CN113110072B

CN113110072B - Unmanned aerial vehicle residual power real-time monitoring method and system based on smart lamp post

Info

Publication number: CN113110072B
Application number: CN202110658213.3A
Authority: CN
Inventors: 杨翰翔
Original assignee: Shenzhen Lianhe Intelligent Technology Co ltd
Current assignee: Shenzhen Lianhe Intelligent Technology Co ltd
Priority date: 2021-06-15
Filing date: 2021-06-15
Publication date: 2021-09-03
Anticipated expiration: 2041-06-15
Also published as: CN113110072A

Abstract

The invention relates to the technical field of unmanned aerial vehicles, in particular to an unmanned aerial vehicle residual electricity quantity real-time monitoring method and system based on a smart lamp post. The electric quantity of the unmanned aerial vehicle in the target monitoring task time period is obtained based on at least two pieces of second monitoring log information, the second monitoring log information can be reference monitoring log information, compared with a scheme of adopting single reference monitoring log information, reference data are richer, and the accuracy of the electric quantity of the unmanned aerial vehicle is improved; and the first target unmanned aerial vehicle monitoring characteristic information sequence is obtained after screening the reference monitoring log information based on the degree of correlation, the basic data size of the reference monitoring log information of the unmanned aerial vehicle electric quantity used for predicting the target monitoring task time interval is simplified, and the first target monitoring log information screened based on the degree of correlation can obtain the second monitoring log information strongly correlated with the unmanned aerial vehicle electric quantity in the target monitoring task time interval, so that the electric quantity prediction precision is further improved.

Description

Unmanned aerial vehicle residual power real-time monitoring method and system based on smart lamp post

Technical Field

The invention relates to the technical field of unmanned aerial vehicles, in particular to an unmanned aerial vehicle residual power real-time monitoring method and system based on a smart lamp post.

Background

Along with the development of city wisdom lamp pole, wisdom lamp pole presents multi-functionalization development trend. Be provided with unmanned aerial vehicle air park and corresponding unmanned aerial vehicle charging device on the top of wisdom lamp pole, can park, descend to charge or change the battery for unmanned aerial vehicle so that unmanned aerial vehicle lasts the flight automatically for unmanned aerial vehicle.

Unmanned Aerial vehicles (Unmanned Aerial vehicles) are Unmanned aircraft that are operated by radio remote control devices and self-contained program control devices, or are operated autonomously, either completely or intermittently, by an onboard computer. Along with the continuous development of science and technology, unmanned aerial vehicle's application is more and more extensive. For example, in aspects such as aerial photography, survey and drawing, disaster relief, agricultural plant protection, electric power inspection, intelligent logistics and smart city, unmanned aerial vehicle can actively realize the field integration to actively change people's production life.

Unmanned aerial vehicle can carry out the flight of returning a journey at the flight in-process, at the flight in-process that returns a journey, needs ensure unmanned aerial vehicle's safety and return a journey. However, in practical applications, for the real-time monitoring process of the remaining capacity of the unmanned aerial vehicle, although the current problems of timeliness and poor monitoring stability can be compensated to some extent, a remaining capacity prediction scheme for a target monitoring task period in a future period of time still lacks at present.

Disclosure of Invention

In order to overcome the defects in the prior art at least, the invention aims to provide a smart lamp pole-based unmanned aerial vehicle residual power real-time monitoring method and system.

In a first aspect, the invention provides an unmanned aerial vehicle remaining capacity real-time monitoring method based on a smart lamp post, which is applied to a monitoring service platform, wherein the monitoring service platform is in communication connection with a plurality of unmanned aerial vehicles to be monitored, and the method comprises the following steps:

acquiring first monitoring log information of an unmanned aerial vehicle to be monitored of a target intelligent lamp pole in a target monitoring task time period and electric quantity of the unmanned aerial vehicle in a previous reference monitoring task time period of the target monitoring task time period;

acquiring second monitoring log information and a target unmanned aerial vehicle behavior tag sequence corresponding to a plurality of reference monitoring task time periods before the target monitoring task time period; the first monitoring log information and/or the second monitoring log information includes: at least one monitoring log information of flight attitude monitoring information, flight altitude monitoring information, flight cooperative monitoring information and flight speed monitoring information;

determining a first target unmanned aerial vehicle monitoring characteristic information sequence based on the correlation degree of the unmanned aerial vehicle monitoring characteristic information sequence formed by the second monitoring log information and the target unmanned aerial vehicle behavior tag sequence, wherein the first target unmanned aerial vehicle monitoring characteristic information sequence comprises second monitoring log information of at least two reference monitoring task time periods;

and predicting the electric quantity of the unmanned aerial vehicle to be monitored in the target monitoring task period according to the first monitoring log information in the target monitoring task period, the electric quantity of the unmanned aerial vehicle in the previous reference monitoring task period in the target monitoring task period and the first target unmanned aerial vehicle monitoring characteristic information sequence.

In a second aspect, the embodiment of the invention further provides a smart-lamp-post-based real-time monitoring system for the remaining power of the unmanned aerial vehicle, which comprises a monitoring service platform and a plurality of unmanned aerial vehicles to be monitored, wherein the plurality of unmanned aerial vehicles to be monitored are in communication connection with the monitoring service platform;

the monitoring service platform is used for:

According to any one of the aspects, in the embodiment provided by the invention, the electric quantity of the unmanned aerial vehicle in the target monitoring task time period is obtained based on at least two pieces of second monitoring log information, and the second monitoring log information can be reference monitoring log information; moreover, the first target unmanned aerial vehicle monitoring characteristic information sequence is obtained based on the degree of correlation with the target unmanned aerial vehicle behavior tag sequence, namely the first target unmanned aerial vehicle monitoring characteristic information sequence is obtained after screening the reference monitoring log information based on the degree of correlation, the basic data size of the reference monitoring log information used for predicting the electric quantity of the unmanned aerial vehicle in the target monitoring task time interval is simplified, in addition, the first target monitoring log information screened out based on the degree of correlation can obtain second monitoring log information strongly correlated with the electric quantity of the unmanned aerial vehicle in the target monitoring task time interval, and the electric quantity prediction precision is further improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that need to be called in the embodiments are briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention, and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.

Fig. 1 is a schematic view of an application scenario of a smart lamp post-based real-time monitoring system for the remaining power of an unmanned aerial vehicle according to an embodiment of the present invention;

fig. 2 is a schematic flow chart of a method for monitoring the remaining power of an unmanned aerial vehicle in real time based on a smart lamp post according to an embodiment of the present invention;

fig. 3 is a schematic view of functional modules of a device for monitoring the remaining power of an unmanned aerial vehicle in real time based on a smart lamp post according to an embodiment of the present invention;

fig. 4 is a block diagram illustrating a structure of a monitoring service platform for implementing the method for monitoring the remaining power of the unmanned aerial vehicle in real time based on the smart lamp post according to the embodiment of the present invention.

Detailed Description

The present invention is described in detail below with reference to the drawings, and the specific operation methods in the method embodiments can also be applied to the apparatus embodiments or the system embodiments.

Fig. 1 is an explanatory schematic view of a system 10 for real-time monitoring of the remaining power of an unmanned aerial vehicle based on a smart light pole according to an embodiment of the present invention. Unmanned aerial vehicle residual capacity real-time monitoring system 10 based on wisdom lamp pole can include control service platform 100 and with control service platform 100 communication connection's unmanned aerial vehicle 200 of waiting to monitor. The smart light pole based real-time monitoring system 10 for the remaining power of the unmanned aerial vehicle shown in fig. 1 is only one possible example, and in other possible embodiments, the smart light pole based real-time monitoring system 10 for the remaining power of the unmanned aerial vehicle may also include only at least some of the components shown in fig. 1 or may also include other components.

For example, the monitoring service platform 100 and the to-be-monitored unmanned aerial vehicle 200 in the smart lamp pole-based unmanned aerial vehicle remaining power real-time monitoring system 10 may cooperatively perform the smart lamp pole-based unmanned aerial vehicle remaining power real-time monitoring method described in the following method embodiments, and the following detailed description of the method embodiments may be referred to in the execution step sections of the monitoring service platform 100 and the to-be-monitored unmanned aerial vehicle 200.

In order to solve the technical problem in the foregoing background, referring to fig. 2, a flow diagram of a method for monitoring the remaining power of the unmanned aerial vehicle in real time based on the smart lamp pole according to an embodiment of the present invention is provided, and the method for monitoring the remaining power of the unmanned aerial vehicle in real time based on the smart lamp pole according to the embodiment of the present invention can be executed by the monitoring service platform 100 shown in fig. 1, and the method for monitoring the remaining power of the unmanned aerial vehicle in real time based on the smart lamp pole is described in detail below.

Step S110, first monitoring log information of the unmanned aerial vehicle to be monitored of the target smart lamp pole in the target monitoring task time period and electric quantity of the unmanned aerial vehicle in a previous reference monitoring task time period in the target monitoring task time period are obtained.

Step S120, second monitoring log information corresponding to a plurality of reference monitoring task time periods before the target monitoring task time period and a target unmanned aerial vehicle behavior label sequence are obtained.

Step S130, determining a first target unmanned aerial vehicle monitoring characteristic information sequence based on the correlation degree of the unmanned aerial vehicle monitoring characteristic information sequence formed by the second monitoring log information and the target unmanned aerial vehicle behavior tag sequence, wherein the first target unmanned aerial vehicle monitoring characteristic information sequence comprises second monitoring log information of at least two reference monitoring task time periods.

Step S140, predicting the electric quantity of the unmanned aerial vehicle to be monitored in the target monitoring task period according to the first monitoring log information in the target monitoring task period, the electric quantity of the unmanned aerial vehicle in the previous reference monitoring task period in the target monitoring task period and the first target unmanned aerial vehicle monitoring characteristic information sequence.

