Low-altitude intelligent cluster flight management method, equipment, medium and product

Abstract

The invention discloses a low-altitude intelligent cluster flight management method and device, a medium and a product, and relates to the technical field of aviation, and the method comprises the steps: constructing a knowledge graph based on the structured and unstructured data of a low-altitude network connection intelligent carrier; an LSTM model is utilized to analyze spatial domain features in the knowledge graph, and a traffic space-time diagram of the carrier is predicted; dividing a low-altitude airspace based on the traffic space-time diagram, and determining a low-altitude airspace distribution result; according to a low-altitude airspace distribution result, taking the flight safety degree of the carrier as a reward function, taking the time used by the flight path as a value function, and adopting a Q-learning plus traffic factor time penalty algorithm to iterate the initialized low-altitude network connection intelligent carrier path to obtain an optimal path; and according to the low-altitude network connection intelligent carrier cluster, constructing a formation flight cooperative control model with the lowest delay rate as the target. The airspace utilization efficiency and the flight safety of the low-altitude network connection intelligent carrier can be improved.

Description

Technical Field

[0001] The present application relates to the field of aviation technology, and in particular to a low-altitude intelligent cluster flight management method, equipment, medium and product. Background Art

[0002] With the rapid progress of electric propulsion technology, high energy density batteries, precise sensing technology and stable communication technology, low-altitude networked intelligent vehicles are gradually becoming a reality. Low-altitude networked intelligent vehicles have broad application prospects in logistics, monitoring, rescue and other fields.

[0003] However, similar to ground transportation, low-altitude networked intelligent vehicles also require a complete and efficient management system to ensure people's safety, comfort and travel efficiency. Given that new low-altitude networked intelligent vehicles have low entry barriers, flexible driving operations, and large numbers, there is currently a lack of effective management strategies. Summary of the invention

[0004] The purpose of this application is to provide a low-altitude intelligent cluster flight management method, equipment, medium and product, which can improve the airspace utilization efficiency and flight safety of low-altitude networked intelligent vehicles.

[0005] To achieve the above objectives, this application provides the following solutions: In a first aspect, the present application provides a low-altitude intelligent cluster flight management method, comprising: Obtain structured and unstructured data of low-altitude connected intelligent vehicles.

[0006] Based on the structured data and the unstructured data, a structured low-altitude networked intelligent vehicle knowledge graph is constructed; the nodes in the structured low-altitude networked intelligent vehicle knowledge graph are entities, and the edges are the relationships between entities; the relationships between entities are obtained by analyzing the structured data and the unstructured data through information extraction technology and knowledge representation technology.

[0007] The LSTM model is used to analyze the airspace features in the structured low-altitude networked intelligent vehicle knowledge graph to predict the flow space-time diagram of the low-altitude networked intelligent vehicle; the airspace features are the passable direction and maximum flow of the airspace.

[0008] The low-altitude airspace is divided based on the traffic space-time diagram to determine the low-altitude airspace allocation result; the low-altitude airspace allocation result is the passable direction and maximum traffic of the unit airspace.

[0009] Initialize the low-altitude networked intelligent vehicle path based on the low-altitude airspace allocation results.

[0010] Taking the flight safety of the low-altitude networked intelligent vehicle as the reward function, and the time taken for the flight path of the low-altitude networked intelligent vehicle as the value function, the Q-learning + traffic factor penalty algorithm is used to iterate the initialized low-altitude networked intelligent vehicle path to obtain the optimal path of the low-altitude networked intelligent vehicle.

[0011] According to a cluster of low-altitude networked intelligent vehicles, a formation flight collaborative control model is constructed with the goal of minimizing the delay rate; the low-altitude networked intelligent vehicle cluster is composed of low-altitude networked intelligent vehicles with the same path; the formation flight collaborative control model is used to perform flight control on the low-altitude networked intelligent vehicle cluster according to a formation control method.

[0012] In a second aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any one of the above-mentioned low-altitude intelligent cluster flight management methods.

[0013] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described low-altitude intelligent cluster flight management methods.

[0014] In a fourth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any one of the above-mentioned low-altitude intelligent cluster flight management methods.

