Emergency flight plan generation methods, devices, equipment, media and procedures

Abstract

This application relates to the field of emergency communication technology, providing an emergency flight plan generation method, apparatus, equipment, medium, and program product. The method includes: predicting emergency events within an emergency planning area to obtain event description information; generating multi-dimensional labels for the emergency events based on the event description information; inputting the multi-dimensional labels and event description information into a plan generation model, and using the plan generation model to generate low-altitude emergency flight plans within the emergency planning area. The plan generation model is obtained through iterative training based on a sample dataset, which is constructed based on historical emergency events within the emergency planning area and their corresponding historical flight plans. The plan generation model can generate emergency flight plans online in real time based on the prediction results of emergency events, improving the efficiency and accuracy of emergency flight plan generation and facilitating the meeting of the frequent requests for low-altitude emergency flight plans.

Description

Technical Field

[0001] This application relates to the field of emergency communication technology, and in particular to an emergency flight plan generation method, apparatus, equipment, medium and program product. Background Technology

[0002] With the development of drone technology, drones are increasingly used in emergency rescue. As natural disasters and other emergencies become more frequent, intense, and widespread, emergency management units in various regions typically need to submit low-altitude emergency flight plans before an emergency occurs, thereby applying for low-altitude emergency flight permissions after the event. Currently, the generation of low-altitude emergency flight plans mainly relies on manual or expert decision-making, with applications for flight permissions submitted annually or semi-annually. This method is inefficient, time-consuming, labor-intensive, and prone to errors in judgment, making it difficult to meet the current demand for frequent applications for low-altitude emergency airspace. Summary of the Invention

[0003] This application provides an emergency flight plan generation method, apparatus, equipment, medium, and program product to address the shortcomings of existing technologies that rely on manual or expert decision-making to generate emergency flight plans, which are inefficient, time-consuming, and prone to errors, making it difficult to meet the current demand for frequent applications for low-altitude emergency airspace.

[0004] This application provides a method for generating emergency flight plans, including:

[0005] Predict emergency events within the emergency planning area to obtain event description information of the emergency events;

[0006] A multidimensional tag for the emergency event is generated based on the event description information; the multidimensional tag includes at least an event time tag, an event type tag, and a three-part event tag.

[0007] The multidimensional labels and the event description information are input into the plan generation model, and the plan generation model is used to generate an emergency flight plan for the low-altitude airspace within the emergency planning area.

[0008] The plan generation model is obtained through iterative training based on a sample dataset, which is constructed based on historical emergency events within the emergency planning area and the historical flight plans corresponding to those historical emergency events.

[0009] In one embodiment, generating a multidimensional tag for the emergency event based on the event description information includes:

[0010] The event description information is input into the encoder of the tag generator; the encoder includes a first word embedding layer and a long short-term memory layer, and the tag generator also includes a decoder, which includes a first fully connected layer and a first output layer.

[0011] The event description information is converted into word vectors using the first word embedding layer in the encoder, and feature vectors are extracted from the word vectors using the long short-term memory layer.

[0012] The feature vector is input into the first fully connected layer of the decoder for decoding processing, and the decoded feature vector is classified and predicted based on the first output layer of the decoder to obtain the multidimensional label of the emergency event.

[0013] In one embodiment, the plan generation model includes a multi-label input layer, a second word embedding layer, a large language model layer, a second fully connected layer, a dropout layer, and a second output layer; the step of inputting the multi-dimensional labels and the event description information into the plan generation model, and using the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area, includes:

[0014] The multidimensional labels and the event description information are input into the plan generation model. The multidimensional labels and the event description information are then concatenated using the multi-label input layer of the plan generation model and input into the second word embedding layer of the plan generation model.

[0015] The second word embedding layer converts the concatenated multidimensional tags and event description information into target word vectors.

[0016] The target word vector is processed by sequentially extracting features from the large language model layer, the second fully connected layer, and the dropout layer of the aforementioned plan generation model to obtain the target feature vector.

[0017] The target feature vector is input into the second output layer of the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area; the emergency flight plan includes the flight time period, flight area, and flight mission.

[0018] In one embodiment, inputting the multidimensional labels and the event description information into a plan generation model, and using the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area, includes:

[0019] Obtain the time granularity at which the emergency planning area generates the emergency flight plan;

[0020] The multidimensional labels and the event description information are input into the plan generation model, and the plan generation model is used to generate an emergency flight plan for the low airspace within the emergency planning area at the time granularity.

[0021] In one embodiment, before inputting the multidimensional labels and the event description information into the plan generation model, and using the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area, the method further includes:

[0022] Obtain information on historical emergency events within the emergency planning area;

[0023] Historical description information and multi-dimensional information of the historical emergency events are extracted from the information; the multi-dimensional information includes event time, event type, and three-disruption event identifier; the three-disruption event identifier is used to characterize whether the historical emergency event is a three-disruption event.