The scheme provided by the invention can be applied to but not limited to the following scenes: the first monitoring log information of the target smart lamp pole of the target monitoring task time interval is obtained, the unmanned aerial vehicle electric quantity of the previous benchmark monitoring task time interval of the target monitoring task time interval, the second monitoring log information and the target unmanned aerial vehicle behavior tag sequence corresponding to the multiple benchmark monitoring task time intervals before the target monitoring task time interval, the target unmanned aerial vehicle behavior tag sequence comprises unmanned aerial vehicle behavior tags corresponding to the multiple benchmark monitoring task time intervals, the unmanned aerial vehicle behavior tags can be represented through tag IDs, namely the corresponding unmanned aerial vehicle behavior tags are 1 when the triggering behaviors exist, and the corresponding unmanned aerial vehicle behavior tags are 0 when the triggering behaviors do not exist. The "monitoring task time interval" may be a time interval with a certain time length, and based on the degree of correlation between the unmanned aerial vehicle monitoring feature information sequence and the target unmanned aerial vehicle behavior tag sequence, a first target unmanned aerial vehicle monitoring feature information sequence is screened from second monitoring log information, where the first target unmanned aerial vehicle monitoring feature information sequence includes second monitoring log information of at least two reference monitoring task time intervals, and the first target unmanned aerial vehicle monitoring feature information sequence may be a set formed by monitoring log information corresponding to a plurality of reference monitoring task time intervals adjacent to the target monitoring task time interval. And then, determining the electric quantity of the unmanned aerial vehicle in the target monitoring task period based on the first monitoring log information in the target monitoring task period, the electric quantity of the unmanned aerial vehicle in the previous reference monitoring task period in the target monitoring task period and the monitoring characteristic information sequence of the first target unmanned aerial vehicle, so as to perform subsequent service processing based on the result.

In order to clarify the technical solutions provided by the present invention, the following explains the solutions provided by the present invention by specific examples:

assuming that a target monitoring task time interval is (s + 1) a monitoring task time interval, first monitoring log information of a corresponding target smart lamp pole is represented as Qs +1, monitoring log information of a previous reference monitoring task time interval (i.e., the s monitoring task time interval) of the target monitoring task time interval is represented as Qs, a plurality of reference monitoring task time intervals before the target monitoring task time interval, the plurality of reference monitoring task time intervals may be a plurality of reference monitoring task time intervals of the previous reference monitoring task time interval of the target monitoring task time interval, that is, the plurality of reference monitoring task time intervals may be P reference monitoring task time intervals before the previous reference monitoring task time interval (the s monitoring task time interval) of the target monitoring task time interval, and second monitoring log information corresponding to the P reference monitoring task time intervals is: qs-1, Qs-2, …, Qs-p. Compared with the target monitoring task time interval (s + 1), the s monitoring task time interval and the P monitoring task time intervals before the s monitoring task time interval are both the reference monitoring task time intervals of the target monitoring task time interval. In this case, the second monitoring log information corresponding to the plurality of reference monitoring task time periods before the target monitoring task time period includes: and the sequence of the corresponding target unmanned aerial vehicle behavior tag is { Ws, Ws-1, Ws-2, …, Ws-p }.

Acquiring a plurality of unmanned aerial vehicle monitoring characteristic information sequences formed by monitoring log information corresponding to s monitoring task time periods before a target monitoring task time period and P reference monitoring task time periods before the s monitoring task time periods, then determining a first target drone monitoring characteristic information sequence based on the correlation of the plurality of drone monitoring characteristic information sequences and the target drone behavior tag sequence, in the present example, the first target drone monitoring characteristic information sequence is a drone monitoring characteristic information sequence composed of monitoring log information corresponding to s-monitoring task period to (s-P) monitoring task period, and determining a first target unmanned aerial vehicle monitoring characteristic information sequence according to the correlation degree between the target unmanned aerial vehicle behavior tag sequences corresponding to the s-to- (s-P) monitoring task time interval, wherein the first target unmanned aerial vehicle monitoring characteristic information sequence is a combination of at least two pieces of screened second monitoring log information.

The first target unmanned aerial vehicle monitoring characteristic information sequence is determined based on the degree of correlation between the unmanned aerial vehicle monitoring characteristic information sequence formed by the second monitoring log information and the target unmanned aerial vehicle behavior tag sequence, namely, the first target unmanned aerial vehicle monitoring characteristic information sequence is screened out from the plurality of unmanned aerial vehicle monitoring characteristic information sequences by utilizing the degree of correlation between the unmanned aerial vehicle monitoring characteristic information sequence and the target unmanned aerial vehicle behavior tag sequence, so that the information simplification of the second monitoring log information is realized, and the size of basic data for predicting the electric quantity of the unmanned aerial vehicle is reduced.

After the first target unmanned aerial vehicle monitoring characteristic information sequence is obtained, the unmanned aerial vehicle electric quantity in the target monitoring task time period is determined according to the first monitoring log information in the target monitoring task time period, the unmanned aerial vehicle electric quantity in the previous reference monitoring task time period in the target monitoring task time period and the first target unmanned aerial vehicle monitoring characteristic information sequence. The electric quantity of the unmanned aerial vehicle in the previous reference monitoring task period of the target monitoring task period can be obtained by adopting the scheme provided by the invention, and the following examples are shown: if the former reference monitoring task time interval of the target monitoring task time interval is the s monitoring task time interval, the electric quantity of the unmanned aerial vehicle in the s monitoring task time interval can be obtained according to (s-1) the electric quantity of the unmanned aerial vehicle in the monitoring task time interval, the first monitoring log information in the s monitoring task time interval and the first target unmanned aerial vehicle monitoring characteristic information sequence corresponding to the s monitoring task time interval. In two adjacent monitoring task periods, the monitoring log information of the former reference monitoring task period is used as basic data for obtaining the electric quantity of the unmanned aerial vehicle in the latter monitoring task period, and iteration is performed continuously along with the lapse of time, so that the monitoring log information of the s monitoring task period and all the monitoring task periods before the s monitoring task period are used as basic data for the prediction result of the s monitoring task period, the basic data for predicting the electric quantity of the unmanned aerial vehicle in the target monitoring task period is greatly expanded, and the accuracy of the electric quantity of the unmanned aerial vehicle is greatly improved.

Therefore, compared with the scheme of monitoring the log information by adopting a single reference, the method has the advantages that the monitoring log information is richer, and the accuracy of the electric quantity of the unmanned aerial vehicle is improved; moreover, the first target unmanned aerial vehicle monitoring characteristic information sequence is obtained based on the degree of correlation with the target unmanned aerial vehicle behavior tag sequence, that is, the first target unmanned aerial vehicle monitoring characteristic information sequence is obtained after screening the second monitoring log information based on the degree of correlation, the second monitoring log information of the unmanned aerial vehicle electric quantity in the target monitoring task period is simplified, the reduction of data processing quantity in a prediction scheme is facilitated, in addition, the unmanned aerial vehicle monitoring characteristic information sequence is screened based on the degree of correlation, the first target unmanned aerial vehicle monitoring characteristic information sequence strongly correlated with the unmanned aerial vehicle electric quantity in the target monitoring task period can be obtained, the accuracy of a prediction result is ensured, and meanwhile, the efficiency of obtaining the prediction result can be improved.

In order to make the target intelligent lamp pole prediction scheme and the technical effect thereof provided by the present invention more clear to those skilled in the art, the following detailed description will be given to specific embodiments by using a plurality of alternative embodiments.

In an independent embodiment, the determining of the first target drone monitoring characteristic information sequence based on the correlation degree between the drone monitoring characteristic information sequence composed of the second monitoring log information and the target drone behavior tag sequence provided in step S130 may be implemented by:

step S131, determining a second target unmanned aerial vehicle monitoring characteristic information sequence with the maximum correlation degree with the target unmanned aerial vehicle behavior tag sequence from the second monitoring log information.

Step S132, determining a first target unmanned aerial vehicle monitoring characteristic information sequence based on the second target unmanned aerial vehicle monitoring characteristic information sequence.

And constructing a plurality of unmanned aerial vehicle monitoring characteristic information sequences according to second monitoring log information corresponding to a plurality of reference monitoring task time periods before the target monitoring task time period, wherein each unmanned aerial vehicle monitoring characteristic information sequence comprises at least one piece of second monitoring log information, and different unmanned aerial vehicle monitoring characteristic information sequences can contain the same second monitoring log information, namely the same service behavior elements exist in different unmanned aerial vehicle monitoring characteristic information sequences. And calculating the correlation degree between each unmanned aerial vehicle monitoring characteristic information sequence and the target unmanned aerial vehicle behavior tag sequence. Optionally, the unmanned aerial vehicle monitoring feature information sequence with the largest correlation degree is determined as the second target unmanned aerial vehicle monitoring feature information sequence.

And then determining a first target unmanned aerial vehicle monitoring characteristic information sequence based on a second target unmanned aerial vehicle monitoring characteristic information sequence, wherein the first target unmanned aerial vehicle monitoring characteristic information sequence and the second target unmanned aerial vehicle monitoring characteristic information sequence both comprise at least two pieces of second monitoring log information. The first target unmanned aerial vehicle monitoring characteristic information sequence is a set formed by second monitoring log information of a plurality of reference monitoring task time periods before the target monitoring task time period, and the number of the specific reference monitoring task time periods is determined by the second target unmanned aerial vehicle monitoring characteristic information sequence. Because the degree of correlation between the second target unmanned aerial vehicle monitoring characteristic information sequence and the target unmanned aerial vehicle behavior tag sequence is maximum, on this basis, the first target unmanned aerial vehicle monitoring characteristic information sequence is determined based on the second target unmanned aerial vehicle monitoring characteristic information sequence, which is favorable for enabling the degree of correlation between the first target unmanned aerial vehicle monitoring characteristic information sequence and the prediction result of the target monitoring task time interval to be higher, if the first target unmanned aerial vehicle monitoring characteristic information sequence can be the unmanned aerial vehicle monitoring characteristic information sequence with the maximum degree of correlation with the prediction result of the target monitoring task time interval, the accuracy of the obtained unmanned aerial vehicle electric quantity of the target monitoring task time interval is favorable for being improved.

It is worth mentioning that the second target drone monitoring characteristic information sequence and the first target drone monitoring characteristic information sequence may contain coincident second monitoring log information, which is exemplified as follows: the monitoring characteristic information sequence of the second target unmanned aerial vehicle is { Qs-1, Qs-2, …, Qs-P-1}, the monitoring characteristic information sequence of the first target unmanned aerial vehicle can be { Qs, Qs-1}, that is, the monitoring log information contained in the monitoring characteristic information sequence of the second target unmanned aerial vehicle and the monitoring log information contained in the monitoring characteristic information sequence of the first target unmanned aerial vehicle can be overlapped or have different monitoring log information, and the set length of the monitoring characteristic information sequence of the first target unmanned aerial vehicle and the set length of the monitoring characteristic information sequence of the second target unmanned aerial vehicle can be different.

