Unmanned Device Autonomous Task Optimization Method and System Based on Cross-Modal Data Alignment

Abstract

The present invention relates to the technical field of task optimization, and particularly to an autonomous task optimization method and system for unmanned devices based on cross-modal data alignment. The method includes: collecting logistics demand data of different UAV delivery areas and resource data of different UAVs, and simultaneously collecting environmental data; constructing an order quantity prediction model for different UAV delivery areas and outputting the predicted order quantity of each delivery area in the next 2 hours; calculating the comprehensive priority index of different UAV delivery areas, and calculating the comprehensive resource gap index of different UAV delivery areas to generate a priority sequence; constructing a UAV scheduling objective function and setting up constraint conditions, where the constraint conditions include power constraint, load constraint, and airspace constraint, to generate an optimal scheduling plan.

Description

Background Art

[0002] In recent years, with the rapid development of e-commerce and the surging demand of consumers for instant delivery, drone logistics has become an important technical direction for solving delivery problems due to its strong flexibility, fast delivery speed, and good adaptability to complex terrains. However, with the complication and scale of application scenarios, drone logistics faces severe challenges in aspects such as dynamic task allocation, multi-modal data fusion, and resource optimization scheduling.

[0003] Disadvantages of the prior art: Traditional systems often only preset fixed delivery routes and allocate tasks according to time windows, unable to respond to sudden orders or environmental changes, and ignoring the power attenuation of drones (such as insufficient remaining power for return after long flights); Most existing methods do not integrate multi-dimensional environmental data such as weather, traffic, and airspace, with incomplete decision-making information and large prediction errors; Existing systems often rely on priority allocation dominated by manual experience, with strong subjectivity, difficult to quantify the weight relationship between urgency and distance, and no distinction in priority between urgent orders and ordinary tasks, which may lead to delays in high-value orders. Summary of the Invention

[0004] The main object of the present invention is to provide an autonomous task optimization method for unmanned devices based on cross-modal data alignment, and further provide an autonomous task optimization system for unmanned devices based on cross-modal data alignment that can run and implement the above method, effectively solving the above problems mentioned in the background art.

[0005] The technical solution of the present invention is as follows:

[0006] In the first aspect, an autonomous task optimization method for unmanned devices based on cross-modal data alignment is proposed, and the method includes the following steps:

[0007] S1. Collect logistics demand data of different drone delivery areas and resource data of different drones, and at the same time collect environmental data;

[0008] S2. Build an order quantity prediction model for different drone delivery areas and output the predicted order quantity of each delivery area in the next 2 hours;

[0009] S3. Calculate the comprehensive priority index of different drone delivery areas, and calculate the comprehensive resource gap index of different drone delivery areas to generate a priority sequence;

[0010] S4. Build a drone scheduling objective function and set constraint conditions, where the constraint conditions include power constraint, load constraint, and airspace constraint, to generate an optimal scheduling plan.

[0011] A further improvement of the present invention is that the S1 includes the following specific steps:

[0012] S11. Collect the logistics demand data of different UAV delivery areas, where the logistics demand data includes the order quantity , the longitude and latitude of the delivery address , the total weight of the goods and the urgency label , where i represents the i-th delivery area. When the urgency is normal, , when the urgency is a rigid demand, 1;

[0013] S12. Collect the resource data of different UAVs, where the resource data includes the available status label of the UAV , the remaining battery power of the UAV , the maximum load of the UAV and the longitude and latitude of the current position of the UAV , where j represents the j-th UAV. When the j-th UAV is available, , when the j-th UAV is unavailable, ;

[0014] S13. Collect the environmental data, where the environmental data includes the weather label weather, the airspace restriction label of the delivery area and the date type label day; where when the weather is sunny, weather = 0, when the weather is rainy, weather = 1, when the weather is windy, weather = 2; when area i is open, , when area i is restricted, ; when the date type is a working day, day = 0, when the date type is a holiday, day = 1.