[0024] Obtain the historical emergency flight plans of the historical emergency events, use the multidimensional information as the multidimensional label of the historical descriptive information, and use the historical emergency flight plans as the prediction results of the historical descriptive information to construct a sample dataset;

[0025] The preset plan generation model is iteratively trained using the sample dataset to obtain a pre-trained plan generation model.

[0026] In one embodiment, the step of iteratively training a pre-defined plan generation model using the sample dataset to obtain a pre-trained plan generation model includes:

[0027] Based on a preset number of iterations, the sample dataset is input into a preset plan generation model, and the plan generation model is iteratively trained.

[0028] In each iteration, a pre-defined multi-class cross-entropy is used as the loss function, and a pre-defined gradient descent optimization algorithm is employed to adjust the model parameters of the planned generation model in order to optimize the function value of the loss function. When the function value of the loss function is optimal, a pre-trained planned generation model is obtained.

[0029] This application also provides an emergency flight plan generation device, including the following modules:

[0030] The event prediction module is used to predict emergency events within the emergency planning area and obtain event description information of the emergency events.

[0031] The tag generation module is used to generate multi-dimensional tags for the emergency event based on the event description information; the multi-dimensional tags include at least an event time tag, an event type tag, and a three-part event tag.

[0032] The plan generation module is used to input the multidimensional labels and the event description information into the plan generation model, and use the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area;

[0033] The plan generation model is obtained through iterative training based on a sample dataset, which is constructed based on historical emergency events within the emergency planning area and the historical flight plans corresponding to those historical emergency events.

[0034] This application also provides an electronic device, including 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 the emergency flight plan generation method as described above.

[0035] This application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the emergency flight plan generation method as described above.

[0036] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the emergency flight plan generation method as described above.

[0037] The emergency flight plan generation method, apparatus, equipment, medium, and program products provided in this application predict emergency events and generate multi-dimensional tags based on the predicted emergency event description information. Using a plan generation model, corresponding low-altitude emergency flight plans can be generated based on the multi-dimensional tags and description information of the emergency events, achieving automatic generation of emergency flight plans. Through the plan generation model, emergency flight plans for low-altitude airspace within the emergency planning area can be generated online in real time based on the prediction results of emergency events, improving the efficiency and accuracy of emergency flight plan generation, saving time and effort. Furthermore, it enables automatic submission and application for low-altitude emergency flight permissions based on the emergency flight plans, which is beneficial for meeting the frequent needs of low-altitude emergency flight plan applications. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a flowchart illustrating the emergency flight plan generation method provided in the embodiments of this application.

[0040] Figure 2 This is a schematic diagram of the multidimensional tag generation process provided in the embodiments of this application.

[0041] Figure 3 This is one of the schematic diagrams of the emergency flight plan generation process provided in the embodiments of this application.

[0042] Figure 4 This is the second schematic diagram of the emergency flight plan generation process provided in the embodiments of this application.

[0043] Figure 5 This is a schematic diagram of the training process of the planned generation model provided in the embodiments of this application.

[0044] Figure 6 This is a schematic diagram of the emergency flight plan generation device provided by the present invention.

[0045] Figure 7 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0047] Figure 1 This is a flowchart illustrating the emergency flight plan generation method provided in this application embodiment, such as... Figure 1 As shown, the method includes the following steps:

[0048] Step 100: Predict emergency events within the emergency planning area to obtain event description information of the emergency events;

[0049] Step 200: Generate a multidimensional tag for the emergency event based on the event description information; the multidimensional tag includes at least an event time tag, an event type tag, and a three-part event tag;

[0050] Step 300: Input the multidimensional labels and the event description information into the plan generation model, and use the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area;

[0051] The plan generation model is obtained through iterative training based on a sample dataset, which is constructed based on historical emergency events within the emergency planning area and the historical flight plans corresponding to those historical emergency events.

[0052] Predicting emergency events within the emergency planning area yields event description information. Prediction methods include, but are not limited to, using predictive models and mining patterns from historical emergency events to predict their occurrence. When using predictive models, these models can be constructed from a sample dataset of historical emergency events within the emergency planning area, and a pre-defined base prediction model can be iteratively trained.

[0053] Optionally, the prediction of emergency events is based on environmental information within the emergency planning area. This involves monitoring the environmental information within the emergency planning area, extracting environmental characteristics, and matching these characteristics with historical emergency events to determine whether a similar emergency event will occur. The collected environmental information includes meteorological data.

[0054] Furthermore, the event description information of an emergency event is the key information used to describe the emergency event. This key information includes at least the time when the emergency event occurred, the type of the emergency event, and whether the emergency event is a "three-disruption event," which refers to an event involving "road disconnection, power outage, and network outage."