According to the scheme provided by the embodiment of the invention, the electric quantity of the unmanned aerial vehicle in the current monitoring task period is predicted based on the first target unmanned aerial vehicle monitoring characteristic information sequence containing at least two pieces of second monitoring log information, the first target unmanned aerial vehicle monitoring characteristic information sequence is obtained based on the second target unmanned aerial vehicle monitoring characteristic information sequence with the maximum correlation degree with the target unmanned aerial vehicle behavior tag sequence, the first target unmanned aerial vehicle monitoring characteristic information sequence can be the unmanned aerial vehicle monitoring characteristic information sequence with the maximum correlation degree with the prediction result in the target monitoring task period, namely the first target unmanned aerial vehicle monitoring characteristic information sequence can be the optimal sequence for accurately predicting the electric quantity of the unmanned aerial vehicle in the target monitoring task period, the accuracy of the electric quantity of the unmanned aerial vehicle is favorably improved, and the efficiency for obtaining the accurate electric quantity of the unmanned aerial vehicle is favorably improved.

In a separate embodiment, determining from the second monitoring log information, a second target drone monitoring feature information sequence having a greatest degree of correlation with the sequence of target drone behavior tags includes:

step S1311, determining a plurality of initial unmanned aerial vehicle monitoring characteristic information sequences according to the plurality of second monitoring log information;

step S1312, determining the correlation degree of each starting unmanned aerial vehicle monitoring characteristic information sequence and the target unmanned aerial vehicle behavior tag sequence;

step S1313, determining the initial unmanned aerial vehicle monitoring feature information sequence with the largest correlation degree as the first target unmanned aerial vehicle monitoring feature information sequence.

The unmanned aerial vehicle monitoring characteristic information sequence is formed by the second monitoring log information, a plurality of initial unmanned aerial vehicle monitoring characteristic information sequences are constructed based on the unmanned aerial vehicle monitoring characteristic information sequence, and the initial unmanned aerial vehicle monitoring characteristic information sequences can be obtained by combining the second monitoring log information in the unmanned aerial vehicle monitoring characteristic information sequence. The second monitoring log information here may be monitoring log information of P reference monitoring task time periods corresponding to a previous reference monitoring task time period of the target monitoring task time period, such as: the target monitoring task time interval is (s + 1) monitoring task time interval, the former reference monitoring task time interval of the target monitoring task time interval is s monitoring task time interval, the P reference monitoring task time intervals corresponding to the former reference monitoring task time interval of the target monitoring task time interval can be the former P monitoring task time intervals adjacent to the s monitoring task time interval, namely, (s-1), (s-2), …, (s-P), and the unmanned aerial vehicle monitoring characteristic information sequence formed by the second monitoring log information corresponding to the s to (s-P) monitoring task time intervals is { Qs, Qs-1, Qs-2, …, Qs-P }. The initial unmanned aerial vehicle monitoring characteristic information sequence constructed based on the unmanned aerial vehicle monitoring characteristic information sequence can be { Qs, Qs-1, Qs-2}, { Qs-1, Qs-2, Qs-3}, …, { Qs-p-2, Qs-p-1, Qs-p } and the like, and the time orders of the initial unmanned aerial vehicle monitoring characteristic information sequences can be different. And calculating the correlation degree of each initial unmanned aerial vehicle monitoring characteristic information sequence and the target unmanned aerial vehicle behavior tag sequence, and determining the initial unmanned aerial vehicle monitoring characteristic information sequence with the maximum correlation degree as a second target unmanned aerial vehicle monitoring characteristic information sequence.

In an independent embodiment, the determining, according to the second monitoring log information, the starting unmanned aerial vehicle monitoring feature information sequences provided in step S1311 may be obtained as follows:

step S13111, forming a first unmanned aerial vehicle monitoring characteristic information sequence according to the plurality of second monitoring log information;

step S13112, performing information conversion on the first unmanned monitoring characteristic information sequence to obtain a second service behavior distribution space, wherein the second service behavior distribution space is a symmetrical distribution space;

step S13113, determine each horizontal distribution or each vertical distribution in the second traffic behavior distribution space as a starting drone monitoring feature information sequence.

Determining a plurality of second monitoring log information based on the initial monitoring time interval, forming the plurality of second monitoring log information into a first unmanned monitoring characteristic information sequence, constructing a second service behavior distribution space based on the first unmanned monitoring characteristic information sequence by utilizing information conversion, wherein the second service behavior distribution space is a symmetrical distribution space, and the ith transverse distribution (row) in the second service behavior distribution space is the same as the ith unmanned monitoring characteristic information sequence longitudinally distributed (column), therefore, the second service behavior distribution space may be divided into starting unmanned aerial vehicle monitoring feature information sequences in the same number as the number of horizontal distributions/the number of vertical distributions of the distribution space by rows, that is, each horizontal distribution or each vertical distribution in the second service behavior distribution space is used as one starting unmanned aerial vehicle monitoring feature information sequence, and the time order of each starting unmanned aerial vehicle monitoring feature information sequence is the same. And then, similarity calculation is carried out on the basis of the initial unmanned aerial vehicle monitoring characteristic information sequence and a target unmanned aerial vehicle behavior tag sequence with the same vector size, and the unmanned aerial vehicle monitoring characteristic information sequence with the maximum similarity is determined from the initial unmanned aerial vehicle monitoring characteristic information sequence and serves as a second target unmanned aerial vehicle monitoring characteristic information sequence.

Optionally, the information conversion is performed on the first unmanned monitoring characteristic information sequence to obtain a second service behavior distribution space, which may be implemented as follows: taking the first unmanned monitoring characteristic information sequence as a first transverse distribution or a first longitudinal distribution of a second business behavior distribution space, taking the first irman-machine monitoring characteristic information sequence as the first longitudinal distribution of the second service behavior distribution space as an example, the first bit of monitoring log information distributed from the first transverse distribution to the last transverse distribution in the second service behavior distribution space is determined as the monitoring log information arranged in sequence in the first irman-machine monitoring characteristic information sequence, each transverse distribution in the distribution space includes the monitoring log information with the first bit of monitoring log information as the starting data, the time order of each transverse distribution is the same as that of the first irman-machine monitoring characteristic information sequence, in this embodiment, for the monitoring log information in each horizontal distribution in the second service behavior distribution space, the later monitoring log information is the reference data of the earlier monitoring log information.

In a separate embodiment, the determining the first target drone monitoring feature information sequence based on the second target drone monitoring feature information sequence provided in step S132 may be implemented by:

step S1321, determining the monitoring time interval of the second target unmanned aerial vehicle monitoring characteristic information sequence as an optimal monitoring time interval;

step S1322, determining a target reference monitoring task time interval based on the target monitoring task time interval and the optimal monitoring time interval;

and step S1323, taking a set formed by monitoring log information of the target reference monitoring task time period as a first target unmanned aerial vehicle monitoring characteristic information sequence.

The sequence number corresponding to the first target unmanned aerial vehicle monitoring characteristic information sequence is used as the optimal monitoring time interval of the unmanned aerial vehicle monitoring characteristic information sequence, if the first target unmanned aerial vehicle monitoring characteristic information sequence is the p-th, the starting unmanned aerial vehicle monitoring characteristic information sequence, then, the p is used as the optimal monitoring time interval of the target unmanned aerial vehicle monitoring characteristic information sequence, namely, the target unmanned aerial vehicle monitoring characteristic information sequence comprises p and monitoring log information.

For example, a plurality of starting unmanned aerial vehicle monitoring feature information sequences may be sequenced from near to far in time with a target monitoring task time period, and if there are p starting unmanned aerial vehicle monitoring feature information sequences, then the sequence number of the starting unmanned aerial vehicle monitoring feature information sequence closest to the target monitoring task time period is set to 1, the sequence number of the starting unmanned aerial vehicle monitoring feature information sequence farthest in time from the target monitoring task time period is set to p, and the degree of correlation between the starting unmanned aerial vehicle monitoring feature information sequence and the target unmanned aerial vehicle behavior tag sequence is calculated, and the degree of correlation between each starting unmanned aerial vehicle monitoring feature information sequence and the target unmanned aerial vehicle behavior tag sequence is represented in a sequence form, such as: the sequence of the degree of correlation is, wherein, is the degree of correlation of the starting unmanned aerial vehicle monitoring characteristic information sequence with sequence number 1 and the target unmanned aerial vehicle behavior tag sequence, is the degree of correlation that the starting unmanned aerial vehicle monitoring characteristic information sequence with sequence number p corresponds, and so on. The initial unmanned aerial vehicle monitoring feature information sequence corresponding to the maximum correlation degree of the absolute value of the correlation degree can be selected to be determined as the second target unmanned aerial vehicle monitoring feature information sequence, and the expression for determining the maximum correlation degree can be as follows: . Taking the sequence number of the initial unmanned aerial vehicle monitoring characteristic information sequence corresponding to the maximum correlation degree as an optimal monitoring time interval, if: the correlation degree sequence is {0.1, 0.2, 0.4, 0.6, 0.3}, then the maximum correlation degree is 0.6, its corresponding sequence number is 4, then the optimal monitoring period interval is 4.

And determining a target reference monitoring task time interval of the electric quantity of the unmanned aerial vehicle for predicting the target monitoring task time interval according to the optimal monitoring time interval, and taking a set formed by monitoring log information corresponding to the target reference monitoring task time interval as a first target unmanned aerial vehicle monitoring characteristic information sequence. If the target monitoring task time interval is (s + 1) monitoring task time interval and the optimal monitoring time interval is 4, the target reference monitoring task time interval can be s monitoring task time interval, (s-1) monitoring task time interval, (s-2) monitoring task time interval, or (s-3) monitoring task time interval, and the first target unmanned aerial vehicle monitoring characteristic information sequence is { Qs, Qs-1, Qs-2, Qs-3 }.

When the first target unmanned aerial vehicle monitoring characteristic information sequence is determined, the optimal monitoring time interval is determined firstly, then the first target unmanned aerial vehicle monitoring characteristic information sequence corresponding to the target monitoring task time interval is determined based on the optimal monitoring time interval, the first target unmanned aerial vehicle monitoring characteristic information sequence corresponding to each target monitoring task time interval can be determined according to each target monitoring task time interval, and the unmanned aerial vehicle electric quantity prediction accuracy of the target monitoring task time interval is improved.