[0015] A further improvement of the present invention is that the S2 includes the following specific steps:

[0016] S21. Construct an order quantity prediction model for different UAV delivery areas. The order quantity prediction model takes the feature vector formed by the historical 24-hour order quantity, weather label, and date type label of a single delivery area as input, takes the order quantity prediction value corresponding to the feature vector of a single delivery area as output, takes the actual order quantity value corresponding to the feature vector of a single delivery area as the prediction target, and takes minimizing the mean square error between the actual order quantity value and the order quantity prediction value as the training target, and stops training until the mean square error between the actual order quantity value and the order quantity prediction value reaches convergence; output the order quantity of different UAV delivery areas in the next 2 hours predicted by the order quantity prediction model; the order quantity prediction model is an LSTM neural network model;

[0017] S22. The training target formula of the order quantity prediction model is:

[0018] ;

[0019] Among them, represents the actual value of the order quantity in the i-th delivery area, represents the predicted value of the order quantity in the i-th delivery area, and n represents the total number of delivery areas.

[0020] A further improvement of the present invention lies in that calculating the comprehensive priority index of different UAV delivery areas in S3 includes the following specific steps:

[0021] S31. Calculate the task urgency weight of different UAV delivery areas. The calculation formula of the task urgency weight is:

[0022] ;

[0023] Among them, represents the task urgency weight of the i-th delivery area, represents the urgency label of the i-th delivery area, is the emergency order coefficient, and its value is 0.7;

[0024] S32. Calculate the distance weight of different UAV delivery areas. The calculation formula of the distance weight is:

[0025] ;

[0026] Among them, represents the longitude and latitude of the delivery address in the i-th delivery area and the longitude and latitude of the distribution center the spherical distance between them, represents the distance weight of the i-th delivery area;

[0027] S33. Calculate the comprehensive priority index of different UAV delivery areas. The calculation formula of the comprehensive priority index is:

[0028] ;

[0029] Among them, represents the comprehensive priority index of the i-th delivery area, is the proportion coefficient of the task urgency weight, and its value is 0.6.

[0030] A further improvement of the present invention lies in that calculating the comprehensive resource gap index of different UAV delivery areas in S3 includes the following specific steps:

[0031] S34. Calculate the number of UAVs required for different UAV delivery areas. The calculation formula is:

[0032] ;

[0033] Among them, represents the number of drones required for the i-th delivery area, represents rounding up, represents the maximum load of the j-th drone;

[0034] S35. Calculate the power gap index of different drones. The calculation formula of the power gap index is:

[0035] ;

[0036] Among them, represents the power gap index of the j-th drone, represents the minimum return power of the drone, represents the remaining battery power of the j-th drone;

[0037] S36. Calculate the comprehensive resource gap index of different drone delivery areas. The calculation formula of the comprehensive resource gap index is:

[0038] ;

[0039] Among them, represents the comprehensive resource gap index of the i-th delivery area, represents the resource gap weight, with a value of 0.8, represents the available status label of the j-th drone.

[0040] A further improvement of the present invention is that the S3 further includes: sorting the comprehensive resource gap indexes of different drone delivery areas in descending order to generate a priority sequence.

[0041] A further improvement of the present invention is that the drone scheduling objective function in the S4 is:

[0042] ;

[0043] Among them, is the fuel cost per unit distance of the drone, is the path length from the drone j to the delivery area i, is the penalty cost per unit time of order delay, is the estimated time from the drone j to the delivery area i, is the latest delivery time of the order in the delivery area i.

[0044] A further improvement of the present invention is that the constraint conditions in the S4 include power constraint, load constraint, and airspace constraint; the calculation formula of the power constraint is:

[0045] ;

[0046] wherein, is the flight speed of the j-th unmanned aerial vehicle, is the energy consumption rate per unit distance of the unmanned aerial vehicle; the calculation formula for the load constraint is:

[0047] ;

[0048] wherein, represents the set of all delivery areas assigned to the unmanned aerial vehicle j; the calculation formula for the airspace constraint is:

[0049] .

[0050] Second, an autonomous task optimization system for unmanned devices based on cross-modal data alignment is proposed. The system includes: a data collection module, an order volume prediction module, a priority generation module, and a scheduling module;

[0051] The data collection module is used to collect logistics demand data of different unmanned aerial vehicle delivery areas and resource data of different unmanned aerial vehicles, and at the same time collect environmental data;

[0052] The order volume prediction module is used to construct an order quantity prediction model for different unmanned aerial vehicle delivery areas and output the predicted order volume of each delivery area in the next 2 hours;

[0053] The priority generation module is used to calculate the comprehensive priority index of different unmanned aerial vehicle delivery areas, calculate the comprehensive resource gap index of different unmanned aerial vehicle delivery areas, and generate a priority sequence;

[0054] The scheduling module is used to construct an unmanned aerial vehicle scheduling objective function and set constraint conditions, and the constraint conditions include power constraint, load constraint, and airspace constraint, and generate an optimal scheduling plan.