[0055] Based on the predicted event description information, multidimensional tags are generated for the emergency events. These tags include an event time tag, an event type tag, and a three-disruption event tag. The event time tag represents the occurrence time of the emergency event and is generated based on the time information extracted from the event description information. The event type tag represents the event type of the emergency event and is generated based on the event type extracted from the event description information. The three-disruption event tag represents whether the emergency event is a three-disruption event, and is generated based on the three-disruption event identifier extracted from the event description information.

[0056] The multidimensional labels and event description information of the generated emergency events are input into the plan generation model. The plan generation model is then used to generate an emergency flight plan for the low-altitude airspace within the emergency planning area. This emergency flight plan includes at least the flight time period, flight area, and flight mission.

[0057] In this embodiment, by predicting emergency events and generating multi-dimensional tags based on the predicted event descriptions, a plan generation model can be used to generate corresponding low-altitude emergency flight plans based on the multi-dimensional tags and descriptions of the emergency events, thus achieving automatic generation of emergency flight plans. Through the plan generation model, emergency flight plans for low-altitude airspace within the emergency planning area can be generated online in real time based on the predicted event results, improving the efficiency and accuracy of emergency flight plan generation, saving time and effort. Furthermore, it allows for the automatic submission and application of low-altitude emergency flight permissions based on the emergency flight plans, which is beneficial for meeting the frequent needs of low-altitude emergency flight plan applications.

[0058] Furthermore, the multidimensional labels for emergency events are generated based on a label generator, which is pre-trained based on historical emergency events. Step 200 may also include:

[0059] Step 201: Input the event description information into the encoder of the tag generator; the encoder includes a first word embedding layer and a long short-term memory layer, and the tag generator also includes a decoder, the decoder includes a first fully connected layer and a first output layer;

[0060] Step 202: Using the first word embedding layer in the encoder, the event description information is converted into word vectors, and feature vectors are extracted from the word vectors through the long short-term memory layer;

[0061] Step 203: Input the feature vector into the first fully connected layer of the decoder for decoding processing, and perform label classification prediction on the decoded feature vector based on the first output layer of the decoder to obtain the multidimensional label of the emergency event.

[0062] The label generator is built on a Long Short-Term Memory (LSTM) network and mainly consists of an encoder and a decoder. The encoder further includes a first word embedding layer and an LSTM layer, while the decoder includes a first fully connected layer and a first output layer. The label generator also includes an input layer; optionally, the LSTM layer may include one or more layers, and the first fully connected layer in the decoder may also include one or more layers.

[0063] Reference Figure 2 The label generation process shown is as follows: First, the event description information of the emergency event is input into the encoder of the label generator through the input layer of the label generator. The first word embedding layer of the encoder is used to convert the event description information into word vectors, and the feature vector of the emergency event is extracted from the word vectors through the long short-term memory layer.

[0064] Then, the extracted feature vector of the emergency event is input into the decoder for decoding. Specifically, the decoding is performed through the first fully connected layer in the decoder. Based on the first output layer of the decoder, the decoded feature vector is classified and predicted to obtain the multidimensional label of the emergency event.

[0065] Optionally, the input layer is also used to index the event description information of the emergency event. The input to the label generator is the indexed event description text, and the length of each index sequence is a preset length value `emergency_length`. In the encoder, the event description text is first transformed by the first word embedding layer, which converts each word in the event description text into a corresponding vector. The data dimension of the input to the first word embedding layer is represented as `emergency_vocab_size`. The output of the first embedding layer is set to convert words into spatial vectors of a preset dimension, such as 128 dimensions. If the input sequence length is `emergency_length`, then the output data of this layer is in the form of (None, emergency_length, 128). The role of the first word embedding layer is to perform vector mapping on the input description text, converting the index of each word in the description text into a fixed 128-dimensional vector. Then, the feature vector of the event description text of the emergency event is extracted through multiple LSTM layers. Each layer in the LSTM layer contains a first preset number of LSTM neurons, and the activation function of each layer is ReLU.

[0066] Furthermore, the first fully connected layer in the decoder contains a second preset number of neurons, and the activation function is also set to ReLU. The output layer also consists of a fully connected (Dense) layer, with the number of neurons being the same as the number of label dimensions to be generated for the multidimensional labels. The activation function is sigmoid. The output layer is used to output the label prediction results. For example, the first preset number is 64, and the second preset number is 128.

[0067] In the pre-training phase of the label generator, based on a preset number of iterations, historical emergency events and their multidimensional labels are used as training samples. The multidimensional labels generated by the label generator are compared with the real labels of historical emergency events. The error between the two is calculated based on the logarithmic loss function. With the goal of minimizing the error, the gradient descent optimization algorithm is used to adjust the parameters of the label generator, thereby optimizing the loss function and finding the optimal parameters that minimize the loss function.

[0068] In one embodiment, the planned generation model is built on a large language model. Specifically, the planned generation model includes a multi-label input layer, a second word embedding layer, a large language model layer, a second fully connected layer, a dropout layer, and a second output layer. The large language model layer can be built using a general large language model (LLM).

[0069] Based on this, step 300 may also include:

[0070] Step 301: Input the multidimensional labels and the event description information into the plan generation model. After the multidimensional labels and the event description information are concatenated by the multi-label input layer of the plan generation model, they are input into the second word embedding layer of the plan generation model.

[0071] Step 302: The concatenated multidimensional tags and event description information are converted into target word vectors through the second word embedding layer;

[0072] Step 303: The target word vector is processed by extracting features sequentially using the large language model layer, the second fully connected layer and the dropout layer of the planned generation model to obtain the target feature vector;

[0073] Step 304: Input the target feature vector into the second output layer of the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area; the emergency flight plan includes the flight time period, flight area and flight mission.

[0074] When generating emergency flight plans for low-altitude airspace within an emergency planning area using a plan generation model, the event description information and generated multidimensional labels are first input into the plan generation model. The multi-label input layer of the plan generation model concatenates the multidimensional labels and event description information, and then inputs the concatenated multidimensional labels and event description information into the second word embedding layer of the plan generation model. Next, the second word embedding layer transforms the concatenated multidimensional labels and event description information into target word vectors. The large language model layer, the second fully connected layer, and the dropout layer of the plan generation model are used to perform feature extraction processing on the target word vectors to obtain the target feature vectors. Finally, the target feature vectors are input into the second output layer of the plan generation model to generate the emergency flight plan for low-altitude airspace within the emergency planning area. This emergency flight plan includes the flight time period, flight area, and flight mission.

[0075] In one embodiment, refer to Figure 3The emergency flight plan generation process shown includes an emergency event extractor, an emergency event classifier, a three-part event filter, and an emergency event collector. Specifically, the emergency event extractor extracts event time tags from the multi-dimensional tags of emergency events; the emergency event classifier classifies emergency events based on event type tags in the multi-dimensional tags, extracting the event type; the three-part event filter filters out three-part events from the emergency events based on their tags; and the emergency event collector collects event description information for the emergency events.

[0076] Furthermore, the event time labels extracted by the emergency event time extractor, the event type labels extracted by the emergency event classifier, the three-part event labels extracted by the three-part event filter, and the time description information extracted by the emergency event collector are input into the second word embedding layer through the multi-label input layer for word mapping. The event time labels, event type labels, three-part event labels, and event description information are converted into target word vectors respectively. Then, the target word vectors are processed by feature extraction through the pre-trained large language model, the second fully connected layer, the dropout layer, and the output layer in sequence, and finally, the emergency flight plan for the low airspace within the emergency planning area is generated.

[0077] In another embodiment, refer to Figure 4 The emergency flight plan generation process, as shown, first predicts emergency events within a given emergency planning area, obtaining event description information. Then, based on a pre-trained tag generator, it generates event time tags, event type tags, and three-part event tags for the emergency events according to the event description information. Further, the generated event time tags, event type tags, three-part event tags, and time description information are input into the plan generation model to automatically generate an emergency flight plan for the low-altitude airspace within the emergency planning area. Finally, the generated emergency flight plan is sent to the low-altitude emergency control platform to request emergency flight permissions for the corresponding low-altitude airspace.

[0078] Furthermore, the generated emergency flight plan can be at different time granularities; therefore, step 300 may also include:

[0079] Step 310: Obtain the time granularity at which the emergency planning area generates the emergency flight plan;

[0080] Step 320: Input the multidimensional labels and the event description information into the plan generation model, and use the plan generation model to generate an emergency flight plan for the low airspace within the emergency planning area at the time granularity.

[0081] When using the plan generation model to generate emergency flight plans for low-altitude airspace within the emergency planning area, the time granularity for generating emergency flight plans in the emergency planning area is first obtained. Event description information and generated multidimensional labels are then input into the plan generation model, which is used to generate emergency flight plans for low-altitude airspace within the emergency planning area at the corresponding time granularity.

[0082] Optionally, the time granularity is configurable and can be monthly, quarterly, or annual. Based on the configured time granularity, an emergency flight plan is generated at the configured time granularity in the manner shown in steps 301 to 304 above.

[0083] Optionally, before generating an emergency flight plan using the plan generation model based on multi-dimensional labels and event description information of the emergency event, it is also necessary to pre-train the plan generation model based on historical emergency events and their corresponding historical emergency flight plans. Therefore, before step 300, the following may also be included:

[0084] Step 501: Obtain information on historical emergency events within the emergency planning area;

[0085] Step 502: Extract historical description information and multi-dimensional information of the historical emergency event from the information information; the multi-dimensional information includes event time, event type, and three-disruption event identifier; the three-disruption event identifier is used to characterize whether the historical emergency event is a three-disruption event;

[0086] Step 503: Obtain the historical emergency flight plan of the historical emergency event, use the multidimensional information as the multidimensional label of the historical description information, and use the historical emergency flight plan as the prediction result of the historical description information to construct a text dataset;

[0087] Step 504: Use the text dataset to iteratively train the preset plan generation model to obtain the pre-trained plan generation model.

[0088] First, information on historical emergency events within the emergency planning area is acquired, and key information from these events is extracted to obtain historical descriptive information and multidimensional information. The multidimensional information includes event time, event type, and a "three-disruption event" identifier. The "three-disruption event" identifier is specifically marked as "yes" or "no," indicating whether the historical emergency event was a "three-disruption event."

[0089] Furthermore, historical emergency flight plans corresponding to historical emergency events are obtained. The extracted multidimensional information is used as multidimensional labels for historical descriptive information, and historical emergency flight plans are used as prediction results for historical descriptive information to construct a text dataset. The constructed text dataset is then used to iteratively train a pre-trained plan generation model to obtain a pre-trained plan generation model.

[0090] In one embodiment, information on historical emergency events and low-altitude emergency flight plans for historical emergency events annotated by humans or experts are obtained from the low-altitude emergency management platform. Multidimensional information and descriptive information of historical emergency events are extracted from the information, and a total dataset of emergency flight plans for historical emergency events is generated.

[0091] Optionally, before constructing the sample dataset, preprocessing operations such as text cleaning and text serialization can be performed on the total dataset.

[0092] When constructing the sample dataset, the extracted multidimensional information is used as multidimensional labels for historical emergency events. Specifically, the event time of historical emergency events is used as the time label to form a set of event time labels for historical emergency events. ,in, This represents the word vector of the time term corresponding to the time tag of the Nth emergency event. Time terms can be words at different time granularities, such as monthly, quarterly, and annual. A set of event type tags for historical emergency events is formed using the event types of historical emergency events as event type tags. , This represents the word vector of the event type word corresponding to the event type label of the T-th emergency event. Event type words include, for example, typhoon, flood, and earthquake. Using the three-disruption event identifiers of historical emergency events as three-disruption event labels, a set of three-disruption event labels for historical emergency events is formed. , This represents the word vector of the three-part event identifier corresponding to the three-part event label of the Mth emergency event. The label result for each three-part event identifier includes either "yes" or "no". Furthermore, a descriptive text set is formed using descriptive information from historical emergency events. , This represents the word vector of the Wth word in the descriptive text of an emergency event. Historical emergency flight plans from past emergency events can form a plan set: , The word vector representing the Lth word in a historical emergency flight plan can be the flight time period, flight area, or flight mission. Based on multi-label datasets including event time label sets, event type label sets, and three-part event label sets from historical emergency events, combined with descriptive text sets of historical emergency events and historical emergency flight plan sets, a sample dataset is constructed to train the plan generation model.

[0093] Optionally, the sample dataset is cleaned and serialized, including unifying letter case, converting uppercase letters to lowercase letters, removing all punctuation marks, and then the text is segmented. At the same time, each word obtained from the segmentation is tokenized, so that each piece of text is converted into a piece of index number, and zeros are padded for sequences that do not reach the preset maximum text length.

[0094] The longest length of the multi-label dataset, Multi-tag_length, is taken as its uniform index sequence length, and the dictionary size of the multi-label dataset is set to Multi-tag_vocab_size; the longest length of the emergency event description text set, emergency_length, is taken as its uniform index sequence length, and the dictionary size of the emergency event description text set is set to emergency_vocab_size; the longest length of the low-altitude emergency flight plan set, plan_length, is taken as its uniform index sequence length, and the dictionary size of the low-altitude emergency flight plan set is set to plan_vocab_size.

[0095] The sample dataset is divided into a training set and a test set. For example, 90% of the sample dataset is allocated to the training set and 10% to the test set. The training set is used to train the plan generation model, and the test set is used to test the model performance of the trained plan generation model.

[0096] Optionally, the multidimensional labels for historical emergency events in the sample dataset can also be generated using a label generator. (See reference...) Figure 5 The model training process shown extracts historical description information of historical emergency events from historical emergency event information. Then, through a tag generator, it extracts the event time of historical emergency events from the historical description information to generate event time tags for historical emergency events, extracts the event type of historical emergency events from the historical description information to generate event type tags for historical emergency events, and extracts the three-part event identifier of historical emergency events from the historical description information to generate three-part event tags for historical emergency events.

[0097] Furthermore, text preprocessing is performed on the event time tags, event type tags, three-part event tags, and historical descriptive information of historical emergency events to construct a sample dataset. The constructed sample dataset is then used to train a pre-defined plan generation model, resulting in a pre-trained plan generation model. This model can then generate emergency flight plans online in real time based on predictions of emergency events within the emergency planning area.

[0098] Optionally, step 504 may further include:

[0099] Step 601: Based on a preset number of iterations, input the sample dataset into a preset plan generation model and iteratively train the plan generation model;

[0100] Step 602: In each iteration, a preset multi-class cross-entropy is used as the loss function, and a preset gradient descent optimization algorithm is adopted to adjust the model parameters of the planned generation model in order to optimize the function value of the loss function. When the function value of the loss function is optimal, the pre-trained planned generation model is obtained.

[0101] Based on a preset number of iterations, such as 1500 (epochs=1500), the constructed sample dataset is input into a preset plan generation model for iterative training. In each iteration, a preset multi-class cross-entropy loss function is used, along with a preset gradient descent optimization algorithm, to adjust the model parameters of the plan generation model to optimize the loss function value. When the loss function reaches its optimal value, the pre-trained plan generation model is obtained.

[0102] Optionally, based on the general large language model, a multi-label input layer, word embedding layer, fully connected layer, dropout layer, and output layer can be added to construct a planned generation model, thereby improving the performance of the large language model in specific task scenarios. During model training, the weights of the large language model are used as the initial weights for model training. Only the fully connected layer, dropout layer, and output layer in the planned generation model are trained. That is, in each iteration, a preset multi-class cross-entropy is used as the loss function, and a preset gradient descent optimization algorithm is employed to adjust the weights of the fully connected layer, dropout layer, and output layer in the planned generation model to optimize the value of the loss function. The gradient descent optimization algorithm is used to improve the learning speed of traditional gradient descent. Through gradient descent, the model can find the optimal weight values ​​that minimize the loss function.

[0103] In this embodiment, the plan generation model can generate emergency flight plans for low-altitude airspace online in real time based on the prediction results of emergency events in the emergency planning area. This is accurate and efficient, and can meet the needs of frequent applications for low-altitude emergency airspace.

[0104] Furthermore, the planned generation model, based on the large language model, adds a multi-label input layer, word embedding layer, fully connected layer, dropout layer, and output layer, enabling the large language model to better adapt to tasks in special scenarios. It can more accurately and efficiently generate emergency flight plans for low-altitude emergency airspace, including flight time periods, flight areas, and flight missions, thus better ensuring the smooth implementation of emergency rescue missions.

[0105] The emergency flight plan generation apparatus provided in the embodiments of this application is described below. The emergency flight plan generation apparatus described below and the emergency flight plan generation method described above can be referred to in correspondence.

[0106] Reference Figure 6 The emergency flight plan generation device provided in this application includes:

[0107] The event prediction module 10 is used to predict emergency events within the emergency planning area and obtain event description information of the emergency events.

[0108] Tag generation module 20 is used to generate multi-dimensional tags for the emergency event based on the event description information; the multi-dimensional tags include at least an event time tag, an event type tag, and a three-part event tag;

[0109] The plan generation module 30 is used to input the multidimensional labels and the event description information into the plan generation model, and use the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area;

[0110] The plan generation model is obtained through iterative training based on a sample dataset, which is constructed based on historical emergency events within the emergency planning area and the historical flight plans corresponding to those historical emergency events.

[0111] In one embodiment, the label generation module 20 is further configured to:

[0112] The event description information is input into the encoder of the tag generator; the encoder includes a first word embedding layer and a long short-term memory layer, and the tag generator also includes a decoder, which includes a first fully connected layer and a first output layer.

[0113] The event description information is converted into word vectors using the first word embedding layer in the encoder, and feature vectors are extracted from the word vectors using the long short-term memory layer.

[0114] The feature vector is input into the first fully connected layer of the decoder for decoding processing, and the decoded feature vector is classified and predicted based on the first output layer of the decoder to obtain the multidimensional label of the emergency event.

[0115] In one embodiment, the plan generation model includes a multi-label input layer, a second word embedding layer, a large language model layer, a second fully connected layer, a dropout layer, and a second output layer; the plan generation module 30 is further configured to:

[0116] The multidimensional labels and the event description information are input into the plan generation model. The multidimensional labels and the event description information are then concatenated using the multi-label input layer of the plan generation model and input into the second word embedding layer of the plan generation model.

[0117] The second word embedding layer converts the concatenated multidimensional tags and event description information into target word vectors.

[0118] The target word vector is processed by sequentially extracting features from the large language model layer, the second fully connected layer, and the dropout layer of the aforementioned plan generation model to obtain the target feature vector.

[0119] The target feature vector is input into the second output layer of the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area; the emergency flight plan includes the flight time period, flight area, and flight mission.

[0120] In one embodiment, the plan generation module 30 is further configured to:

[0121] Obtain the time granularity at which the emergency planning area generates the emergency flight plan;

[0122] The multidimensional labels and the event description information are input into the plan generation model, and the plan generation model is used to generate an emergency flight plan for the low airspace within the emergency planning area at the time granularity.

[0123] In one embodiment, the emergency flight plan generation device further includes a model pre-training module, used for:

[0124] Obtain information on historical emergency events within the emergency planning area;

[0125] Historical description information and multi-dimensional information of the historical emergency events are extracted from the information; the multi-dimensional information includes event time, event type, and three-disruption event identifier; the three-disruption event identifier is used to characterize whether the historical emergency event is a three-disruption event.

[0126] Obtain the historical emergency flight plans of the historical emergency events, use the multidimensional information as the multidimensional label of the historical descriptive information, and use the historical emergency flight plans as the prediction results of the historical descriptive information to construct a sample dataset;

[0127] The preset plan generation model is iteratively trained using the sample dataset to obtain a pre-trained plan generation model.

[0128] In one embodiment, the model pre-training module is further configured to:

[0129] Based on a preset number of iterations, the sample dataset is input into a preset plan generation model, and the plan generation model is iteratively trained.

[0130] In each iteration, a pre-defined multi-class cross-entropy is used as the loss function, and a pre-defined gradient descent optimization algorithm is employed to adjust the model parameters of the planned generation model in order to optimize the function value of the loss function. When the function value of the loss function is optimal, a pre-trained planned generation model is obtained.

[0131] Figure 7 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 7 As shown, the electronic device may include: a processor 710, a communications interface 720, a memory 730, and a communication bus 740, wherein the processor 710, the communications interface 720, and the memory 730 communicate with each other via the communication bus 740. The processor 710 can call logical instructions in the memory 730 to execute an emergency flight plan generation method, which includes:

[0132] Predict emergency events within the emergency planning area to obtain event description information of the emergency events;

[0133] A multidimensional tag for the emergency event is generated based on the event description information; the multidimensional tag includes at least an event time tag, an event type tag, and a three-part event tag.

[0134] The multidimensional labels and the event description information are input into the plan generation model, and the plan generation model is used to generate an emergency flight plan for the low-altitude airspace within the emergency planning area.

[0135] The plan generation model is obtained through iterative training based on a sample dataset, which is constructed based on historical emergency events within the emergency planning area and the historical flight plans corresponding to those historical emergency events.

[0136] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0137] On the other hand, embodiments of this application also provide a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the emergency flight plan generation method provided by the above methods, the method including:

[0138] Predict emergency events within the emergency planning area to obtain event description information of the emergency events;

[0139] A multidimensional tag for the emergency event is generated based on the event description information; the multidimensional tag includes at least an event time tag, an event type tag, and a three-part event tag.

[0140] The multidimensional labels and the event description information are input into the plan generation model, and the plan generation model is used to generate an emergency flight plan for the low-altitude airspace within the emergency planning area.

[0141] The plan generation model is obtained through iterative training based on a sample dataset, which is constructed based on historical emergency events within the emergency planning area and the historical flight plans corresponding to those historical emergency events.

[0142] In another aspect, embodiments of this application also provide a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the emergency flight plan generation method provided by the methods described above, the method comprising:

[0143] Predict emergency events within the emergency planning area to obtain event description information of the emergency events;

[0144] A multidimensional tag for the emergency event is generated based on the event description information; the multidimensional tag includes at least an event time tag, an event type tag, and a three-part event tag.

[0145] The multidimensional labels and the event description information are input into the plan generation model, and the plan generation model is used to generate an emergency flight plan for the low-altitude airspace within the emergency planning area.

[0146] The plan generation model is obtained through iterative training based on a sample dataset, which is constructed based on historical emergency events within the emergency planning area and the historical flight plans corresponding to those historical emergency events.

[0147] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0148] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0149] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for generating emergency flight plans, characterized in that, include: Predict emergency events within the emergency planning area to obtain event description information of the emergency events; Generate a multidimensional tag for the emergency event based on the event description information; The multidimensional tags include at least event time tags, event type tags, and three-part event tags; The multidimensional labels and the event description information are input into the plan generation model, and the plan generation model is used to generate an emergency flight plan for the low-altitude airspace within the emergency planning area. The plan generation model is obtained through iterative training based on a sample dataset, which is constructed based on historical emergency events within the emergency planning area and the corresponding historical flight plans. The plan generation model includes a multi-label input layer, a second word embedding layer, a large language model layer, a second fully connected layer, a dropout layer, and a second output layer. Before inputting the multidimensional labels and the event description information into the plan generation model, and using the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area, the method further includes: Obtain information on historical emergency events within the emergency planning area; Historical description information and multidimensional information of the historical emergency events are extracted from the information. The multidimensional information includes event time, event type, and three-disruption event identifier. The three-disruption event identifier is used to indicate whether the historical emergency event is a three-disruption event. The three-disruption event is an event involving a road break, power outage, or network outage. Obtain the historical emergency flight plans of the historical emergency events, use the multidimensional information as the multidimensional label of the historical descriptive information, and use the historical emergency flight plans as the prediction results of the historical descriptive information to construct a sample dataset; The preset plan generation model is iteratively trained using the sample dataset to obtain a pre-trained plan generation model.

2. The emergency flight plan generation method according to claim 1, characterized in that, The step of generating a multidimensional tag for the emergency event based on the event description information includes: The event description information is input into the encoder of the tag generator; the encoder includes a first word embedding layer and a long short-term memory layer, and the tag generator also includes a decoder, which includes a first fully connected layer and a first output layer. The event description information is converted into word vectors using the first word embedding layer in the encoder, and feature vectors are extracted from the word vectors using the long short-term memory layer. The feature vector is input into the first fully connected layer of the decoder for decoding processing, and the decoded feature vector is classified and predicted based on the first output layer of the decoder to obtain the multidimensional label of the emergency event.

3. The emergency flight plan generation method according to claim 1, characterized in that, The step of inputting the multidimensional labels and the event description information into the plan generation model, and using the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area, includes: The multidimensional labels and the event description information are input into the plan generation model. The multidimensional labels and the event description information are then concatenated using the multi-label input layer of the plan generation model and input into the second word embedding layer of the plan generation model. The second word embedding layer converts the concatenated multidimensional tags and event description information into target word vectors. The target word vector is processed by sequentially extracting features from the large language model layer, the second fully connected layer, and the dropout layer of the aforementioned plan generation model to obtain the target feature vector. The target feature vector is input into the second output layer of the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area; the emergency flight plan includes the flight time period, flight area, and flight mission.

4. The emergency flight plan generation method according to claim 1, characterized in that, The step of inputting the multidimensional labels and the event description information into the plan generation model, and using the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area, includes: Obtain the time granularity at which the emergency planning area generates the emergency flight plan; The multidimensional labels and the event description information are input into the plan generation model, and the plan generation model is used to generate an emergency flight plan for the low airspace within the emergency planning area at the time granularity.

5. The emergency flight plan generation method according to claim 1, characterized in that, The step of iteratively training a pre-defined plan generation model using the sample dataset to obtain a pre-trained plan generation model includes: Based on a preset number of iterations, the sample dataset is input into a preset plan generation model, and the plan generation model is iteratively trained. In each iteration, a pre-defined multi-class cross-entropy is used as the loss function, and a pre-defined gradient descent optimization algorithm is employed to adjust the model parameters of the planned generation model in order to optimize the function value of the loss function. When the function value of the loss function is optimal, a pre-trained planned generation model is obtained.

6. An emergency flight plan generation device, characterized in that, include: The event prediction module is used to predict emergency events within the emergency planning area and obtain event description information of the emergency events. The tag generation module is used to generate multidimensional tags for the emergency event based on the event description information; The multidimensional tags include at least event time tags, event type tags, and three-part event tags; The plan generation module is used to input the multidimensional labels and the event description information into the plan generation model, and use the plan generation model to generate an emergency flight plan for the low-altitude airspace within the emergency planning area; The plan generation model is obtained through iterative training based on a sample dataset, which is constructed based on historical emergency events within the emergency planning area and the corresponding historical flight plans. The plan generation model includes a multi-label input layer, a second word embedding layer, a large language model layer, a second fully connected layer, a dropout layer, and a second output layer. The model pre-training module is used to acquire information on historical emergency events within the emergency planning area; and to extract historical description information and multi-dimensional information of the historical emergency events from the information. The multidimensional information includes event time, event type, and a three-disruption event identifier; the three-disruption event identifier is used to characterize whether the historical emergency event is a three-disruption event; the three-disruption event is an event involving a circuit breaker, a power outage, and a network outage; Obtain the historical emergency flight plans of the historical emergency events, use the multidimensional information as the multidimensional label of the historical descriptive information, and use the historical emergency flight plans as the prediction results of the historical descriptive information to construct a sample dataset; The preset plan generation model is iteratively trained using the sample dataset to obtain a pre-trained plan generation model.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the emergency flight plan generation method as described in any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the emergency flight plan generation method as described in any one of claims 1 to 5.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the emergency flight plan generation method as described in any one of claims 1 to 5.

Images