An implementation process for determining a target unmanned aerial vehicle monitoring characteristic information sequence provided by an optional embodiment of the invention is as follows: firstly, determining a second target unmanned aerial vehicle monitoring characteristic information sequence with the maximum correlation degree with the target unmanned aerial vehicle behavior tag sequence from second monitoring log information, wherein the step can be performed through the schemes provided in the steps S1311 to S1313, and the step of determining a plurality of starting unmanned aerial vehicle monitoring characteristic information sequences according to a set formed by the second monitoring log information can be performed through the schemes provided in the steps S13111 to S13113; then, the first target drone monitoring characteristic information sequence is determined based on the second target drone monitoring characteristic information sequence, and this step may be performed by the scheme provided in steps S1321 to S1323. According to the scheme provided by the embodiment, the first target unmanned aerial vehicle monitoring characteristic information sequence can be rapidly determined from a large amount of second monitoring log information, so that the unmanned aerial vehicle electric quantity can be more accurately predicted subsequently based on the first target unmanned aerial vehicle monitoring characteristic information sequence.

To further clarify the embodiments provided by the present invention, reference may be made to specific examples. Assume that the input Ws of the s monitoring task period includes the input (Ws-1) of the last monitoring task period (s-1) thereof, the second monitoring log information { Qs-1, Qs-2, …, Qs-p } of the plurality of reference monitoring task periods { s-1, s-2, …, s-p }, and the first monitoring log information Qs of the s monitoring task period. The output of the s-monitor task period is related to the input Ws of the s-monitor task period. In the embodiment of the invention, the input Ws of the s monitoring task time interval is used as the electric quantity of the unmanned aerial vehicle in the s monitoring task time interval, and the Qs output based on the electric quantity of the unmanned aerial vehicle is used as the data output after further processing.

The second service behavior distribution space Q is a symmetric distribution space composed of monitoring log information sequences { Qs, Qs-1, …, Qs-P, …, Qs-2P } corresponding to each reference monitoring task time period (i.e., the monitoring log information in the jth column of the ith row is equal to the monitoring log information in the ith column of the jth row).

In the related technology, the input data of the s monitoring task time interval only adopts the monitoring log information comprising the current target monitoring task time interval and the previous period thereof, namely when the s monitoring task time interval is the target monitoring task time interval, only the monitoring log information of (s-1) monitoring task time interval is used as the input data of the target monitoring task time interval. Moreover, the second monitoring log information can be obtained by the scheme of screening the target monitoring log information, so that the prediction result is further improved, namely the accuracy of the electric quantity of the unmanned aerial vehicle in the target monitoring task time interval is improved, and the electric quantity prediction precision can be improved.

In addition, the unmanned aerial vehicle remaining power real-time monitoring method based on the intelligent lamp pole provided by the optional embodiment of the invention can be carried out in a model mode, so that the efficiency and the accuracy of obtaining a prediction result are further improved.

For example, the step S140 of predicting the electric quantity of the unmanned aerial vehicle to be monitored in the target monitoring task period according to the first monitoring log information in the target monitoring task period, the electric quantity of the unmanned aerial vehicle in the previous reference monitoring task period in the target monitoring task period, and the first target unmanned aerial vehicle monitoring feature information sequence may be performed in the following manner:

the method comprises the steps that first monitoring log information of a target monitoring task time interval, unmanned aerial vehicle electric quantity of a previous reference monitoring task time interval of the target monitoring task time interval and a first target unmanned aerial vehicle monitoring characteristic information sequence are used as input of an unmanned aerial vehicle electric quantity prediction model, and unmanned aerial vehicle electric quantity of the target monitoring task time interval is obtained;

in a separate embodiment, the training process of the unmanned aerial vehicle electric quantity prediction model is as follows:

b1, acquiring collected monitoring log information, wherein the collected monitoring log information comprises a calibrated unmanned aerial vehicle monitoring characteristic information sequence corresponding to a plurality of reference monitoring task time intervals and corresponding unmanned aerial vehicle behavior labels; the calibration unmanned aerial vehicle monitoring characteristic information sequence corresponding to each reference monitoring task time interval comprises a plurality of calibration monitoring log information.

And B2, performing iterative training on the initial machine learning model by using the collected monitoring log information until the prediction loss corresponding to the initial machine learning model reaches a preset requirement, taking the initial machine learning model corresponding to the condition that the prediction loss meets a preset training termination condition as an unmanned aerial vehicle electric quantity prediction model, and representing the difference between the trigger behavior prediction information output by the network and the unmanned aerial vehicle behavior label by using the value of the prediction loss.

In the embodiment of the invention, each piece of collected monitoring log information of the unmanned aerial vehicle electric quantity prediction model comprises a calibrated unmanned aerial vehicle monitoring characteristic information sequence corresponding to a plurality of reference monitoring task time intervals and a corresponding unmanned aerial vehicle behavior label, each piece of collected monitoring log information comprises a calibrated unmanned aerial vehicle monitoring characteristic information sequence corresponding to one reference monitoring task time interval and a corresponding unmanned aerial vehicle behavior label, the calibrated unmanned aerial vehicle monitoring characteristic information sequence corresponding to one reference monitoring task time interval comprises a plurality of pieces of calibrated monitoring log information, and the calibrated monitoring log information is the reference monitoring log information.

The method comprises the steps of sequentially inputting a monitoring characteristic information sequence of a calibration unmanned aerial vehicle corresponding to a reference monitoring task time interval into a target neural network, namely sequentially inputting a plurality of calibration monitoring log information corresponding to one reference monitoring task time interval into an initial machine learning model, outputting a prediction result, calculating the difference between the prediction result and an unmanned aerial vehicle behavior label corresponding to the reference monitoring task time interval, obtaining a prediction loss value of the model corresponding to the reference monitoring task time interval, if the prediction loss value does not meet the condition of model convergence, adjusting network weight by using the prediction loss value until the initial machine learning model converges, and enabling the initial machine learning model during convergence to be an unmanned aerial vehicle electric quantity prediction model.

Of course, for the acquired collected monitoring log information, a part of test data used as the unmanned aerial vehicle electric quantity prediction model may be reserved, for example, the collected monitoring log information may be randomly divided into a calibration training data set (ratio a) and a calibration test data set (ratio 1-a) according to a preset ratio. And when the value of the prediction loss corresponding to the calibration test data set meets the preset requirement of the unmanned aerial vehicle electric quantity prediction model, finishing the network training. The prediction loss can be represented as indexes such as accuracy, when the prediction loss is lower than the preset prediction loss, the initial machine learning model can be judged to meet the preset requirement, and the initial machine learning model during convergence is the unmanned aerial vehicle electric quantity prediction model. The unmanned aerial vehicle electric quantity prediction model is trained based on calibrated unmanned aerial vehicle monitoring characteristic information sequences corresponding to a plurality of reference monitoring task periods and corresponding unmanned aerial vehicle behavior labels, wherein each calibrated unmanned aerial vehicle monitoring characteristic information sequence comprises a plurality of pieces of second monitoring log information. Network training is carried out based on the second monitoring log information, so that the influence factors of the second monitoring log information are considered in the trained electric quantity prediction model of the unmanned aerial vehicle, compared with a mode of carrying out network training by adopting single second monitoring log information, the prediction accuracy of the electric quantity prediction model of the unmanned aerial vehicle obtained by utilizing the second monitoring log information is improved.

According to the target intelligent lamp pole prediction scheme provided by the invention, the electric quantity of the unmanned aerial vehicle in the target monitoring task time period is predicted by utilizing the pre-trained electric quantity prediction model of the unmanned aerial vehicle, so that the electric quantity of the unmanned aerial vehicle in the target monitoring task time period can be rapidly determined. For example, it may be used to perform the steps of the aforementioned steps S110 to S140.

Optionally, the collected monitoring log information of the unmanned aerial vehicle electric quantity prediction model includes calibration unmanned aerial vehicle monitoring feature information sequences corresponding to a plurality of reference monitoring task time intervals and corresponding unmanned aerial vehicle behavior tags, and the calibration unmanned aerial vehicle monitoring feature information sequence corresponding to each reference monitoring task time interval may include a set formed by the calibration monitoring log information corresponding to the reference monitoring task time interval and a first training target unmanned aerial vehicle monitoring feature information sequence corresponding to the reference monitoring task time interval. If one reference monitoring task time interval in the collected monitoring log information is an s monitoring task time interval, the monitoring feature information sequence of the calibration unmanned aerial vehicle corresponding to the s monitoring task time interval comprises calibration monitoring log information corresponding to the s monitoring task time interval and a first training target unmanned aerial vehicle monitoring feature information sequence corresponding to the s monitoring task time interval, and the first training target unmanned aerial vehicle monitoring feature information sequence corresponding to the s monitoring task time interval can be obtained based on the optimal monitoring time interval obtained by the method. Examples are as follows: if the optimal monitoring time interval is p, the monitoring feature information sequence of the calibrated unmanned aerial vehicle corresponding to the s monitoring task time interval comprises first monitoring log information Qs corresponding to the s monitoring task time interval and a first target unmanned aerial vehicle monitoring feature information sequence { Qs-1, …, Qs-p ', which corresponds to the s monitoring task time interval, namely the monitoring feature information sequence of the calibrated unmanned aerial vehicle corresponding to the s monitoring task time interval is { Qs, Qs-1, …, Qs-p'.

In the scheme provided in this embodiment, the calibration unmanned aerial vehicle monitoring feature information sequence is a set formed by calibration monitoring log information corresponding to a reference monitoring task time period and a first training target unmanned aerial vehicle monitoring feature information sequence corresponding to the reference monitoring task time period. That is to say, each set of model input data in the collected monitoring log information is composed of a plurality of pieces of second monitoring log information having the greatest correlation with the behavior tag of the drone. The scheme that this embodiment provided has carried out the screening on a plurality of second unmanned aerial vehicle control characteristic information sequence's basis, utilizes and has the input data of the higher calibration unmanned aerial vehicle control characteristic information sequence of a plurality of second monitoring log information component of degree of correlation that triggers the action result as the model, carries out the network training, can further promote the accuracy of unmanned aerial vehicle electric quantity prediction model, is favorable to promoting the efficiency of obtaining unmanned aerial vehicle electric quantity prediction model simultaneously.