[0055] The technical effects of the present invention are as follows:

[0056] An autonomous task optimization method for unmanned devices based on cross-modal data alignment is constructed. This method overcomes the problem of data fragmentation in traditional systems and ensures that the scheduling plan comprehensively considers order attributes, unmanned aerial vehicle capabilities, and environmental restrictions; uses a long short-term memory network (LSTM) to model the spatio-temporal dependence of historical orders, and the input includes multi-modal features such as weather and date type to accurately predict the order distribution in the next 2 hours and avoid resource waste or shortage caused by prediction deviation; by combining the urgency weight and the distance weight, a dynamic priority sequence is generated to ensure that high-urgency and short-distance tasks are executed first, improving the quality of delivery services. At the same time, this method dynamically adjusts the path under multiple constraints such as power, load, and airspace to minimize the operating cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Other features, objects, and advantages of the present invention will become more apparent by reading the detailed description of non - restrictive embodiments with reference to the following drawings:

[0058] Figure 1 It is a schematic flowchart of an autonomous task optimization method for unmanned devices based on cross - modal data alignment in Embodiment 1 of the present invention;

[0059] Figure 2 It is a schematic structural diagram of an autonomous task optimization system for unmanned devices based on cross - modal data alignment in Embodiment 2 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] Embodiment 1

[0061] In this embodiment, an autonomous task optimization method for unmanned devices based on cross - modal data alignment is constructed. This method overcomes the problem of data fragmentation in traditional systems, ensuring that the scheduling scheme comprehensively considers order attributes, drone capabilities, and environmental constraints; uses a long short - term memory network (LSTM) to model the spatio - temporal dependence of historical orders, with inputs including multi - modal features such as weather and date type, to accurately predict the order distribution in the next 2 hours, avoiding resource waste or shortage caused by prediction deviation; generates a dynamic priority sequence by combining urgency weights and distance weights to ensure that high - urgency and short - distance tasks are executed first, improving the quality of delivery services. At the same time, this method dynamically adjusts the path under multiple constraints such as battery power, load capacity, and airspace to minimize operating costs.

[0062] The autonomous task optimization method for unmanned devices based on cross - modal data alignment, as Figure 1 shown, includes the following specific steps:

[0063] S1. Collect logistics demand data for different drone delivery areas, resource data for different drones, and environmental data at the same time;

[0064] S2. Construct an order quantity prediction model for different drone delivery areas and output the predicted order quantities for each delivery area in the next 2 hours;

[0065] S3. Calculate the comprehensive priority index for different drone delivery areas and calculate the comprehensive resource gap index for different drone delivery areas to generate a priority sequence;

[0066] S4. Construct a drone scheduling objective function and set up constraint conditions, where the constraint conditions include battery power constraint, load capacity constraint, and airspace constraint, to generate an optimal scheduling scheme.

[0067] In this embodiment, S1 includes the following specific steps:

[0068] S11. Collect the logistics demand data of different UAV delivery areas. The logistics demand data includes the order quantity , the longitude and latitude of the delivery address , the total weight of the goods and the urgency level label , where i represents the i-th delivery area. When the urgency level is normal, , when the urgency level is a rigid demand, 1;

[0069] S12. Collect the resource data of different UAVs. The resource data includes the available status label of the UAV , the remaining battery power of the UAV , the maximum load capacity of the UAV and the longitude and latitude of the current position of the UAV , where j represents the j-th UAV. When the j-th UAV is available, , when the j-th UAV is unavailable, ;

[0070] S13. Collect the environmental data. The environmental data includes the weather label weather, the airspace restriction label of the delivery area and the date type label day; where when the weather is sunny, weather = 0, when the weather is rainy, weather = 1, when the weather is windy, weather = 2; when area i is open, , when area i is restricted, ; when the date type is a working day, day = 0, when the date type is a holiday, day = 1.

