determination flow chart is as follows: Figure 2 shown.

[0050] Based on the above analysis, this embodiment sets a patrol crossing threshold, recorded as a preset threshold. If the normalized value of the electromagnetic influence coefficient at the current moment is greater than the patrol crossing threshold, it is determined that the drone has a patrol crossing risk. Otherwise, it is determined that the drone does not have a patrol crossing risk. In this embodiment, the patrol crossing threshold is set to 0.8, and the implementer can set it according to the actual situation. This embodiment does not limit it here.

[0051] This embodiment uses the Sigmoid function to obtain the normalized value of the electromagnetic influence coefficient, and the implementer can choose other existing feasible normalization methods.

[0052] Based on the same inventive concept as the above method, an embodiment of the present application also provides an inspection and cross-boundary warning system applied to distribution network drones, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of any one of the above-mentioned inspection and cross-boundary warning methods applied to distribution network drones are implemented.

[0053] It should be noted that the above sequence of the embodiments of the present application is for description only and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0054] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0055] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application should be included in the protection scope of the present application.

Claims

1. A patrol crossing warning method for a UAV in a distribution network, characterized in that: The method comprises the following steps: Obtain the inspection path of the drone, measure the location of the drone and the electromagnetic field strength at the location in real time, and measure the signal-to-noise ratio of the communication link at the location of the drone in real time; Based on the inspection path and the measured position of the drone, determine the path deviation of the drone at each moment; construct a feature matrix at the current moment using the electromagnetic field strength and the rate of change of the electromagnetic field strength at the current moment and at each moment before, the path deviation and the rate of change of the path deviation at the current moment and at each moment before; Through the feature matrix at the current moment, taking the inspection path as the target value and combining the integrated learning algorithm, the prediction deviation, mean square error, and determination coefficient at the current moment are obtained, and the sensitivity of the drone path to the electromagnetic field at the current moment is obtained; Determine the discrete degree of the path deviation at the current moment and its adjacent moments, the first correlation between the electromagnetic field strength and the path deviation, and the second correlation between the electromagnetic field strength and the signal-to-noise ratio; After blurring and deblurring the discrete degree, the first correlation, and the second correlation, the output results are used as the first weight, the second weight, and the third weight, and the weighted fusion factor at the current moment is determined by combining the path deviation, the electromagnetic field strength, and the signal-to-noise ratio at the current moment; The weighted fusion factor is combined with the sensitivity to obtain the electromagnetic influence coefficient at the current moment, and an early warning is issued for the inspection crossing of the UAV.

2. The inspection crossing warning method for a distribution network drone as claimed in claim 1 is characterized in that: The path deviation is the shortest distance between the measurement position of the UAV at each moment and the inspection path.

3. The inspection and crossing-boundary warning method for a distribution network drone as claimed in claim 1 is characterized in that: The construction of the feature matrix at the current moment includes: For the current moment and each moment before it, the mean of the electromagnetic field strength at each moment and its adjacent moments is calculated, the electromagnetic field change gradient at each moment is calculated, and the path deviation change rate at each moment is calculated. The electromagnetic field strength, the mean, the electromagnetic field change gradient, the path deviation, and the path deviation change rate at all moments are respectively formed into column vectors to obtain a characteristic matrix.

4. The inspection crossing warning method for a distribution network drone as claimed in claim 1 is characterized in that: The determination of sensitivity includes: The ratio of the determination coefficient to the mean square error at the current moment is calculated, and the difference between the prediction deviation and the path deviation at the current moment is calculated. The sensitivity is positively correlated with the ratio and negatively correlated with the difference.

5. The inspection crossing warning method for a distribution network drone as claimed in claim 4 is characterized in that: The difference is calculated as: Calculate the sum of the path deviation at the current moment and a preset value greater than 0, calculate the ratio of the predicted deviation at the current moment to the sum, record it as the first ratio, and the difference is the absolute value of the difference between the first ratio and the value 1.

6. The inspection crossing-boundary warning method for a distribution network drone as claimed in claim 1, characterized in that: The fuzzy processing includes: Calculating the discrete degree, the first correlation, and the second correlation at each moment, obtaining the maximum value of the discrete degree at the current moment and all previous moments, dividing the numerical interval from 0 to the maximum value into three fuzzy sets of low, medium, and high, and obtaining the fuzzy set to which the discrete degree at the current moment belongs according to the interval in which the discrete degree at the current moment is located; For the first correlation and the second correlation at the current moment, fuzzy processing is performed using a fuzzy processing process that is the same as the discrete degree at the current moment.

7. The inspection crossing warning method for a distribution network drone as claimed in claim 6 is characterized in that: The output results are used as the first weight, the second weight, and the third weight, including: The output results of the fuzzy sets of the discrete degree, the first correlation, and the second correlation at the current moment obtained by using the centroid method are used as the first weight, the second weight, and the third weight respectively.

8. The inspection crossing warning method for a distribution network drone as claimed in claim 1 is characterized in that: The determination of the weighted fusion factor includes: Calculate the product of the normalized value of the path deviation at the current moment and the first weight, calculate the product of the normalized value of the electromagnetic field strength at the current moment and the second weight, calculate the product of the normalized value of the signal-to-noise ratio at the current moment and the third weight, and the weighted fusion factor is the cumulative sum of the product, the multiplication result, and the product.

9. The inspection crossing warning method for a distribution network drone as claimed in claim 1, characterized in that: The electromagnetic influence coefficient at the current moment is obtained, and an early warning is issued for the inspection crossing of the UAV, including: The electromagnetic influence coefficient is the product of the sensitivity and the weighted fusion factor, recorded as the first product. When the normalized value of the first product is greater than a preset threshold, it is determined that the drone has a risk of crossing the inspection boundary. Otherwise, it is determined that the drone does not have a risk of crossing the inspection boundary.

10. A patrol crossing warning system for a distribution network drone, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 9 are implemented.

Images