Optionally, in the training process of the unmanned aerial vehicle electric quantity prediction model, the method further includes:

c1, classifying the calibration monitoring log information in the collected monitoring log information to obtain a first calibration monitoring log information and a second calibration monitoring log information;

c2, performing feature processing on the first calibration monitoring log information and the second calibration monitoring log information respectively;

and C3, taking the first calibration monitoring log information and the second calibration monitoring log information after feature processing and corresponding unmanned aerial vehicle behavior labels as input data of the initial machine learning model, so as to train the neural network model.

And dividing the calibration monitoring log information in the collected monitoring log information according to the data types, wherein the calibration monitoring log information can be second monitoring log information, and corresponding characteristic processing is performed on the calibration monitoring log information of different data types, so that the efficiency of monitoring log information is improved, and the characteristic loss can be reduced to the maximum extent.

And then, network training is carried out by utilizing the first calibrated monitoring log information after the characteristic processing and the densely collected monitoring log information, so that the data processing amount of the model is favorably reduced, and the training efficiency of the model is favorably improved.

Optionally, the feature processing may be performed on the first calibration monitoring log information and the second calibration monitoring log information respectively in the following manner, including:

c21, performing feature extraction on the first calibration monitoring log information to obtain a sparse feature vector corresponding to the first calibration monitoring log information; and/or the presence of a gas in the gas,

c22, at least one of decorrelating, normalizing or feature discretizing is performed on the second calibration monitoring log information.

And performing feature extraction on the first calibration monitoring log information, so that the sparse training features are expressed into high-order features through feature extraction, and feature information of original sparse features can be reserved under lower feature dimensionality.

At least one of decorrelation processing, normalization processing and feature discretization processing is performed on the second calibration monitoring log information so as to solve the problems of correlation, standardization, discretization and the like among the second calibration monitoring log information, and therefore the method is beneficial to reducing the subsequent calculation complexity of the monitoring log information and improving the processing efficiency of the monitoring log information.

Optionally, a pre-trained feature extraction model may be used to perform feature extraction on the first calibration monitoring log information, and the feature extraction model is pre-trained by collecting the monitoring log information, so that rapid feature extraction may be performed on the first monitoring log information in the subsequent unmanned aerial vehicle electric quantity prediction model.

Specifically, the first monitoring log information is processed into a sparse feature vector by using a feature extraction model, the sparse feature is expressed into a high-order feature, and feature information is reserved while feature dimensionality is reduced.

In an independent embodiment, the first monitoring log information of the target monitoring task time interval, the electric quantity of the unmanned aerial vehicle in the previous reference monitoring task time interval of the target monitoring task time interval, and the first target unmanned aerial vehicle monitoring characteristic information sequence are used as the input of an unmanned aerial vehicle electric quantity prediction model to obtain the electric quantity of the unmanned aerial vehicle in the target monitoring task time interval, and the method can be carried out in the following manner, and includes the following steps:

d1, performing data division on second monitoring log information in the first target unmanned aerial vehicle monitoring characteristic information sequence and first monitoring log information of a target monitoring task time period to obtain first monitoring log information and second monitoring log information;

d2, performing feature extraction on sparse feature data to obtain sparse feature vectors corresponding to the first monitoring log information;

and D3, inputting the sparse feature vector, the dense feature and the electric quantity of the unmanned aerial vehicle in the previous reference monitoring task period of the target monitoring task period into the electric quantity prediction model of the unmanned aerial vehicle to obtain the electric quantity of the unmanned aerial vehicle in the target monitoring task period.

If the optimal monitoring time interval is p', for a target monitoring task time interval (s + 1), the first monitoring log information of the target monitoring task time interval is Qs +1, and the corresponding first target unmanned aerial vehicle monitoring characteristic information sequence is { Qs, …, Qs-p, +1 }. And performing data category division on second monitoring log information in the first target unmanned aerial vehicle monitoring characteristic information sequence and first monitoring log information of a target monitoring task time period to obtain the first monitoring log information and the second monitoring log information. And then respectively carrying out different feature processing on the data features, and carrying out feature extraction on the first monitoring log information to obtain a sparse feature vector corresponding to the first monitoring log information, wherein the sparse feature vector is a high-order feature of the sparse feature, so that feature dimensionality can be reduced, and feature information can be reserved. At least one of decorrelation processing, normalization processing and feature discretization processing is performed on the second monitoring log information to process the problems of correlation, standardization, discretization and the like among the second monitoring log information. By carrying out targeted processing on two different types of monitoring log information, the characteristic information can be retained to the greatest extent, and meanwhile, the data processing amount is reduced, such as: and feature extraction processing and the like are not required to be carried out on the second monitoring log information, so that the data processing efficiency is improved.

In order to further clarify the method for monitoring the remaining power of the unmanned aerial vehicle based on the smart lamp post in real time, a target smart lamp post prediction scheme provided by the invention is described below by combining an example.

Assuming that the target monitoring task time interval is (s + 1) monitoring task time interval, the following is specific:

obtaining second monitoring log information corresponding to the initial monitoring period interval P according to the input initial monitoring period interval P, obtaining an unmanned aerial vehicle monitoring characteristic information sequence { Qs, Qs-1, …, Qs-2P } and a corresponding target unmanned aerial vehicle behavior tag sequence (i.e. tag set) { Ws, Ws-1, …, Ws-P } based on the second monitoring log information, performing information conversion on the unmanned aerial vehicle monitoring characteristic information sequence by using an information conversion device to obtain a second service behavior distribution space Q, wherein P initial unmanned aerial vehicle monitoring characteristic information sequences Qi (i =1, 2, …, P) are arranged in the second service behavior distribution space, calculating the correlation degree of each initial unmanned aerial vehicle monitoring characteristic information sequence Qi and the tag set W, and corresponding to the initial unmanned aerial vehicle monitoring characteristic information sequence with the maximum correlation degree to the monitoring period interval P, and the time interval is optimally monitored.

Determining a first target unmanned aerial vehicle monitoring characteristic information sequence { Qs, Qs-1, …, Qs-p, } of a target monitoring task time interval based on an optimal monitoring time interval, performing network training based on the first target unmanned aerial vehicle monitoring characteristic information sequence, before the network training, firstly performing data classification on the first target unmanned aerial vehicle monitoring characteristic information sequence, and classifying second monitoring log information in the first target unmanned aerial vehicle monitoring characteristic information sequence into second calibration monitoring log information (corresponding to dense characteristics) and first calibration monitoring log information (corresponding to sparse characteristics); the method includes the steps that different data processing is conducted on different data types, multi-layer feature extraction can be conducted on first calibration monitoring log information, for example, feature extraction can be conducted through a deep neural network Model (DNN Model), the deep neural network Model can be a multi-layer network architecture Model, such as 3 layers and 5 layers, sparse feature vectors corresponding to the first monitoring log information are obtained through the deep neural network Model, then dense features, the sparse feature vectors and labels corresponding to the features of the two types are used as collected monitoring log information of a cyclic neural network Model, whether a network training process is completed or not is judged through loss parameters, if the loss parameters reach preset training termination conditions, the network training is completed, and an unmanned aerial vehicle electric quantity prediction Model is obtained.

The method comprises the steps of dividing monitoring log information of an unmanned aerial vehicle monitoring characteristic information sequence { Qs +1, Qs, …, Qs-p +1} which is composed of first monitoring log information corresponding to a target monitoring task time interval (s + 1) and the first target unmanned aerial vehicle monitoring characteristic information sequence, dividing the monitoring log information into first monitoring log information and second monitoring log information, carrying out characteristic extraction on the first monitoring log information, carrying out characteristic extraction by using a pre-trained deep neural network model, and obtaining sparse characteristic vectors corresponding to the first monitoring log information in the unmanned aerial vehicle monitoring characteristic information sequence corresponding to the target monitoring task time interval. And finally, inputting the sparse feature vector and dense features in the unmanned aerial vehicle monitoring feature information sequence corresponding to the target monitoring task time interval into the trained unmanned aerial vehicle electric quantity prediction model to obtain the unmanned aerial vehicle electric quantity of the target monitoring task time interval.

In the target smart light pole prediction scheme provided in the foregoing embodiment, optionally, the first monitoring log information and/or the second monitoring log information may include: at least one monitoring log information of flight attitude monitoring characteristic information, flight altitude monitoring information, flight cooperative monitoring information and flight speed monitoring information.

Correspondingly, an optional embodiment of the invention further provides a smart lamp pole-based real-time monitoring method for the residual electric quantity of the unmanned aerial vehicle, which comprises the following steps:

e1, acquiring collected monitoring log information and an initial machine learning model; the collected monitoring log information comprises a calibrated unmanned aerial vehicle monitoring characteristic information sequence corresponding to a plurality of reference monitoring task time intervals and a corresponding unmanned aerial vehicle behavior label, and the calibrated unmanned aerial vehicle monitoring characteristic information sequence corresponding to each reference monitoring task time interval comprises a plurality of calibrated monitoring log information;

e2, inputting the monitoring characteristic information sequence of the calibrated unmanned aerial vehicle corresponding to each reference monitoring task time interval into an initial machine learning model to obtain the triggering behavior prediction information corresponding to each reference characteristic monitoring task time interval, and determining the value of the prediction loss based on the triggering behavior prediction information and the unmanned aerial vehicle behavior label corresponding to the reference monitoring task time interval;

e3, training the initial machine learning model based on the value of the prediction loss until the prediction loss of the initial machine learning model meets the preset training termination condition, and taking the initial machine learning model when the prediction loss meets the preset training termination condition as the unmanned aerial vehicle electric quantity prediction model to predict the unmanned aerial vehicle electric quantity.

In the embodiment of the invention, each piece of collected monitoring log information of the unmanned aerial vehicle electric quantity prediction model comprises a calibrated unmanned aerial vehicle monitoring characteristic information sequence corresponding to a plurality of reference monitoring task time intervals and a corresponding unmanned aerial vehicle behavior label, each piece of collected monitoring log information comprises a calibrated unmanned aerial vehicle monitoring characteristic information sequence corresponding to one reference monitoring task time interval and a corresponding unmanned aerial vehicle behavior label, the calibrated unmanned aerial vehicle monitoring characteristic information sequence corresponding to one reference monitoring task time interval comprises a plurality of calibrated monitoring log information, and the calibrated monitoring log information can be second monitoring log information.