[0071] In this embodiment, the S2 includes the following specific steps:

[0072] S21. Construct an order quantity prediction model for different UAV delivery areas. The order quantity prediction model takes the feature vector formed by the historical 24-hour order quantity, weather label and date type label of a single delivery area as input, takes the order quantity prediction value corresponding to the feature vector of a single delivery area as output, takes the actual order quantity value corresponding to the feature vector of a single delivery area as the prediction target, and takes minimizing the mean square error between the actual order quantity value and the order quantity prediction value as the training target, and stops training until the mean square error between the actual order quantity value and the order quantity prediction value reaches convergence; output the order quantity of different UAV delivery areas in the next 2 hours predicted by the order quantity prediction model; the order quantity prediction model is an LSTM neural network model;

[0073] S22. The training target formula of the order quantity prediction model is:

[0074] ;

[0075] Among them, represents the actual value of the order quantity in the i-th delivery area, represents the predicted value of the order quantity in the i-th delivery area, and n represents the total number of delivery areas.

[0076] In this embodiment, calculating the comprehensive priority index of different UAV delivery areas in S3 includes the following specific steps:

[0077] S31. Calculate the task urgency weight of different UAV delivery areas. The calculation formula for the task urgency weight is:

[0078] ;

[0079] Among them, represents the task urgency weight of the i-th delivery area, represents the urgency label of the i-th delivery area, is the emergency order coefficient, with a value of 0.7;

[0080] S32. Calculate the distance weight of different UAV delivery areas. The calculation formula for the distance weight is:

[0081] ;

[0082] Among them, represents the longitude and latitude of the delivery address in the i-th delivery area and the longitude and latitude of the distribution center the spherical distance between them, represents the distance weight of the i-th delivery area;

[0083] S33. Calculate the comprehensive priority index of different UAV delivery areas. The calculation formula for the comprehensive priority index is:

[0084] ;

[0085] Among them, represents the comprehensive priority index of the i-th delivery area, is the proportion coefficient of the task urgency weight, with a value of 0.6.

[0086] In this embodiment, calculating the resource comprehensive gap index of different UAV delivery areas in S3 includes the following specific steps:

[0087] S34. Calculate the number of UAVs required for different UAV delivery areas. The calculation formula is:

[0088] ;

[0089] Among them, represents the number of drones required for the i-th delivery area, represents rounding up, represents the maximum load of the j-th drone;

[0090] S35. Calculate the power gap index of different drones. The calculation formula of the power gap index is:

[0091] ;

[0092] Among them, represents the power gap index of the j-th drone, represents the minimum return power of the drone, represents the remaining battery power of the j-th drone;

[0093] S36. Calculate the comprehensive resource gap index of different drone delivery areas. The calculation formula of the comprehensive resource gap index is:

[0094] ;

[0095] Among them, represents the comprehensive resource gap index of the i-th delivery area, represents the resource gap weight, with a value of 0.8, represents the available status label of the j-th drone.

[0096] In this embodiment, S3 further includes: arranging the comprehensive resource gap indexes of different drone delivery areas in descending order to generate a priority sequence.

[0097] In this embodiment, the drone scheduling objective function in S4 is:

[0098] ;

[0099] Among them, is the fuel cost per unit distance of the drone, is the path length from drone j to delivery area i, is the penalty cost per unit time of order delay, is the estimated time from drone j to delivery area i, is the latest delivery time of the order in delivery area i.

[0100] In this embodiment, the constraint conditions in S4 include power constraint, load constraint, and airspace constraint; the calculation formula of the power constraint is:

[0101] ;

[0102] Among them, is the flight speed of the j-th drone, is the energy consumption rate per unit distance of the drone; the calculation formula of the load constraint is:

[0103] ;

[0104] Among them, represents the set of all delivery areas assigned to drone j; the calculation formula of the airspace constraint is:

[0105] .

[0106] Embodiment 2

[0107] This embodiment proposes an autonomous task optimization system for unmanned devices based on cross-modal data alignment, as Figure 2 shown, including: a data acquisition module, an order volume prediction module, a priority generation module, and a scheduling module;

[0108] The data acquisition module is used to collect logistics demand data of different drone delivery areas and resource data of different drones, and at the same time collect environmental data;

[0109] The order volume prediction module is used to build an order quantity prediction model for different drone delivery areas and output the predicted order volume of each delivery area in the next 2 hours;

[0110] The priority generation module is used to calculate the comprehensive priority index of different drone delivery areas and calculate the comprehensive resource gap index of different drone delivery areas, and generate a priority sequence;

[0111] The scheduling module is used to build a drone scheduling objective function and set up constraint conditions, and the constraint conditions include power constraint, load constraint, and airspace constraint, and generate an optimal scheduling plan.