[0015] According to the specific embodiments provided in this application, this application discloses the following technical effects: The present application provides a low-altitude intelligent cluster flight management method, equipment, medium and product. The method relies on the structured and unstructured data of low-altitude networked intelligent vehicles to construct a detailed knowledge graph. By using the long short-term memory network (LSTM) model, the airspace characteristics in the knowledge graph are deeply analyzed to predict the spatiotemporal distribution map of vehicle traffic. Based on this spatiotemporal distribution map of traffic, the present application proposes a low-altitude airspace division method to determine the airspace allocation plan. Subsequently, based on the airspace allocation results, the flight safety of the vehicle is used as the reward function, and the time required for the flight path is used as the value function. The Q-learning algorithm is combined with the traffic factor penalty strategy to iteratively optimize the initial low-altitude networked intelligent vehicle path to obtain the optimal flight path. In addition, the present application also constructs a formation flight collaborative control model with the goal of minimizing the delay rate, which is suitable for low-altitude networked intelligent vehicle clusters. Through these technical means, the present application significantly improves the airspace utilization efficiency and flight safety performance of low-altitude networked intelligent vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 This is an application environment diagram of a low-altitude intelligent cluster flight management method in one embodiment of the present application.

[0018] Figure 2 A flowchart of a low-altitude intelligent cluster flight management method provided for one embodiment of the present application.

[0019] Figure 3 A schematic diagram of formation control based on a goose-like bionic formation provided in one embodiment of the present application.

[0020] Figure 4 A system overall architecture diagram provided for one embodiment of the present application.

[0021] Figure 5 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0023] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0024] The low-altitude intelligent cluster flight management method provided in the embodiment of the present application can be applied to Figure 1In the application environment shown, the terminal 102 communicates with the server 104 through a network. The data storage system can store data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. Terminal 102 can send the structured data and unstructured data of the low-altitude networked intelligent vehicle to be processed to server 104. After server 104 receives the structured data and unstructured data of the low-altitude networked intelligent vehicle to be processed, server 104 constructs a structured low-altitude networked intelligent vehicle knowledge graph based on the structured data and the unstructured data for the low-altitude networked intelligent vehicle to be processed; the nodes in the structured low-altitude networked intelligent vehicle knowledge graph are entities, and the edges are the relationships between entities; the relationships between entities are obtained by analyzing structured data and unstructured data through information extraction technology and knowledge representation technology; the airspace features in the structured low-altitude networked intelligent vehicle knowledge graph are analyzed using the LSTM model to predict the spatiotemporal flow diagram of the low-altitude networked intelligent vehicle; the airspace features are the passable directions of the airspace and maximum flow; based on the flow time-space diagram, the low-altitude airspace is divided to determine the low-altitude airspace allocation result; the low-altitude airspace allocation result is the passable direction and maximum flow of the unit airspace; according to the low-altitude airspace allocation result, the low-altitude networked intelligent vehicle path is initialized; the flight safety of the low-altitude networked intelligent vehicle is used as the reward function, the time used for the flight path of the low-altitude networked intelligent vehicle is used as the value function, and the Q-learning + traffic factor penalty time algorithm is used to iterate the initialized low-altitude networked intelligent vehicle path to obtain the optimal path of the low-altitude networked intelligent vehicle; according to the low-altitude networked intelligent vehicle cluster, a formation flight collaborative control model with the lowest delay rate as the goal is constructed; the low-altitude networked intelligent vehicle cluster is composed of low-altitude networked intelligent vehicles with the same path; the formation flight collaborative control model is used to control the flight of the low-altitude networked intelligent vehicle cluster according to the formation control method. The server 104 can feedback the obtained formation flight collaborative control model to the terminal 102. In addition, in some embodiments, the low-altitude intelligent cluster flight management method can also be implemented independently by the server 104 or the terminal 102. For example, the terminal 102 can directly perform data processing on the structured data and unstructured data of the low-altitude networked intelligent vehicle to be processed, or the server 104 can obtain the structured data and unstructured data of the low-altitude networked intelligent vehicle to be processed from the data storage system for data processing.

[0025] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart vehicle-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.

[0026] In an exemplary embodiment, Figure 2 As shown, a low-altitude intelligent cluster flight management method is provided. The method is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used as an example to illustrate, including the following steps 201 to 207. Among them: Step 201: Obtain structured data and unstructured data of the low-altitude networked intelligent vehicle.

[0027] Step 202: Based on the structured data and the unstructured data, a structured low-altitude networked intelligent vehicle knowledge graph is constructed; the nodes in the structured low-altitude networked intelligent vehicle knowledge graph are entities, and the edges are the relationships between entities; the relationships between entities are obtained by analyzing the structured data and the unstructured data through information extraction technology and knowledge representation technology.