The method comprises the steps of sequentially inputting a monitoring characteristic information sequence of a calibrated unmanned aerial vehicle corresponding to a reference monitoring task time interval into a target neural network, namely sequentially inputting a plurality of pieces of second monitoring log information corresponding to one reference monitoring task time interval into an initial machine learning model, outputting a prediction result, calculating the difference between the prediction result and an unmanned aerial vehicle behavior label corresponding to the reference monitoring task time interval, obtaining a prediction loss value of the model corresponding to the reference monitoring task time interval, if the prediction loss value does not meet a model convergence condition, adjusting network weight by using the prediction loss value until the initial machine learning model converges, and enabling the initial machine learning model during convergence to be an unmanned aerial vehicle electric quantity prediction model.

The unmanned aerial vehicle electric quantity prediction model is trained based on calibrated unmanned aerial vehicle monitoring characteristic information sequences corresponding to a plurality of reference monitoring task periods and corresponding unmanned aerial vehicle behavior labels, wherein each calibrated unmanned aerial vehicle monitoring characteristic information sequence comprises a plurality of pieces of second monitoring log information. Network training is carried out based on the second monitoring log information, so that the influence factors of the second monitoring log information are considered in the trained electric quantity prediction model of the unmanned aerial vehicle, compared with a mode of carrying out network training by adopting single second monitoring log information, the prediction accuracy of the electric quantity prediction model of the unmanned aerial vehicle obtained by utilizing the second monitoring log information is improved.

Optionally, the calibration unmanned aerial vehicle monitoring feature information sequence corresponding to each reference monitoring task time interval includes calibration monitoring log information corresponding to the reference monitoring task time interval and a first training target unmanned aerial vehicle monitoring feature information sequence corresponding to the reference monitoring task time interval.

In this embodiment, the collected monitoring log information of the model includes calibration unmanned aerial vehicle monitoring feature information sequences corresponding to a plurality of reference monitoring task periods and corresponding unmanned aerial vehicle behavior tags, and the calibration unmanned aerial vehicle monitoring feature information sequence corresponding to each reference monitoring task period may include a set formed by the calibration monitoring log information corresponding to the reference monitoring task period and the first training target unmanned aerial vehicle monitoring feature information sequence corresponding to the reference monitoring task period. If one reference monitoring task time interval in the collected monitoring log information is the s monitoring task time interval, the monitoring feature information sequence of the calibration unmanned aerial vehicle corresponding to the s monitoring task time interval comprises the calibration monitoring log information corresponding to the s monitoring task time interval and the monitoring feature information sequence of the first training target unmanned aerial vehicle corresponding to the s monitoring task time interval, the monitoring feature information sequence of the first training target unmanned aerial vehicle corresponding to the s monitoring task time interval can be the monitoring feature information sequence of the first target unmanned aerial vehicle obtained based on the method (the monitoring feature information sequence of the first target unmanned aerial vehicle corresponding to the s monitoring task time interval is determined based on the optimal monitoring time interval), if the monitoring feature information sequence of the first target unmanned aerial vehicle is used for network training, and taking the first target unmanned aerial vehicle monitoring characteristic information sequence as a first training target unmanned aerial vehicle monitoring characteristic information sequence.

For example: assuming that the optimal monitoring time interval is p, the monitoring feature information sequence of the calibrated unmanned aerial vehicle corresponding to the s monitoring task time interval comprises calibration monitoring log information Qs corresponding to the s monitoring task time interval and a first training target unmanned aerial vehicle monitoring feature information sequence { Qs-1, …, Qs-p } corresponding to the s monitoring task time interval, namely the monitoring feature information sequence of the calibrated unmanned aerial vehicle corresponding to the s monitoring task time interval is { Qs, Qs-1, …, Qs-p }.

In the scheme provided by this embodiment, the unmanned aerial vehicle monitoring feature information sequence is calibrated to be a set formed by monitoring log information corresponding to a reference monitoring task time period and a first training target unmanned aerial vehicle monitoring feature information sequence corresponding to the reference monitoring task time period. That is to say, each set of model input data in the collected monitoring log information is composed of a plurality of pieces of second monitoring log information having the greatest correlation with the behavior tag of the drone. The scheme that this embodiment provided has carried out the screening on a plurality of benchmark control log information's basis, utilizes and has the input data of the higher calibration unmanned aerial vehicle control characteristic information sequence of a plurality of second control log information constitutions of degree of relevance that triggers the action result as the model, carries out the network training, can further promote the accuracy of unmanned aerial vehicle electric quantity prediction model, is favorable to promoting the efficiency that obtains unmanned aerial vehicle electric quantity prediction model simultaneously.

On the basis of the above description, the method provided by the embodiment of the present application may further include the following steps.

Step S150, when the electric quantity of the unmanned aerial vehicle to be monitored in the target monitoring task period is larger than the preset electric quantity, adding the unmanned aerial vehicle to be monitored into a current scheduling unmanned aerial vehicle queue;

and step S160, distributing the tasks of all the unmanned aerial vehicles in the current scheduling unmanned aerial vehicle queue according to the flight tasks of the newly distributed unmanned aerial vehicles.

In a possible design concept, the task allocation of each drone in the currently scheduled drone queue according to the currently newly allocated drone flight task in step S160 may be specifically implemented through the following steps.

Step S161, acquiring an unmanned aerial vehicle node map associated with a flight prediction model corresponding to a flight space environment obtained by adapting to a current flight scene environment in a current scheduling unmanned aerial vehicle queue, acquiring first unmanned aerial vehicle plan information of an unmanned aerial vehicle launching configuration information sequence in an unmanned aerial vehicle flight task according to the unmanned aerial vehicle node map, and acquiring second unmanned aerial vehicle plan information of unmanned aerial vehicle transfer configuration information in the unmanned aerial vehicle flight task;

step S162, carrying out scene environment association on first unmanned aerial vehicle plan information of the unmanned aerial vehicle starting configuration information sequence and second unmanned aerial vehicle plan information of the unmanned aerial vehicle transfer configuration information to obtain scene environment association characteristic information;

step S163, identifying matching probabilities of scene environment associated characteristic information and a plurality of grid objects in a first flight zone grid according to flight distribution nodes in the first flight zone grid corresponding to the unmanned aerial vehicle flight mission, and binding the matching probabilities obtained by the first flight zone grid with scheduling mission objects corresponding to the plurality of grid objects in the first flight zone grid to obtain a first scheduling mission object list;

step S164, identifying matching probabilities of the scene environment associated characteristic information and a plurality of grid objects in a second flight partition grid according to flight distribution nodes in the second flight partition grid, and binding the matching probabilities obtained by the second flight partition grid with scheduling task objects corresponding to the plurality of grid objects in the second flight partition grid to obtain a second scheduling task object list, wherein the first flight partition grid and the second flight partition grid are used for performing flight scheduling on the scene environment associated characteristic information from different flight scheduling categories;

and S165, performing scene environment association on the first scheduling task object list and the second scheduling task object list to obtain a scheduling task object corresponding to the flight task of the unmanned aerial vehicle, and performing corresponding unmanned aerial vehicle task configuration on the flight task of the unmanned aerial vehicle according to the scheduling task object.

Based on the above steps, the present embodiment obtains, according to the node map of the drone, first drone plan information of the drone launching configuration information sequence and second drone plan information of the drone transfer configuration information in the flight mission of the drone, further in the process of performing flight scheduling may be combined to take into account the link between drone plan information between question bank configuration inputs and question bank configuration outputs, therefore, the scene environment associated characteristic information is obtained after the scene environment association is carried out on the unmanned aerial vehicle and the flight scheduling, the input characteristic when the flight scheduling is carried out on the flight task of the unmanned aerial vehicle can be enriched, the accuracy of the flight scheduling is convenient to improve, in addition, the classification of the scheduling task object is carried out on the scene environment associated characteristic information by combining the flight partition grids of two different flight scheduling categories, the flight scheduling can be carried out from the composite flight scheduling dimension, and the accuracy of the flight scheduling is further improved.

In a possible design, according to the unmanned aerial vehicle node map, first unmanned aerial vehicle plan information of an unmanned aerial vehicle launching configuration information sequence in an unmanned aerial vehicle flight task is acquired, and second unmanned aerial vehicle plan information of unmanned aerial vehicle transfer configuration information in the unmanned aerial vehicle flight task is acquired, including: the method comprises the steps that transfer configuration information of the unmanned aerial vehicle is extracted from a flight task of the unmanned aerial vehicle, a plurality of pieces of unmanned aerial vehicle starting configuration information are generated according to return flight demand information and a plurality of task demands in the flight task of the unmanned aerial vehicle, and the plurality of pieces of unmanned aerial vehicle starting configuration information are combined into an unmanned aerial vehicle starting configuration information sequence; the method comprises the steps of obtaining first characteristic information and second characteristic information corresponding to an unmanned aerial vehicle node map, wherein the first characteristic information comprises unmanned aerial vehicle plans matched with a plurality of characteristics, and the second characteristic information comprises unmanned aerial vehicle plans matched with characteristics of transfer configuration information of a plurality of unmanned aerial vehicles. And then, extracting first unmanned plane plan information of the unmanned plane starting configuration information sequence through the first characteristic information, and extracting second unmanned plane plan information of the unmanned plane transfer configuration information through the second characteristic information.