[0112] In this embodiment, the implementation of the data acquisition module includes the following specific steps: First, collect the logistics demand data of different drone delivery areas, and the logistics demand data includes the order quantity , the longitude and latitude of the delivery address , the total weight of the goods and the urgency label , where i represents the i-th delivery area. When the urgency is normal, , when the urgency is a rigid demand, 1; Further collect the resource data of different drones, and the resource data includes the available status label of the drone , remaining battery power of the drone , maximum load of the drone and the longitude and latitude of the current position of the drone , where j represents the j-th drone. When the j-th drone is available, , when the j-th drone is unavailable, ; finally, environmental data is collected. The environmental data includes weather label weather, airspace restriction label of the delivery area and date type label day; where, when the weather is sunny, weather = 0, when the weather is rainy, weather = 1, when the weather is windy, weather = 2; when area i is open, , when area i is restricted, ; when the date type is a working day, day = 0, when the date type is a holiday, day = 1.

[0113] In this embodiment, the implementation of the order volume prediction module includes the following specific steps: First, an order quantity prediction model is constructed for different drone delivery areas. The order quantity prediction model takes the feature vector formed by the historical 24-hour order quantity, weather label, and date type label of a single delivery area as input, takes the order quantity prediction value corresponding to the feature vector of a single delivery area as output, takes the actual order quantity value corresponding to the feature vector of a single delivery area as the prediction target, and takes minimizing the mean square error between the actual order quantity value and the order quantity prediction value as the training target, and stops training until the mean square error between the actual order quantity value and the order quantity prediction value reaches convergence; output the order quantity of different drone delivery areas predicted by the order quantity prediction model in the next 2 hours; the order quantity prediction model is an LSTM neural network model; the training target formula of the order quantity prediction model is:

[0114] ;

[0115] where, represents the actual order quantity value of the i-th delivery area, represents the order quantity prediction value of the i-th delivery area, and n represents the total number of delivery areas.

[0116] In this embodiment, calculating the comprehensive priority index of different drone delivery areas includes the following specific steps: First, calculate the task urgency weight of different drone delivery areas. The calculation formula of the task urgency weight is:

[0117] ;

[0118] where, represents the task urgency weight of the i-th delivery area, Represents the emergency level label of the i-th delivery area Is the emergency order coefficient, with a value of 0.7; Further calculate the distance weights of different UAV delivery areas, and the calculation formula for the distance weights is:

[0119] ;

[0120] Wherein, Represents the longitude and latitude of the delivery address in the i-th delivery area And the longitude and latitude of the distribution center The spherical distance therebetween, Represents the distance weight of the i-th delivery area; Finally, calculate the comprehensive priority index of different UAV delivery areas, and the calculation formula for the comprehensive priority index is:

[0121] ;

[0122] Wherein, Represents the comprehensive priority index of the i-th delivery area, Is the proportion coefficient of the task urgency weight, with a value of 0.6.

[0123] In this embodiment, calculating the resource comprehensive gap index of different UAV delivery areas includes the following specific steps: First, calculate the number of UAVs required for different UAV delivery areas, and the calculation formula is:

[0124] ;

[0125] Wherein, Represents the number of UAVs required for the i-th delivery area, Represents rounding up, Represents the maximum load of the j-th UAV; Further calculate the power gap index of different UAVs, and the calculation formula for the power gap index is:

[0126] ;

[0127] Wherein, Represents the power gap index of the j-th UAV, Represents the minimum return power of the UAV, Represents the remaining battery power of the j-th UAV; Finally, calculate the resource comprehensive gap index of different UAV delivery areas, and the calculation formula for the resource comprehensive gap index is:

[0128] ;

[0129] Wherein, Represents the resource comprehensive gap index of the i-th delivery area, represents the resource gap weight, with a value of 0.8, represents the available status label of the j-th drone. The comprehensive resource gap indices of different drone delivery areas are sorted in descending order to generate a priority sequence.

[0130] In this embodiment, the drone scheduling objective function is:

[0131] ;

[0132] where is the fuel cost per unit distance of the drone, is the path length from the j-th drone to the delivery area i, is the penalty cost per unit time of order delay, is the estimated time from the j-th drone to the delivery area i, is the latest delivery time of the order in the delivery area i.

[0133] In this embodiment, the constraint conditions include power constraint, load constraint, and airspace constraint; the calculation formula of the power constraint is:

[0134] ;

[0135] where is the flight speed of the j-th drone, is the energy consumption rate per unit distance of the drone; the calculation formula of the load constraint is:

[0136] ;

[0137] where represents the set of all delivery areas assigned to the j-th drone; the calculation formula of the airspace constraint is: .

[0138] For the steps of the above-mentioned parameters and each unit module in the unmanned device autonomous task optimization system based on cross-modal data alignment of the present invention to implement corresponding functions, reference can be made to the parameters and steps in the embodiment of the unmanned device autonomous task optimization method based on cross-modal data alignment in Embodiment 1 above.

[0139] Embodiment 3

[0140] This embodiment provides an electronic device, including: a processor and a memory, where a computer program that can be called by the processor is stored in the memory; the processor executes the above-mentioned unmanned device autonomous task optimization method based on cross-modal data alignment by calling the computer program stored in the memory.

[0141] The electronic device may vary greatly due to different configurations or performances, and can include one or more processors (Central Processing Units, CPUs) and one or more memories. Among them, at least one computer program is stored in the memory, and the computer program is loaded and executed by the processor to implement the method for optimizing the autonomous task of the unmanned device based on cross-modal data alignment provided by the above method embodiment. The electronic device can also include other components for implementing device functions. For example, the electronic device can also have components such as wired or wireless network interfaces and input / output interfaces for data input and output. This embodiment will not be elaborated here.

[0142] Those skilled in the art know that the present invention can be implemented as a system, a method, or a computer program product. Therefore, the present disclosure can be specifically implemented in the following forms: it can be completely hardware, can be completely software (including firmware, resident software, microcode, etc.), or can also be a combination of hardware and software, which is generally referred to as "circuit", "module", or "system" in this article. In addition, in some embodiments, the present invention can also be implemented in the form of a computer program product in one or more computer-readable media, and the computer-readable media contains computer-readable program code.

[0143] Any combination of one or more computer-readable media can be adopted. The computer-readable media can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the computer-readable storage medium can be any tangible medium that contains or stores a program, and the program can be used by or combined with an instruction execution system, device, or component.

[0144] The present invention will be described with reference to the flowcharts and block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow or block in the flowcharts and block diagrams, and the combination of flows and blocks in the flowcharts or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in the Figure 1 one or more flows and blocks Figure 1 one or more blocks.

[0145] These computer program instructions can also be loaded onto a computer or other programmable data processing devices, such that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process, so that the instructions executed on the computer or other programmable devices provide steps for implementing the functions specified in the Figure 1 one or more flows and blocks Figure 1 one or more blocks.

[0146] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the spirit of the present invention and the scope protected by the claims. All of these are within the protection scope of the present invention.

Claims

1. An unmanned equipment autonomous task optimization method based on cross-modal data alignment, characterized by: The specific steps include: S1, collect logistics demand data of different drone delivery areas and resource data of different drones, and collect environmental data at the same time; S2, build an order quantity prediction model for different drone delivery areas, and output the predicted order quantity for each delivery area in the next 2 hours; S3, calculating the comprehensive priority index of different drone delivery areas, and calculating the comprehensive resource gap index of different drone delivery areas to generate a priority sequence; S4. Constructing the UAV scheduling objective function and setting constraints, wherein the constraints include power constraints, load constraints, and airspace constraints, and generating the optimal scheduling solution; The calculation of the comprehensive priority index of different drone delivery areas in S3 includes the following specific steps: S31. Calculate the mission urgency weights of different drone delivery areas. The calculation formula for the mission urgency weights is: ； in, represents the task urgency weight of the i-th delivery area, represents the urgency label of the i-th delivery area, is the emergency order coefficient, which takes a value of 0.7; S32. Calculate the distance weights of different drone delivery areas. The distance weight calculation formula is: ； in, Indicates the longitude and latitude of the delivery address in the i-th delivery area Latitude and longitude of the distribution center The spherical distance between represents the distance weight of the i-th delivery area; S33. Calculate the comprehensive priority index of different drone delivery areas. The calculation formula of the comprehensive priority index is: ； in, represents the comprehensive priority index of the i-th distribution area, is the coefficient of task urgency weight, and its value is 0.6; The calculation of the comprehensive resource gap index of different drone delivery areas in S3 includes the following specific steps: S34. Calculate the number of drones required for different drone delivery areas. The calculation formula is: ； in, represents the number of drones required for the i-th delivery area, Indicates rounding up. represents the maximum load of the jth UAV; S35. Calculate the power shortage index of different UAVs. The calculation formula of the power shortage index is: ； in, represents the power shortage index of the jth UAV, Indicates the minimum return power of the drone. represents the remaining battery power of the jth drone; S36. Calculate the comprehensive resource gap index of different drone delivery areas. The calculation formula of the comprehensive resource gap index is: ； in, represents the comprehensive resource gap index of the i-th distribution area, Indicates the resource gap weight, with a value of 0.

8. Represents the available status label of the j-th drone.

2. The method for optimizing autonomous tasks of unmanned equipment based on cross-modal data alignment according to claim 1, characterized in that: The S1 comprises the following specific steps: S11. Collect logistics demand data of different drone delivery areas, where the logistics demand data includes order quantity , Latitude and longitude of delivery address , Total weight of cargo and urgency labels , where i represents the i-th delivery area. When the urgency is normal, , when the urgency is justified, 1; S12. Collect resource data of different drones, where the resource data includes drone availability status tags , Remaining battery power of the drone 、Maximum payload of drone And the latitude and longitude of the drone’s current location , where j represents the jth UAV. When the jth UAV is available, , when the jth UAV is unavailable, ; S13, collecting environmental data, the environmental data includes weather tags, airspace restriction tags of the delivery area and date type label day; where weather is sunny, weather=0, rainy, weather=1, and windy, weather=2; when area i is open, , when area i is restricted, ; When the date type is a weekday, day=0; when the date type is a holiday, day=1.

3. The unmanned equipment autonomous task optimization method based on cross-modal data alignment according to claim 2, characterized in that: The S2 comprises the following specific steps: S21. Construct an order quantity prediction model for different drone delivery areas, wherein the order quantity prediction model uses a feature vector formed by a combination of the number of orders in the history of a single delivery area for 24 hours, a weather label, and a date type label as input, uses the predicted value of the order quantity corresponding to the feature vector of a single delivery area as output, uses the actual value of the order quantity corresponding to the feature vector of a single delivery area as a prediction target, and uses minimizing the mean square error between the actual value of the order quantity and the predicted value of the order quantity as a training target, until the mean square error between the actual value of the order quantity and the predicted value of the order quantity reaches convergence, and stops training; outputs the order quantity predicted by the order quantity prediction model for different drone delivery areas in the next 2 hours; the order quantity prediction model is an LSTM neural network model; S22. The training objective formula of the order quantity prediction model is: ； in, represents the actual value of the order quantity in the i-th delivery area, represents the predicted value of the order quantity of the i-th delivery area, and n represents the total number of delivery areas.

4. The unmanned equipment autonomous task optimization method based on cross-modal data alignment according to claim 3 is characterized in that: The S3 also includes: arranging the comprehensive resource gap indexes of different drone delivery areas in descending order to generate a priority sequence.

5. The method for optimizing autonomous tasks of unmanned equipment based on cross-modal data alignment according to claim 4, characterized in that: The objective function of the UAV scheduling in S4 is: ； in, is the fuel cost per unit distance of the drone, is the path length from drone j to delivery area i, is the penalty cost per unit time of order delay, is the estimated time for drone j to arrive at delivery area i, The latest delivery time for orders in delivery area i.

6. The method for optimizing autonomous tasks of unmanned equipment based on cross-modal data alignment according to claim 5, characterized in that: The constraints in S4 include power constraints, load constraints and airspace constraints; the calculation formula of the power constraints is: ； in, is the flight speed of the jth UAV, is the energy consumption rate per unit distance of the UAV; the calculation formula of the load constraint is: ； in, represents the set of all delivery areas assigned to drone j; the calculation formula of the airspace constraint is: 。 7. An unmanned equipment autonomous task optimization system based on cross-modal data alignment, which is implemented based on the unmanned equipment autonomous task optimization method based on cross-modal data alignment according to any one of claims 1 to 6, characterized in that: The system includes: a data collection module, an order quantity prediction module, a priority generation module, and a scheduling module; The data collection module is used to collect logistics demand data of different drone delivery areas and resource data of different drones, and collect environmental data at the same time; The order quantity prediction module is used to build an order quantity prediction model for different drone delivery areas and output the predicted order quantity for each delivery area in the next 2 hours; The priority generation module is used to calculate the comprehensive priority index of different drone delivery areas, and calculate the comprehensive resource gap index of different drone delivery areas to generate a priority sequence; The scheduling module is used to construct a UAV scheduling objective function and set constraints, which include power constraints, load constraints, and airspace constraints, to generate an optimal scheduling solution.

Images