[0028] Step 203: Use the LSTM model to analyze the airspace features in the structured low-altitude networked intelligent vehicle knowledge graph to predict the flow space-time diagram of the low-altitude networked intelligent vehicle; the airspace features are the passable direction and maximum flow of the airspace.

[0029] Step 204: Divide the low-altitude airspace based on the traffic space-time diagram and determine the low-altitude airspace allocation result; the low-altitude airspace allocation result is the passable direction and maximum traffic of the unit airspace.

[0030] Step 205: Initialize the low-altitude networked intelligent vehicle path based on the low-altitude airspace allocation result.

[0031] Step 206: Taking the flight safety of the low-altitude networked intelligent vehicle as the reward function and the time taken for the flight path of the low-altitude networked intelligent vehicle as the value function, the Q-learning + traffic factor penalty algorithm is used to iterate the initialized low-altitude networked intelligent vehicle path to obtain the optimal path of the low-altitude networked intelligent vehicle.

[0032] Step 207: Based on the low-altitude networked intelligent vehicle cluster, a formation flight collaborative control model with the goal of minimizing the delay rate is constructed; the low-altitude networked intelligent vehicle cluster is composed of low-altitude networked intelligent vehicles with the same path; the formation flight collaborative control model is used to perform flight control on the low-altitude networked intelligent vehicle cluster according to the formation control method.

[0033] When executing step 201, the specific steps may be as follows: The structured data obtained for the low-altitude networked intelligent vehicle is the flight log, and the unstructured data obtained for the low-altitude networked intelligent vehicle is the data entries that do not follow a fixed format or structure.

[0034] When executing step 202, the specific steps may be as follows: Step 301: Preprocess the structured data and unstructured data of the low-altitude networked intelligent vehicle to obtain preprocessed information.

[0035] Step 302: Use a feature-based machine learning method to extract traffic entities from the preprocessed information.

[0036] Step 303: Use the semi-supervised learning Traffic-Snowball algorithm to extract the association relationship in the traffic information corpus; the traffic entity information corpus is composed of information of several traffic entities.

[0037] Step 304: Use data mining methods to mine the relationship patterns between traffic entity attributes and attribute values ​​from the travel information of low-altitude networked intelligent vehicles.

[0038] Step 305: Create nodes and edges in a structured low-altitude networked intelligent vehicle knowledge graph based on the extracted traffic entities, association relationships, and relationship patterns to obtain a structured low-altitude networked intelligent vehicle knowledge graph.

[0039] Specifically, the structured data (such as flight logs) and unstructured data (such as entries) of low-altitude networked intelligent vehicles are preprocessed to remove privacy information and noise. Data cleaning and data filtering methods are used to ensure data quality and consistency. Structured and unstructured data from different sources are matched and aligned to ensure data consistency and integrity.

[0040] Among them, information extraction is: using feature-based machine learning methods to extract traffic entities, using the BiLSTM model. Given an input sequence , for t The flow of low-altitude networked intelligent vehicles at all times, the output of the model is: (1).

[0041] In the formula, and are the hidden states of the forward and backward LSTM at time t, It is the final feature representation at time t, forward means forward, and backword means backward. BiLSTM is used for information extraction, emphasizing bidirectionality. The output result is a combination of sequential analysis and reverse analysis. is the hidden state of the forward LSTM at time t-1, is the hidden state of the backward LSTM at time t+1.

[0042] Specifically, the Traffic-Snowball algorithm of semi-supervised learning is used to extract correlation relationships from traffic information corpus, including: 1) Initial seed set: First, some initial seed relation instances (seed samples) are selected from the basic traffic corpus. These instances include known entity pairs and their relations.

[0043] 2) Feature extraction: Extract features from seed relationship instances to represent these relationship instances. These features can be word vectors, context patterns, part-of-speech tags, etc. is the feature set, For the initial seed set: (2).

[0044] 3) Pattern generation: Generate relational patterns based on the extracted features. These patterns can be string patterns, part-of-speech patterns, etc. For a collection of patterns: (3).

[0045] 4) Pattern matching: Use the generated patterns to match new relation instances in unlabeled corpus. is an unlabeled corpus, and represents the two entities in an entity pair, For existing relationships, The new set of matched relationship instances is: (4).

[0046] 5) Instance filtering: Filter the newly matched relationship instances, retain instances with high confidence, and remove noise. is the confidence threshold, is a confidence function that returns the probability of extracting the correct content from the two entities e1 and e2 during training under the existing relationship r. To filter the results: (5).

[0047] 6) Iteration process: Add the filtered new instances to the seed set, repeat the above process, and gradually expand the set of relationship instances.

[0048] Specifically, data mining methods are used to mine the relationship patterns between traffic entity attributes and attribute values ​​from the travel information of low-altitude networked intelligent vehicles, including: 1) Generate candidate item sets: Starting from a single item set, gradually generate candidate item sets for the item set.

[0049] 2) Calculate support: Calculate the support of each candidate item set and retain the item sets whose support is greater than or equal to the minimum support threshold.

[0050] 3) Generate frequent item sets: Generate candidate +1 item sets based on frequent item sets, and repeat the above steps until no new candidate item sets can be generated.

[0051] 4) Generate association rules: Generate association rules from frequent item sets, calculate the confidence and lift of each rule, and retain the rules that meet the minimum confidence threshold.

[0052] 5) Represent the extracted knowledge as entities, relationships, and attributes in the knowledge graph, using RDF (Resource Description Framework) as the semantic representation format, and eliminate ambiguity and duplication. Based on the extracted knowledge and fused data, create nodes (entities) and edges (relationships) in the knowledge graph to build a structured low-altitude networked intelligent vehicle knowledge graph.

[0053] When executing step 203, the specific steps may be as follows: The LSTM model is used to analyze the airspace features in the structured low-altitude networked intelligent vehicle knowledge graph, predict the flow space-time graph of low-altitude networked intelligent vehicles, and thus perform airspace division: (6).

[0054] In the formula, is the input gate, For the Gate of Forgetfulness, is the output gate, is the sigmoid function, and is the weight matrix and bias vector of the model, subscript x and h Corresponding to X and h vector, i , f , o They correspond to the input gate formula, the forget gate formula and the output gate formula respectively. for t The traffic volume of low-altitude connected intelligent vehicles at all times, yes t The final feature representation at time -1. The cell state update equation is: (7).

[0055] In the formula, C t for t The cell state at the moment, t The information left behind at all times, C t-1 for t -1 cell state at time instant, W is the weight matrix, subscript x and h To represent the corresponding vector, c Coefficient representing the cell state calculation.

[0056] The hidden state update equation is: (8).

[0057] Finally, the output of the LSTM model As input data The encoding is then passed through a linear layer or a convolutional layer to predict the spatiotemporal graph .

[0058] When executing steps 204-206, the specific steps may be as follows: The continuous low-altitude airspace is subdivided into rasterized areas, and then the low-altitude airspace is divided based on the traffic space-time diagram to determine the low-altitude airspace allocation result; the low-altitude airspace allocation result is the passable direction and maximum traffic of the unit airspace.

[0059] An environmental model is constructed based on the low-altitude airspace allocation results; the environmental model includes low-altitude intelligent vehicles, route networks, flight rules and potential conflict points.

[0060] Using reinforcement learning methods, the optimal path and speed under the current environmental state are determined in the environmental model.

[0061] The flight safety composed of time consumption, speed change smoothness, power consumption and conflict risk level is used as the reward function, and the time used for the flight path of the low-altitude networked intelligent vehicle is used as the value function. The Q-learning + traffic factor penalty algorithm is used to iterate the current environmental state to obtain the optimal path.

[0062] Specifically, define the state transfer function: (9).

[0063] In the formula, It is the initial state of the low-altitude networked intelligent vehicle. For this moment, is the posterior probability of taking this action in this state.

[0064] Define the reward function: (10).

[0065] In the formula, Indicates the time taken for different paths. is the longest time consumed in the currently explored path, is the delay coefficient, and are the comfort penalty factors for acceleration magnitude and direction, For speed, is the acceleration, For the energy consumed, For spatial location. For time consumption, For the speed change smoothness, For power consumption, The conflict risk level.

[0066] The "Q-learning + traffic penalty" algorithm is used to iteratively find the optimal path based on the current airspace conditions.

[0067] (11).

[0068] in is the learning rate, is the discount factor, is to execute the action to get the next state, and is the traffic penalty coefficient, is in state The maximum Q value after executing all possible actions, t is the reduction factor determined based on the historical traffic conditions of the city. R(s,d) Indicates in status s Take action d Instant rewards when you win.

[0069] In the process of executing step 207, the specific operation steps can be described in detail as follows: First, establish a low-altitude networked intelligent vehicle cluster: according to the path planning results of low-altitude networked intelligent vehicles, low-altitude networked intelligent vehicles with the same path are classified into the same cluster. This process involves analyzing and classifying the paths of the vehicles to ensure that the vehicles in the cluster have similar flight routes, thereby improving the efficiency of cluster management and the effectiveness of collaborative work.

[0070] Secondly, build a dynamic model: establish a six-degree-of-freedom low-altitude networked intelligent vehicle dynamic model to describe the vehicle's motion characteristics in detail. This model will include parameters such as the vehicle's position, velocity, acceleration, and attitude in three-dimensional space, so as to accurately simulate the dynamic behavior of the vehicle during flight.

[0071] Next, establish a collaborative control model: build a formation flight collaborative control model with the goal of minimizing the delay rate. By adjusting the formation flight strategy, the optimal formation control is achieved. Specifically, this model includes the following aspects: 1) Pilot mode: In this state, the low-altitude intelligent vehicle acts as a communication node between the entire cluster and the central control system. According to the overall route planning and real-time environmental conditions, the flight posture of the entire team is determined, and the instructions are issued to each following low-altitude networked intelligent vehicle. This mode ensures the unity and coordination of cluster flight.

[0072] 2) Follow mode: In this state, the low-altitude intelligent vehicle acts as the execution terminal of flight decisions and as a sensor to transmit key information such as flight speed, atmospheric pressure, wind speed, wind direction, etc. The follower vehicle flies according to the instructions of the pilot vehicle to ensure the flight safety and efficiency of the entire cluster.

[0073] 3) Transformation mode: This mode is used to respond to different weather conditions and change the three-dimensional configuration of the entire formation. The pilot low-altitude networked intelligent vehicle issues a transformation command, and the following vehicle cancels the following state and searches for the optimal position from the middle to both sides. This process ensures the stability and safety of the formation in adverse weather conditions.

[0074] 4) Disengagement mode: In this state, the low-altitude networked intelligent vehicle reaches the destination airspace and leaves the formation. The pilot low-altitude networked intelligent vehicle responds to the separation request and gives a separation gesture. This mode ensures that the vehicle can safely and orderly disengage from the formation and complete the mission.

[0075] Specifically, the basic control methods of the flight process include: (12).

[0076] In the formula, and is the position and velocity of the navigator, is the time step, and the current relative position to the pilot vehicle is , the relative position of the target is , and To control the gain, is the control input, and is the position and speed of the vehicle that has left the formation. Calculate the fitness and select a solution, and find the optimal solution through iterative updates.

[0077] In real-scene control, algorithms with different focuses are selected to control different modes.

[0078] In the pilot mode, the low-altitude networked intelligent vehicle acts as the communication node of the cluster, determining the flight posture of the entire team based on the overall route planning and real-time environmental conditions.

[0079] In follow mode, the low-altitude connected intelligent vehicle executes flight decisions and transmits sensor data back.

[0080] The transformation mode is used to cope with different weather conditions, and the three-dimensional configuration transformation of the formation is achieved by issuing commands through the piloted low-altitude networked intelligent vehicle.

[0081] In the disengagement mode, the low-altitude networked intelligent vehicle reaches the destination airspace and disengages from the formation.

[0082] The present application also provides an application scenario, which applies the above-mentioned low-altitude intelligent cluster flight management method. Specifically: Use search engines, navigation software and other data sources to collect triple information on space and time occupied by vehicles, affiliations, and paths to build a knowledge graph, and produce a low-altitude networked intelligent vehicle traffic spatiotemporal diagram. Use the LSTM model to dynamically divide the low-altitude airspace and assign headings, and use the spatiotemporal attention mechanism to reinforce learning of key spatiotemporal features. Establish an environmental model that matches the real environment, including low-altitude intelligent vehicles, road networks, traffic rules, and potential conflict points. Use reinforcement learning methods to select the best path and speed adjustment actions under the current environmental conditions. Through time consumption , Speed ​​change stability , Power consumption and conflict risk level Set the reward function and get the optimal path.

[0083] (12).

[0084] In the formula, it represents Different paths take time. is the longest time consumed in the currently explored path, is the delay coefficient, and are the comfort penalty factors for acceleration magnitude and direction, For speed, is the acceleration, For the energy consumed, The optimal solution is obtained through the interaction between low-altitude networked intelligent vehicles and the environment.

[0085] The flight process includes pilot mode, follow mode, change mode and disengage mode. The basic control methods include: (13).

[0086] in, and is the position and velocity of the navigator, is the time step, and the current relative position to the pilot vehicle is , the relative position of the target is , and To control the gain, is the control input, and is the position and speed of the vehicle that has left the formation. Calculate the fitness and select a solution, and find the optimal solution through iterative updates: like Figure 3 As shown in the figure, the control method based on the formation control schematic diagram of the goose flock bionic formation is as follows: In real-scene control, algorithms with different focuses are selected to control different modes.

[0087] In the pilot mode, the low-altitude networked intelligent vehicle acts as the communication node of the cluster, determining the flight posture of the entire team based on the overall route planning and real-time environmental conditions.

[0088] In follow mode, the low-altitude connected intelligent vehicle executes flight decisions and transmits sensor data back.

[0089] The transformation mode is used to cope with different weather conditions, and the three-dimensional configuration transformation of the formation is achieved by issuing commands through the piloted low-altitude networked intelligent vehicle.

[0090] In the disengagement mode, the low-altitude networked intelligent vehicle reaches the destination airspace and disengages from the formation.

[0091] In an exemplary embodiment, Figure 4 As shown in the figure, the low-altitude networked intelligent vehicle control system adopts a hierarchical control strategy to manage all vehicles and high-altitude flying vehicles. The system predicts the empty volume for all vehicles and uses knowledge graphs to obtain the following information: 1) travel planning clues; 2) resource distribution in the region; 3) historical data routes. In addition, trajectory planning achieves collective optimization through reinforcement learning. For high-altitude flying vehicles, the system realizes high-altitude formation flight and uses goose-like cluster control technology to determine the pilot mode, follow mode, change mode, and disengagement mode.

[0092] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 5As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store control data of the formation flight collaborative control model. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a low-altitude intelligent cluster flight management method is implemented.

[0093] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0094] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0095] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0096] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0097] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0098] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0099] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0100] In summary, this application has the following technical effects: Based on the search terms and travel orders combined with historical data, a highly accurate forecast of the number of empty rows on the day can be obtained. The empty row forecast method based on the knowledge graph can clarify the starting point, end point and most likely route of most trips, and can resolve congestion when airspace ownership is divided in advance. In terms of real-time route planning, it can reduce the situation where a better route is found halfway through the trip, ensuring that the travel route is the optimal solution within the travel time period in most cases.

[0101] The path planning algorithm based on reinforcement learning can find the path planning method with the smallest overall delay for all low-altitude connected intelligent vehicles per unit time while ensuring safety. Its main advantage is to reduce the overall delay rate, thereby improving the reliability of the estimated time required for individual trips, reducing the time passengers have to reserve for emergencies, and improving time utilization efficiency.

[0102] The low-altitude networked intelligent vehicle formation system based on goose-like cluster control can greatly reduce the total energy consumption of low-altitude networked intelligent vehicles in a cluster by utilizing the wake of the preceding vehicle. Since the entire formation simultaneously responds to airflow disturbances in the air and influences each other to form a system, it can reduce turbulence during flight and improve the safety, stability and comfort of the journey.

[0103] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0104] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A low-altitude intelligent cluster flight management method, characterized in that: include: Obtain structured and unstructured data of low-altitude networked intelligent vehicles; Based on the structured data and the unstructured data, construct a structured low-altitude networked intelligent vehicle knowledge graph; The nodes in the structured low-altitude networked intelligent vehicle knowledge graph are entities, and the edges are relationships between entities; the relationships between entities are obtained by analyzing structured data and unstructured data through information extraction technology and knowledge representation technology; The LSTM model is used to analyze the airspace features in the structured low-altitude networked intelligent vehicle knowledge graph to predict the flow space-time graph of the low-altitude networked intelligent vehicle; the airspace features are the passable direction and maximum flow of the airspace; Dividing the low-altitude airspace based on the traffic space-time diagram and determining the low-altitude airspace allocation result; The low-altitude airspace allocation result is the passable direction and maximum flow rate of the unit airspace; Initialize the low-altitude networked intelligent vehicle path according to the low-altitude airspace allocation result; Taking the flight safety of the low-altitude networked intelligent vehicle as the reward function and the time taken by the low-altitude networked intelligent vehicle's flight path as the value function, the Q-learning + traffic factor penalty algorithm is used to iterate the initialized low-altitude networked intelligent vehicle path to obtain the optimal path of the low-altitude networked intelligent vehicle; According to a cluster of low-altitude networked intelligent vehicles, a formation flight collaborative control model is constructed with the goal of minimizing the delay rate; the low-altitude networked intelligent vehicle cluster is composed of low-altitude networked intelligent vehicles with the same path; the formation flight collaborative control model is used to perform flight control on the low-altitude networked intelligent vehicle cluster according to a formation control method.

2. A low-altitude intelligent cluster flight management method according to claim 1, characterized in that: Based on the structured data and the unstructured data, a structured low-altitude networked intelligent vehicle knowledge graph is constructed, specifically including: Preprocess the structured data and unstructured data of the low-altitude networked intelligent vehicle to obtain preprocessed information; Use feature-based machine learning methods to extract traffic entities from preprocessed information; Using the semi-supervised learning Traffic-Snowball algorithm to extract the correlation relationship in the traffic information corpus; the traffic entity information corpus is composed of information of several traffic entities; Using data mining methods, we can mine the relationship patterns between the attributes of traffic entities and their values ​​from the travel information of low-altitude networked intelligent vehicles. According to the extracted traffic entities, associations and relationship patterns, nodes and edges in the structured low-altitude networked intelligent vehicle knowledge graph are created to obtain the structured low-altitude networked intelligent vehicle knowledge graph.

3. A low-altitude intelligent cluster flight management method according to claim 1, characterized in that: The LSTM model is used to analyze the airspace features in the structured low-altitude networked intelligent vehicle knowledge graph to predict the flow space-time graph of low-altitude networked intelligent vehicles, including: According to the formula Predict the spatiotemporal flow of low-altitude networked intelligent vehicles; in, is the input gate, For the Gate of Forgetfulness, is the output gate, is the sigmoid function, and is the weight matrix and bias vector of the model, subscript x and h Corresponding to X and h vector, i , f , o They correspond to the input gate formula, the forget gate formula and the output gate formula respectively. for t The traffic volume of low-altitude connected intelligent vehicles at all times, yes t The final feature representation at time -1.

4. A low-altitude intelligent cluster flight management method according to claim 1, characterized in that: Taking the flight safety of the low-altitude networked intelligent vehicle as the reward function and the time taken by the low-altitude networked intelligent vehicle's flight path as the value function, the Q-learning + traffic factor penalty algorithm is used to iterate the initialized low-altitude networked intelligent vehicle path to obtain the optimal path of the low-altitude networked intelligent vehicle, which includes: According to the low-altitude airspace allocation results, an environmental model is constructed; the environmental model includes low-altitude intelligent vehicles, route networks, flight rules and potential conflict points; Using reinforcement learning methods, determine the optimal path and speed in the current environment state in the environment model; The flight safety composed of time consumption, speed change smoothness, power consumption and conflict risk level is used as the reward function, and the time used for the flight path of the low-altitude networked intelligent vehicle is used as the value function. The Q-learning + traffic factor penalty algorithm is used to iterate the current environmental state to obtain the optimal path.

5. A low-altitude intelligent cluster flight management method according to claim 1, characterized in that: The Q-learning + traffic penalty algorithm is iterated to obtain the optimal path of the low-altitude networked intelligent vehicle: ； in is the learning rate, is the discount factor, is to execute the action to get the next state, and is the traffic penalty coefficient, is in state The maximum Q value after executing all possible actions, t is the reduction factor determined based on the historical traffic conditions of the city. R(s,d) It represents the immediate reward obtained when taking action d in state s.

6. A low-altitude intelligent cluster flight management method according to claim 1, characterized in that: The formation control method includes a pilot mode, a follow mode, a change mode and a breakaway mode.

7. The low-altitude intelligent cluster flight management method according to claim 1, characterized in that: The structured data of the low-altitude networked intelligent vehicle is the flight log, and the unstructured data of the low-altitude networked intelligent vehicle is the data entries that do not follow a fixed format or structure.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a low-altitude intelligent cluster flight management method as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, a low-altitude intelligent cluster flight management method described in any one of claims 1 to 7 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, a low-altitude intelligent cluster flight management method described in any one of claims 1 to 7 is implemented.

Images