The method for carrying out scene environment association on the first scheduling task object list and the second scheduling task object list to obtain the scheduling task object corresponding to the flight task of the unmanned aerial vehicle comprises the following steps: in the first scheduling task object list and the second scheduling task object list, carrying out weighted average on the matching probability associated with the same scheduling task object, and binding the matching probability after weighted average with the scheduling task object to obtain a target scheduling task object list; extracting a scheduling task object associated with the maximum matching probability from a target scheduling task object list, and taking the extracted scheduling task object as a scheduling task object corresponding to the flight task of the unmanned aerial vehicle;

the second flight zone grid is obtained by:

(1) acquiring first calibration scene environment associated characteristic information and second calibration scene environment associated characteristic information;

(2) extracting second unmanned aerial vehicle plan information of the first calibration scene environment associated characteristic information, identifying the matching probability of the second unmanned aerial vehicle plan information of the first calibration scene environment associated characteristic information and a plurality of grid objects in a second flight partition grid according to flight distribution nodes in the second flight partition grid, and binding the matching probability obtained by the second unmanned aerial vehicle plan information of the first calibration scene environment associated characteristic information and scheduling task objects corresponding to the plurality of grid objects in the second flight partition grid to obtain a third scheduling task object list;

(3) extracting second unmanned aerial vehicle plan information of second calibration scene environment associated characteristic information, identifying the matching probability of the second unmanned aerial vehicle plan information of the second calibration scene environment associated characteristic information and a plurality of grid objects according to flight distribution nodes in a second flight partition grid, and binding the matching probability obtained by the second unmanned aerial vehicle plan information of the second calibration scene environment associated characteristic information and scheduling task objects corresponding to the plurality of grid objects in the second flight partition grid to obtain a fourth scheduling task object list;

(4) determining a loss function value according to second unmanned aerial vehicle plan information and a third scheduling task object list of the first calibration scene environment associated characteristic information, and second unmanned aerial vehicle plan information and a fourth scheduling task object list of the second calibration scene environment associated characteristic information, and adjusting network weight in a second flight partition grid according to the loss function value;

the second flight partition grid is used for outputting a second scheduling task object list matched with second unmanned aerial vehicle plan information of unmanned aerial vehicle transfer configuration information in the unmanned aerial vehicle flight task;

the second scheduling task object list is used for obtaining a scheduling task object corresponding to the flight task of the unmanned aerial vehicle after scene environment association is carried out on the second scheduling task object list and the first scheduling task object list;

the first scheduling task object list is a label set which is output by the first flight partition grid and matched with first unmanned aerial vehicle planning information of an unmanned aerial vehicle starting configuration information sequence in the unmanned aerial vehicle flight task;

the loss function value includes an identification loss parameter value and a validation loss parameter value;

determining a loss function value according to second unmanned aerial vehicle plan information and a third scheduling task object list of the first calibration scene environment associated characteristic information, and second unmanned aerial vehicle plan information and a fourth scheduling task object list of the second calibration scene environment associated characteristic information, wherein the determining comprises the following steps: generating an identification loss parameter value of the first calibration scene environment associated characteristic information according to the third scheduling task object list and the labeling grid object corresponding to the first calibration scene environment associated characteristic information; generating an identification loss parameter value of the second calibration scene environment associated characteristic information according to the fourth scheduling task object list and the labeling grid object corresponding to the second calibration scene environment associated characteristic information; generating a verification loss parameter value according to second unmanned aerial vehicle plan information of the first calibration scene environment associated characteristic information, a labeling grid object corresponding to the first calibration scene environment associated characteristic information, second unmanned aerial vehicle plan information of the second calibration scene environment associated characteristic information and a labeling grid object corresponding to the second calibration scene environment associated characteristic information; and generating a loss function value according to the identification loss parameter value of the first calibration scene environment associated characteristic information, the identification loss parameter value of the second calibration scene environment associated characteristic information and the verification loss parameter value.

Fig. 3 is a schematic functional module diagram of the unmanned aerial vehicle remaining power real-time monitoring device 300 based on the smart lamp post according to the embodiment of the present invention, and the functions of the functional modules of the unmanned aerial vehicle remaining power real-time monitoring device 300 based on the smart lamp post are described in detail below.

The first obtaining module 310 is configured to obtain first monitoring log information of the target smart lamp pole at the target monitoring task time interval of the unmanned aerial vehicle to be monitored and an electric quantity of the unmanned aerial vehicle at a previous reference monitoring task time interval of the target monitoring task time interval.

The second obtaining module 320 is configured to obtain second monitoring log information and a target unmanned aerial vehicle behavior tag sequence corresponding to a plurality of reference monitoring task time periods before the target monitoring task time period. The first monitoring log information and/or the second monitoring log information includes: at least one monitoring log information of flight attitude monitoring information, flight altitude monitoring information, flight cooperative monitoring information and flight speed monitoring information.

The first determining module 330 is configured to determine a first target drone monitoring feature information sequence based on a degree of correlation between a drone monitoring feature information sequence composed of second monitoring log information and a target drone behavior tag sequence, where the first target drone monitoring feature information sequence includes second monitoring log information of at least two reference monitoring task periods.

The second determining module 340 is configured to predict the electric quantity of the unmanned aerial vehicle to be monitored in the target monitoring task period according to the first monitoring log information in the target monitoring task period, the electric quantity of the unmanned aerial vehicle in the previous reference monitoring task period in the target monitoring task period, and the monitoring characteristic information sequence of the first target unmanned aerial vehicle.

Fig. 4 shows a hardware structure of the monitoring service platform 100 for implementing the method for monitoring the remaining power of the smart light pole-based drone in real time according to the embodiment of the present invention, as shown in fig. 4, the monitoring service platform 100 may include a processor 110, a machine-readable storage medium 120, a bus 130, and a transceiver 140.

In a specific implementation process, the at least one processor 110 executes the monitoring service platform stored in the machine-readable storage medium 120 to execute the instructions, so that the processor 110 may execute the method for monitoring the remaining power of the smart light pole-based drone in real time as in the above method embodiment, the processor 110, the machine-readable storage medium 120, and the transceiver 140 are connected through the bus 130, and the processor 110 may be configured to control the transceiver 140 to perform transceiving actions, so as to perform data transceiving with the drone 200 to be monitored.

For a specific implementation process of the processor 110, reference may be made to the above-mentioned various method embodiments executed by the monitoring service platform 100, which implement principles and technical effects are similar, and details of this embodiment are not described herein again.

In addition, the embodiment of the invention also provides a readable storage medium, wherein the readable storage medium is preset with a monitoring service platform execution instruction, and when a processor executes the monitoring service platform execution instruction, the unmanned aerial vehicle residual power real-time monitoring method based on the intelligent lamp pole is realized.

Finally, it should be understood that the examples in this specification are only intended to illustrate the principles of the examples in this specification. Other variations are also possible within the scope of this description. Accordingly, by way of example, and not limitation, alternative configurations of the embodiments of the specification can be seen as matching the teachings of the specification. Accordingly, the embodiments of the present description are not limited to only those embodiments explicitly described and depicted herein.

Claims

1. The utility model provides an unmanned aerial vehicle residual capacity real-time monitoring method based on wisdom lamp pole which characterized in that is applied to the monitoring service platform, the monitoring service platform is with a plurality of unmanned aerial vehicle communication connection of waiting to monitor, the method includes:

predicting the electric quantity of the unmanned aerial vehicle to be monitored in the target monitoring task period according to the first monitoring log information of the target monitoring task period, the electric quantity of the unmanned aerial vehicle in the previous reference monitoring task period of the target monitoring task period and the first target unmanned aerial vehicle monitoring characteristic information sequence;

the target unmanned aerial vehicle behavior tag sequence comprises unmanned aerial vehicle behavior tags corresponding to a plurality of reference monitoring task time intervals, the unmanned aerial vehicle behavior tags are represented by tag IDs, the corresponding unmanned aerial vehicle behavior tag is 1 when a triggering behavior exists, and the corresponding unmanned aerial vehicle behavior tag is 0 when the triggering behavior does not exist;

the determining of the correlation degree between the unmanned aerial vehicle monitoring characteristic information sequence formed based on the second monitoring log information and the target unmanned aerial vehicle behavior tag sequence includes:

determining a second target unmanned aerial vehicle monitoring characteristic information sequence with the maximum correlation degree with the target unmanned aerial vehicle behavior tag sequence based on the second monitoring log information;

determining the first target unmanned aerial vehicle monitoring characteristic information sequence based on the second target unmanned aerial vehicle monitoring characteristic information sequence;

wherein the determining, based on the second monitoring log information, a second target drone monitoring feature information sequence having a greatest degree of correlation with the target drone behavior tag sequence includes:

determining a plurality of initial unmanned aerial vehicle monitoring characteristic information sequences according to the second monitoring log information, determining the correlation degree of each initial unmanned aerial vehicle monitoring characteristic information sequence and the target unmanned aerial vehicle behavior tag sequence, and determining the initial unmanned aerial vehicle monitoring characteristic information sequence with the maximum correlation degree as the second target unmanned aerial vehicle monitoring characteristic information sequence;

the determining the first target unmanned aerial vehicle monitoring characteristic information sequence based on the second target unmanned aerial vehicle monitoring characteristic information sequence includes:

determining a monitoring time interval of the second target unmanned aerial vehicle monitoring characteristic information sequence as an optimal monitoring time interval, determining a target reference monitoring task time interval based on a target monitoring task time interval and the optimal monitoring time interval, and taking a set formed by monitoring log information of the target reference monitoring task time interval as the first target unmanned aerial vehicle monitoring characteristic information sequence;

wherein, according to the first monitoring log information of the target monitoring task period, the unmanned aerial vehicle electric quantity of the previous reference monitoring task period of the target monitoring task period, and the first target unmanned aerial vehicle monitoring characteristic information sequence, predicting the unmanned aerial vehicle electric quantity of the unmanned aerial vehicle to be monitored in the target monitoring task period, includes:

using the first monitoring log information of the target monitoring task time interval, the electric quantity of the unmanned aerial vehicle of a previous reference monitoring task time interval of the target monitoring task time interval and the first target unmanned aerial vehicle monitoring characteristic information sequence as the input of an unmanned aerial vehicle electric quantity prediction model to obtain the electric quantity of the unmanned aerial vehicle of the target monitoring task time interval;

the training process of the unmanned aerial vehicle electric quantity prediction model is as follows:

acquiring collected monitoring log information, wherein the collected monitoring log information comprises a calibrated unmanned aerial vehicle monitoring characteristic information sequence corresponding to a plurality of reference monitoring task time intervals and a corresponding unmanned aerial vehicle behavior tag, and the calibrated unmanned aerial vehicle monitoring characteristic information sequence corresponding to each reference monitoring task time interval comprises a plurality of calibrated monitoring log information;

and performing iterative training on an initial machine learning model by using the collected monitoring log information until the prediction loss corresponding to the initial machine learning model reaches a preset requirement, taking the initial machine learning model corresponding to the initial machine learning model when the prediction loss meets a preset training termination condition as an unmanned aerial vehicle electric quantity prediction model, wherein the value of the prediction loss represents the difference between the trigger behavior prediction information output by the network and the unmanned aerial vehicle behavior label.

2. The method of claim 1, wherein determining a plurality of initial sequences of monitoring characteristic information of the unmanned aerial vehicle according to the second monitoring log information comprises:

forming a first unmanned monitoring characteristic information sequence according to the second monitoring log information;

performing information conversion on the first unmanned monitoring characteristic information sequence to obtain a second service behavior distribution space, wherein the second service behavior distribution space is a symmetrical distribution space;

and determining each horizontal distribution or each vertical distribution in the second service behavior distribution space as an initial unmanned aerial vehicle monitoring characteristic information sequence.

3. The method as claimed in claim 1, wherein before performing iterative training on an initial machine learning model using the collected monitoring log information, the method further comprises:

classifying the calibration monitoring log information in the collected monitoring log information to obtain first calibration monitoring log information and second calibration monitoring log information, wherein the first calibration monitoring log information and the second calibration monitoring log information are respectively sparse calibration monitoring log information and dense calibration monitoring log information;

respectively performing characteristic processing on the first calibration monitoring log information and the second calibration monitoring log information;

and taking the first calibration monitoring log information and the second calibration monitoring log information after the characteristic processing and corresponding unmanned aerial vehicle behavior labels as input data of the initial machine learning model.

4. The method for real-time monitoring of the remaining power of the unmanned aerial vehicle based on the smart lamp post as claimed in claim 1, wherein the step of obtaining the power of the unmanned aerial vehicle in the target monitoring task period by using the first monitoring log information in the target monitoring task period, the power of the unmanned aerial vehicle in the previous reference monitoring task period in the target monitoring task period, and the first target unmanned aerial vehicle monitoring characteristic information sequence as the input of the unmanned aerial vehicle power prediction model comprises:

performing data division on second monitoring log information in the first target unmanned aerial vehicle monitoring characteristic information sequence and first monitoring log information of a target monitoring task time period to obtain first monitoring log information and second monitoring log information;

performing feature extraction on the first monitoring log information to obtain a sparse feature vector corresponding to the first monitoring log information;

and inputting the sparse feature vector, the second monitoring log information and the electric quantity of the unmanned aerial vehicle in the previous reference monitoring task period of the target monitoring task period into the electric quantity prediction model of the unmanned aerial vehicle to obtain the electric quantity of the unmanned aerial vehicle in the target monitoring task period.

5. The real-time monitoring method for the residual power of the unmanned aerial vehicle based on the intelligent lamp post as claimed in any one of claims 1 to 4, wherein the method further comprises:

when the electric quantity of the unmanned aerial vehicle to be monitored in the target monitoring task period is larger than a preset electric quantity, adding the unmanned aerial vehicle to be monitored into a current scheduling unmanned aerial vehicle queue;

and performing task allocation on each unmanned aerial vehicle in the current scheduling unmanned aerial vehicle queue according to the current newly allocated unmanned aerial vehicle flight task.

6. The smart lamp post-based real-time monitoring method for the remaining power of the unmanned aerial vehicle, according to claim 5, wherein the step of performing task allocation on each unmanned aerial vehicle in the currently scheduled unmanned aerial vehicle queue according to the currently newly allocated unmanned aerial vehicle flight task includes:

acquiring an unmanned aerial vehicle node map associated with a flight prediction model corresponding to a flight space environment obtained by adapting to a current flight scene environment in the current scheduling unmanned aerial vehicle queue, acquiring first unmanned aerial vehicle plan information of an unmanned aerial vehicle launching configuration information sequence in an unmanned aerial vehicle flight task according to the unmanned aerial vehicle node map, and acquiring second unmanned aerial vehicle plan information of unmanned aerial vehicle transfer configuration information in the unmanned aerial vehicle flight task;

carrying out scene environment association on first unmanned aerial vehicle plan information of the unmanned aerial vehicle launching configuration information sequence and second unmanned aerial vehicle plan information of the unmanned aerial vehicle transfer configuration information to obtain scene environment association characteristic information;

according to a flight distribution node in a first flight partition grid corresponding to the unmanned aerial vehicle flight task, identifying matching probabilities of the scene environment associated characteristic information and a plurality of grid objects in the first flight partition grid, and binding the matching probabilities obtained by the first flight partition grid with scheduling task objects corresponding to the plurality of grid objects in the first flight partition grid to obtain a first scheduling task object list;

according to a flight distribution node in a second flight partition grid, identifying matching probabilities of the scene environment associated characteristic information and a plurality of grid objects in the second flight partition grid, and binding the matching probabilities obtained by the second flight partition grid with scheduling task objects corresponding to the grid objects in the second flight partition grid to obtain a second scheduling task object list, wherein the first flight partition grid and the second flight partition grid are used for performing flight scheduling on the scene environment associated characteristic information from different flight scheduling categories;

and performing scene environment association on the first scheduling task object list and the second scheduling task object list to obtain a scheduling task object corresponding to the unmanned aerial vehicle flight task, and performing corresponding unmanned aerial vehicle task configuration on the unmanned aerial vehicle flight task according to the scheduling task object.

7. The method as claimed in claim 6, wherein the obtaining of the first drone plan information of the drone enabled configuration information sequence in the drone flight task and the obtaining of the second drone plan information of the drone transfer configuration information in the drone flight task according to the drone node map includes:

extracting transfer configuration information of the unmanned aerial vehicle from the flight task of the unmanned aerial vehicle, generating a plurality of pieces of unmanned aerial vehicle starting configuration information according to the return flight requirement information and a plurality of task requirements in the flight task of the unmanned aerial vehicle, and combining the plurality of pieces of unmanned aerial vehicle starting configuration information into an unmanned aerial vehicle starting configuration information sequence;

acquiring first characteristic information and second characteristic information corresponding to the unmanned aerial vehicle node map, wherein the first characteristic information comprises unmanned aerial vehicle plans matched with a plurality of characteristics, and the second characteristic information comprises unmanned aerial vehicle plans matched with characteristics of transfer configuration information of a plurality of unmanned aerial vehicles;

extracting first unmanned plane plan information of the unmanned plane starting configuration information sequence through the first characteristic information, and extracting second unmanned plane plan information of the unmanned plane transfer configuration information through the second characteristic information;

the performing scene environment association on the first scheduling task object list and the second scheduling task object list to obtain the scheduling task object corresponding to the flight task of the unmanned aerial vehicle includes:

in the first scheduling task object list and the second scheduling task object list, carrying out weighted average on the matching probability associated with the same scheduling task object, and binding the matching probability after weighted average with the scheduling task object to obtain a target scheduling task object list;

extracting a scheduling task object associated with the maximum matching probability from the target scheduling task object list, and taking the extracted scheduling task object as a scheduling task object corresponding to the unmanned aerial vehicle flight task;

the second flight zone grid is obtained by:

acquiring first calibration scene environment associated characteristic information and second calibration scene environment associated characteristic information;

extracting second unmanned aerial vehicle plan information of the first calibration scene environment associated characteristic information, identifying matching probabilities of the second unmanned aerial vehicle plan information of the first calibration scene environment associated characteristic information and a plurality of grid objects in a second flight partition grid according to flight distribution nodes in the second flight partition grid, and binding the matching probabilities obtained by the second unmanned aerial vehicle plan information of the first calibration scene environment associated characteristic information and scheduling task objects corresponding to the plurality of grid objects in the second flight partition grid to obtain a third scheduling task object list;

extracting second unmanned aerial vehicle plan information of the second calibration scene environment associated characteristic information, identifying matching probabilities of the second unmanned aerial vehicle plan information of the second calibration scene environment associated characteristic information and the grid objects according to flight distribution nodes in the second flight partition grid, and binding the matching probabilities obtained by the second unmanned aerial vehicle plan information of the second calibration scene environment associated characteristic information and scheduling task objects corresponding to the grid objects in the second flight partition grid to obtain a fourth scheduling task object list;

determining a loss function value according to second unmanned aerial vehicle plan information and the third scheduling task object list of the first calibration scene environment associated characteristic information, and second unmanned aerial vehicle plan information and the fourth scheduling task object list of the second calibration scene environment associated characteristic information, and adjusting network weight in the second flight partition grid according to the loss function value;

the second flight partition grid is used for outputting a second scheduling task object list matched with second unmanned aerial vehicle plan information of unmanned aerial vehicle transfer configuration information in the unmanned aerial vehicle flight tasks;

the second scheduling task object list is used for obtaining a scheduling task object corresponding to the unmanned aerial vehicle flight task after scene environment association is carried out on the second scheduling task object list and the first scheduling task object list;

the first scheduling task object list is a label set which is output by a first flight partition grid and matched with first unmanned aerial vehicle planning information of an unmanned aerial vehicle starting configuration information sequence in the unmanned aerial vehicle flight task;

the loss function value comprises an identification loss parameter value and a validation loss parameter value;

determining a loss function value according to the second unmanned aerial vehicle plan information and the third scheduling task object list of the first calibration scene environment associated characteristic information, and the second unmanned aerial vehicle plan information and the fourth scheduling task object list of the second calibration scene environment associated characteristic information, including:

generating an identification loss parameter value of the first calibration scene environment associated characteristic information according to the third scheduling task object list and the labeled grid object corresponding to the first calibration scene environment associated characteristic information;

generating an identification loss parameter value of the second calibration scene environment associated characteristic information according to the fourth scheduling task object list and the labeled grid object corresponding to the second calibration scene environment associated characteristic information;

generating the verification loss parameter value according to second unmanned aerial vehicle plan information of the first calibration scene environment associated characteristic information, a labeling grid object corresponding to the first calibration scene environment associated characteristic information, second unmanned aerial vehicle plan information of the second calibration scene environment associated characteristic information, and a labeling grid object corresponding to the second calibration scene environment associated characteristic information;

and generating the loss function value according to the identification loss parameter value of the first calibration scene environment associated characteristic information, the identification loss parameter value of the second calibration scene environment associated characteristic information and the verification loss parameter value.

8. The unmanned aerial vehicle residual electric quantity real-time monitoring system based on the intelligent lamp pole is characterized by comprising a monitoring service platform and a plurality of unmanned aerial vehicles to be monitored, wherein the unmanned aerial vehicles to be monitored are in communication connection with the monitoring service platform;

the monitoring service platform is used for: