Remote sensing control method and system based on natural language model

Abstract

The invention provides a remote sensing control method and system based on a natural language model, and the method comprises the steps: obtaining a user request, recognizing the user request according to a pre-trained natural language model, and obtaining a user intention vector representation according to a recognition result; screening out a target remote sensing task vector closest to the user intention vector from a preset remote sensing task vector library, obtaining a remote sensing task corresponding to the target remote sensing task vector according to the target remote sensing task vector, obtaining a task type of the remote sensing task, and selecting a corresponding remote sensing algorithm based on the task type; and generating a control parameter set according to the remote sensing algorithm and the state parameter of the current remote sensing equipment, and sending the control parameter set to a control system of the remote sensing equipment within a first preset time, so that the control system performs remote sensing control on the remote sensing equipment according to the control parameter set. The accuracy and efficiency of remote sensing control can be improved.

Description

Technical Field

[0001] The present invention relates to the field of remote sensing control technology, and in particular to a remote sensing control method and system based on a natural language model. Background Art

[0002] Remote sensing technology, as a crucial means of acquiring geospatial information, plays an irreplaceable role in numerous fields, including resource surveys, environmental monitoring, disaster warning, and urban planning. With the continuous advancement of technology, remote sensing equipment is becoming increasingly powerful, data acquisition capabilities are significantly enhanced, and the application scenarios of remote sensing data are expanding. Remote sensing control serves as a bridge between user needs and remote sensing equipment execution. Its performance and intelligence directly impact the efficiency and quality of remote sensing missions.

[0003] Currently, existing remote sensing control technologies have numerous limitations. First, operational complexity is high, requiring users to be familiar with specific command formats and operating specifications. This requires specialized remote sensing knowledge and technical background, placing a high learning curve for non-experts and limiting the widespread adoption and application of remote sensing technology. Second, traditional methods struggle to accurately understand the diverse natural language requirements of users. Users cannot directly describe tasks in natural language but must translate them into specific instructions, increasing workload and easily leading to inaccurate task execution due to misunderstandings. Furthermore, traditional remote sensing control systems lack flexibility. Their fixed, pre-set command sets and simple logical judgment rules are ill-suited to complex and ever-changing real-world requirements. When users propose new tasks, the system responds slowly and requires reconfiguration, resulting in inefficient execution. Furthermore, the system lacks self-optimization capabilities, unable to adjust control strategies based on actual performance and user feedback, making it difficult to continuously improve the accuracy and efficiency of remote sensing control. Summary of the Invention

[0004] The purpose of the present invention is to provide a remote sensing control method and system based on a natural language model, aiming to solve the problems of low control accuracy and efficiency existing in traditional remote sensing control technology.

[0005] In a first aspect, the present invention provides a remote sensing control method based on a natural language model, the method comprising:

[0006] Obtaining a user request, recognizing the user request based on a pre-trained natural language model, and obtaining a user intent vector representation based on the recognition result;

[0007] Filtering a target remote sensing task vector that is closest to the user intention vector from a preset remote sensing task vector library, obtaining a remote sensing task corresponding to the target remote sensing task vector, obtaining a task type of the remote sensing task, and selecting a corresponding remote sensing algorithm based on the task type;

[0008] A control parameter set is generated according to the remote sensing algorithm and the state parameters of the current remote sensing device, and the control parameter set is sent to the control system of the remote sensing device within a first preset time, so that the control system performs remote sensing control on the remote sensing device according to the control parameter set.

[0009] Furthermore, the steps of obtaining a user request, identifying the user request according to a pre-trained natural language model, and obtaining a user intent vector representation according to the identification result include:

[0010] The natural language model is a deep learning model based on the Transformer architecture, which obtains natural language text data related to remote sensing tasks and preprocesses the natural language text data, including denoising, word segmentation, and part-of-speech tagging;

[0011] Inputting the preprocessed natural language text data into the natural language model for training to obtain a pre-trained natural language model;

[0012] The expression of the user intention vector is:

[0013] I=(i1,i2,=,i m );

[0014] Among them, I is the user intention vector, i1, i2, i m They are the values ​​of the 1st, 2nd, and mth dimensions respectively, where m is the vector dimension.

[0015] Furthermore, the step of inputting the pre-processed natural language text data into a natural language model for training to obtain a pre-trained natural language model includes:

[0016] Randomly mask the words in the preprocessed natural language text data and dynamically adjust the mask ratio during the natural language model training process:

[0017]

[0018] in, is the target loss function value to be achieved during training, L k is the loss function value of the kth iteration training, α k , α k+1 are the mask ratios for the kth and k+1th iteration training respectively. The value range of the mask ratio is (0, 1), and β is the adjustment step size.

[0019] Furthermore, the step of selecting the target remote sensing task vector closest to the user intention vector representation from the preset remote sensing task vector library includes:

[0020] The user intention vector representation and all remote sensing task vectors in the preset remote sensing task vector library are weighted respectively:

[0021]

[0022] Among them, I' represents the weighted user intention vector, W1 and W2 are weight vectors, S i is the i-th remote sensing mission vector in the preset remote sensing mission vector library, S' i is the weighted i-th remote sensing mission vector;

[0023] The similarity between the weighted user intention vector representation and the weighted i-th remote sensing task vector is calculated according to the following formula:

[0024]

[0025] Among them, Q i is the similarity between the weighted user intention vector and the weighted i-th remote sensing task vector;

[0026] The maximum similarity is screened out from all similarities, and the remote sensing task vector corresponding to the maximum similarity is obtained as the target remote sensing task vector.

[0027] Furthermore, the step of generating a control parameter set according to the remote sensing algorithm and the state parameters of the current remote sensing device includes:

[0028] Assume that the current state parameter set of the remote sensing device is

[0029] Computational model:

[0030] p i =f Ai (S);

[0031] Among them, p i represents the i-th control parameter in the control parameter set, f Ai represents the mapping function;

[0032] The functional relationship of the mapping function is determined according to the task type, and the functional relationship includes linear relationship and nonlinear relationship:

[0033]

[0034] Among them, α i , β i , γ i are the weight coefficients of the i-th control parameter, and C is a constant term;

[0035] The constraint conditions of the control parameter calculation model are constructed according to the following formula:

[0036]

[0037] in, are the upper and lower limits of the i-th control parameter, Δp min ,Δp max are the upper and lower limit constraints between multiple control parameters, p i+1 、p i+j are the i+1th and i+jth control parameters respectively;

[0038] Solve the control parameter calculation model and constraint conditions, and obtain the control parameter set P = (p1, p2, ..., p M ), p1, p2, p M They are the 1st, 2nd and Mth control parameters respectively.

[0039] Furthermore, after the step of sending the control parameter set to the control system of the remote sensing device within the first preset time so that the control system performs remote sensing control on the remote sensing device according to the control parameter set, the method further includes:

[0040] Obtaining user feedback results after executing remote sensing control, and recognizing the user feedback results according to a pre-trained natural language model to obtain a feedback vector;

[0041] The target remote sensing mission vector is adjusted according to the following formula:

[0042] S'0=(1-σ)S0+σl0;

[0043] Where S'0 represents the adjusted target remote sensing mission vector, σ is the adjustment coefficient, and I0 is the feedback vector;

[0044] The preset remote sensing mission vector library is updated according to the adjusted target remote sensing mission vector.

[0045] In a second aspect, the present invention provides a remote sensing control system based on a natural language model, the system comprising:

[0046] A user request acquisition module is used to acquire user requests, identify the user requests according to a pre-trained natural language model, and obtain a user intent vector representation based on the recognition result;

[0047] A vector matching module is used to filter out a target remote sensing task vector that is closest to the user intention vector from a preset remote sensing task vector library, obtain the remote sensing task corresponding to the target remote sensing task vector, obtain the task type of the remote sensing task, and select the corresponding remote sensing algorithm based on the task type;

[0048] A control parameter generation module is used to generate a control parameter set based on the remote sensing algorithm and the state parameters of the current remote sensing device, and send the control parameter set to the control system of the remote sensing device within a first preset time, so that the control system can remotely control the remote sensing device according to the control parameter set.

[0049] In a third aspect, the present invention provides a storage medium storing one or more programs, which, when executed by a processor, implement the above-mentioned remote sensing control method based on a natural language model.

[0050] In a fourth aspect, the present invention provides an electronic device, comprising a memory and a processor, wherein:

[0051] The memory is used to store computer programs;

[0052] When the processor is used to execute the computer program stored in the memory, the above-mentioned remote sensing control method based on the natural language model is implemented.

[0053] Compared with the prior art, the present invention has the following advantages:

[0054] This invention utilizes natural language processing technology, vector similarity calculation, control parameter generation, and feedback optimization to achieve an intelligent process from user natural language requests to precise control of remote sensing equipment. This method significantly reduces user operational difficulty, improves the system's usability and intelligence, and enables rapid and accurate response to user needs and execution of diverse remote sensing tasks. Furthermore, by continuously optimizing the remote sensing task vector library based on user feedback, the system possesses self-learning and optimization capabilities, continuously improving the effectiveness and accuracy of remote sensing control, providing an efficient, convenient, and intelligent control solution for the remote sensing field. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a flow chart of a remote sensing control method based on a natural language model proposed in one embodiment of the present invention;

[0056] Figure 2 This is a schematic structural diagram of a remote sensing control system based on a natural language model proposed in one embodiment of the present invention.

[0057] The following specific embodiments will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. Unless otherwise defined, the technical terms or scientific terms used herein should be the common meanings understood by people with ordinary skills in the field to which the invention belongs. The words "including" and similar words used in this article mean that the elements or objects appearing before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects.

[0059] like Figure 1 As shown, an embodiment of the present invention provides a remote sensing control method based on a natural language model, the method comprising steps S101 to S103, wherein:

[0060] Step S101: obtaining a user request, identifying the user request according to a pre-trained natural language model, and obtaining a user intent vector representation according to the identification result;

[0061] It should be noted that the natural language model is a deep learning model based on the Transformer architecture. It obtains natural language text data related to remote sensing tasks and preprocesses the natural language text data, including denoising, word segmentation, and part-of-speech tagging. The preprocessed natural language text data is then input into the natural language model for training to obtain a pre-trained natural language model.

[0062] The expression of the user intention vector is:

[0063] Ⅰ(i1,i2,…,i m );

[0064] Among them, I is the user intention vector, i1, i2, i m They are the values ​​of the 1st, 2nd, and mth dimensions respectively, where m is the vector dimension.

[0065] By performing preprocessing operations such as denoising, word segmentation, and part-of-speech tagging on natural language text data related to remote sensing tasks, denoising can remove irrelevant information in the data, and word segmentation and part-of-speech tagging can help the model better understand the grammatical structure and semantic information of the text.

[0066] In addition, during the training of the natural language model, the words in the preprocessed natural language text data are randomly masked, and the mask ratio is dynamically adjusted. Dynamically adjusting the mask ratio allows the model to better learn the semantic information of the text, improve the model's generalization ability and robustness, and further improve the natural language model's recognition accuracy of user requests, thereby more accurately capturing user intent. Specifically, the words in the preprocessed natural language text data are randomly masked, and the mask ratio is dynamically adjusted during the natural language model training process:

[0067]

[0068] in, is the target loss function value to be achieved during training, L k is the loss function value of the kth iteration training, α k , α k+1 are the mask ratios for the kth and k+1th iteration training respectively. The value range of the mask ratio is (0, 1), and β is the adjustment step size.

[0069] Step S102: Filtering a target remote sensing task vector that is closest to the user intention vector from a preset remote sensing task vector library, obtaining a remote sensing task corresponding to the target remote sensing task vector, obtaining a task type of the remote sensing task, and selecting a corresponding remote sensing algorithm based on the task type;

[0070] It should be noted that the preset remote sensing task vector library contains a large number of remote sensing task vectors related to remote sensing tasks. In order to filter out the target remote sensing task vector that completely corresponds to the user request, the user intention vector representation and all remote sensing task vectors in the preset remote sensing task vector library are first weighted:

[0071]

[0072] Among them, I' represents the weighted user intention vector, W1 and W2 are weight vectors, S i is the i-th remote sensing mission vector in the preset remote sensing mission vector library, S' i is the weighted i-th remote sensing mission vector;

[0073] Then, the similarity between the weighted user intention vector representation and the weighted i-th remote sensing task vector is calculated according to the following formula:

[0074]

[0075] Among them, Q i is the similarity between the weighted user intention vector and the weighted i-th remote sensing task vector;

[0076] Finally, the maximum similarity is screened out from all the similarities, and the remote sensing task vector corresponding to the maximum similarity is obtained as the target remote sensing task vector.

[0077] In summary, weighted processing can highlight the importance of different dimensions in the vector and improve the accuracy of similarity calculation, thereby more accurately screening out the remote sensing task vector that best matches the user's intention and ensuring the accurate execution of the remote sensing task.

[0078] Step S103: Generate a control parameter set based on the remote sensing algorithm and the current state parameters of the remote sensing device, and send the control parameter set to the control system of the remote sensing device within a first preset time, so that the control system performs remote sensing control on the remote sensing device according to the control parameter set.

[0079] It should be noted that, in this step, it is assumed that the state parameter set of the current remote sensing device is S = (S1, S2, ..., S n ), build a control parameter calculation model based on the state parameter set:

[0080] p i =f Ai (S);

[0081] Among them, p i represents the i-th control parameter in the control parameter set, f Ai represents the mapping function;

[0082] The functional relationship of the mapping function is determined according to the task type, and the functional relationship includes linear relationship and nonlinear relationship:

[0083]

[0084] Among them, α i , β i , γ i are the weight coefficients of the i-th control parameter, and C is a constant term;

[0085] The constraint conditions of the control parameter calculation model are constructed according to the following formula:

[0086]

[0087] in, are the upper and lower limits of the i-th control parameter, Δp min ,Δp max are the upper and lower limit constraints between multiple control parameters, p i+1 、p i+j are the i+1th and i+jth control parameters respectively;

[0088] Solve the control parameter calculation model and constraint conditions, and obtain the control parameter set P = (p1, p2, ..., p M ), p1, p2, p M They are the 1st, 2nd and Mth control parameters respectively.

[0089] In summary, by constructing a control parameter calculation model and constraint conditions, we can comprehensively consider the remote sensing algorithm and equipment state parameters, flexibly generate appropriate control parameters according to different task types, ensure that the remote sensing equipment can be effectively controlled in different states, and improve the quality of remote sensing data and the accuracy of task execution.

[0090] In addition, in some embodiments, in order to further improve the matching accuracy of user requests, after the remote sensing control is completed, the user feedback result after the remote sensing control is executed is obtained, and the user feedback result is recognized according to the pre-trained natural language model to obtain a feedback vector;

[0091] The target remote sensing mission vector is adjusted according to the following formula:

[0092] S'0=(1-σ)S0+σl0;

[0093] Where S'0 represents the adjusted target remote sensing mission vector, σ is the adjustment coefficient, and I0 is the feedback vector;

[0094] Finally, the preset remote sensing mission vector library is updated according to the adjusted target remote sensing mission vector, that is, the original target remote sensing mission vector is replaced by the adjusted target remote sensing mission vector.

[0095] By continuously optimizing the remote sensing mission vector library based on actual user feedback, the accuracy of subsequent remote sensing mission screening and the effect of remote sensing control are improved, and the system's self-learning and optimization are achieved to better meet user needs.

[0096] For example, suppose a user is an environmental monitor who wants to obtain remote sensing images of vegetation recovery following a recent forest fire in a certain region. The user sends a natural language request to the system: "Get remote sensing images of vegetation recovery following a forest fire in region XX between July 1st and August 1st." The system first receives this request and uses a pre-trained natural language model to recognize it. The natural language model analyzes keywords in the request, such as "region XX," "July 1st to August 1st," "forest fire," and "vegetation recovery," and uses this information to generate a vector representation of the user's intent. Next, the system selects the target remote sensing task vector that most closely matches the user's intent vector from a pre-set remote sensing task vector library. This library may contain a variety of remote sensing tasks, such as "Get images of terrain and landforms in a certain region" and "Monitor crop growth in a certain region." By calculating vector similarity, the system finds the task that best matches the user's request: the vector corresponding to "Get remote sensing images of vegetation recovery following a forest fire in a certain region during a specific time period." Based on this target remote sensing task vector, the system obtains the corresponding remote sensing task and determines the task type as "vegetation restoration monitoring". Based on this task type, the system selects a suitable remote sensing algorithm, such as a multispectral image analysis algorithm, which can evaluate the growth status of vegetation by analyzing the reflectivity of different bands. Then, the system generates a set of control parameters based on the remote sensing algorithm and the status parameters of the current remote sensing equipment (such as the resolution, orbit parameters, storage capacity, etc. of the equipment). For example, the control parameter set may include the time, resolution, shooting angle, etc. of the image shooting.

[0097] This invention utilizes natural language processing technology, vector similarity calculation, control parameter generation, and feedback optimization to achieve an intelligent process from user natural language requests to precise control of remote sensing equipment. This method significantly reduces user operational difficulty, improves the system's usability and intelligence, and enables rapid and accurate response to user needs and execution of diverse remote sensing tasks. Furthermore, by continuously optimizing the remote sensing task vector library based on user feedback, the system possesses self-learning and optimization capabilities, continuously improving the effectiveness and accuracy of remote sensing control, providing an efficient, convenient, and intelligent control solution for the remote sensing field.

[0098] like Figure 2 As shown, an embodiment of the present invention provides a remote sensing control system based on a natural language model, the system comprising:

[0099] A user request acquisition module 10 is used to acquire a user request, identify the user request according to a pre-trained natural language model, and obtain a user intent vector representation based on the recognition result;

[0100] The vector matching module 20 is used to select a target remote sensing task vector that is closest to the user intention vector from a preset remote sensing task vector library, obtain a remote sensing task corresponding to the target remote sensing task vector, obtain a task type of the remote sensing task, and select a corresponding remote sensing algorithm based on the task type;

[0101] The control parameter generation module 30 is used to generate a control parameter set based on the remote sensing algorithm and the state parameters of the current remote sensing device, and send the control parameter set to the control system of the remote sensing device within a first preset time, so that the control system can remotely control the remote sensing device according to the control parameter set.

[0102] On the other hand, the present invention further provides a storage medium storing one or more programs, which, when executed by a processor, implement the above-mentioned remote sensing control method based on a natural language model.

[0103] On the other hand, the present invention also proposes an electronic device, including a memory and a processor, wherein the memory is used to store computer programs, and the processor is used to execute the computer programs stored in the memory to implement the above-mentioned remote sensing control method based on the natural language model.

[0104] Those skilled in the art will appreciate that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device), or in conjunction with such instruction execution system, apparatus, or device. For purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by an instruction execution system, apparatus, or device, or in conjunction with such instruction execution system, apparatus, or device.

[0105] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0106] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement the hardware: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0107] While the embodiments of the present invention have been described in detail above, it will be apparent to those skilled in the art that various modifications and variations of these embodiments are possible. However, it should be understood that such modifications and variations are within the scope and spirit of the present invention as set forth in the claims. Furthermore, the invention described herein is susceptible to other embodiments and may be practiced or implemented in a variety of ways.

Claims

1. A remote sensing control method based on a natural language model, characterized in that: The method comprises: Obtaining a user request, recognizing the user request based on a pre-trained natural language model, and obtaining a user intent vector representation based on the recognition result; Filtering a target remote sensing task vector that is closest to the user intention vector from a preset remote sensing task vector library, obtaining a remote sensing task corresponding to the target remote sensing task vector, obtaining a task type of the remote sensing task, and selecting a corresponding remote sensing algorithm based on the task type; A control parameter set is generated according to the remote sensing algorithm and the state parameters of the current remote sensing device, and the control parameter set is sent to the control system of the remote sensing device within a first preset time, so that the control system performs remote sensing control on the remote sensing device according to the control parameter set.

2. The remote sensing control method based on the natural language model according to claim 1, characterized in that: The steps of obtaining a user request, identifying the user request according to a pre-trained natural language model, and obtaining a user intent vector representation according to the identification result include: The natural language model is a deep learning model based on the Transformer architecture, which obtains natural language text data related to remote sensing tasks and preprocesses the natural language text data, including denoising, word segmentation, and part-of-speech tagging; Inputting the preprocessed natural language text data into the natural language model for training to obtain a pre-trained natural language model; The expression of the user intention vector is: I=(i1,i2,…,i m ); Among them, I is the user intention vector, i1, i2, i m They are the values ​​of the 1st, 2nd, and mth dimensions respectively, where m is the vector dimension.

3. The remote sensing control method based on the natural language model according to claim 2, characterized in that: The step of inputting the pre-processed natural language text data into the natural language model for training to obtain the pre-trained natural language model includes: Randomly mask the words in the preprocessed natural language text data and dynamically adjust the mask ratio during the natural language model training process: in, is the target loss function value to be achieved during training, L k is the loss function value of the kth iteration training, α k , α k+1 are the mask ratios for the kth and k+1th iteration training respectively. The value range of the mask ratio is (0, 1), and β is the adjustment step size.

4. The remote sensing control method based on the natural language model according to claim 3, characterized in that: The step of selecting the target remote sensing task vector closest to the user intention vector representation from the preset remote sensing task vector library includes: The user intention vector representation and all remote sensing task vectors in the preset remote sensing task vector library are weighted respectively: Among them, I' represents the weighted user intention vector, W1 and W2 are weight vectors, S i is the i-th remote sensing mission vector in the preset remote sensing mission vector library, S' i is the weighted i-th remote sensing mission vector; The similarity between the weighted user intention vector representation and the weighted i-th remote sensing task vector is calculated according to the following formula: Among them, Q i is the similarity between the weighted user intention vector and the weighted i-th remote sensing task vector; The maximum similarity is screened out from all similarities, and the remote sensing task vector corresponding to the maximum similarity is obtained as the target remote sensing task vector.

5. The remote sensing control method based on the natural language model according to claim 4, characterized in that: The step of generating a control parameter set according to the remote sensing algorithm and the state parameters of the current remote sensing device includes: Assume that the state parameter set of the current remote sensing equipment is S=(S1, S2, ..., S n ), build a control parameter calculation model based on the state parameter set: p i =f A( S); Among them, p i represents the i-th control parameter in the control parameter set, f Ai represents the mapping function; The functional relationship of the mapping function is determined according to the task type, and the functional relationship includes linear relationship and nonlinear relationship: Among them, α i , β i , γ i are the weight coefficients of the i-th control parameter, and C is a constant term; The constraint conditions of the control parameter calculation model are constructed according to the following formula: in, are the upper and lower limits of the i-th control parameter, Δp min ,Δp max are the upper and lower limit constraints between multiple control parameters, p i+1 、p i+j are the i+1th and i+jth control parameters respectively; Solve the control parameter calculation model and constraint conditions, and obtain the control parameter set P = (p1, p2, ..., px) according to the solution results, where p1, p2, p M They are the 1st, 2nd and Mth control parameters respectively.

6. The remote sensing control method based on the natural language model according to claim 5, characterized in that: After the step of sending the control parameter set to the control system of the remote sensing device within the first preset time so that the control system performs remote sensing control on the remote sensing device according to the control parameter set, the step further includes: Obtaining user feedback results after executing remote sensing control, and recognizing the user feedback results according to a pre-trained natural language model to obtain a feedback vector; The target remote sensing mission vector is adjusted according to the following formula: S'0=(1-σ)S0+σI0; Where S'0 represents the adjusted target remote sensing mission vector, σ is the adjustment coefficient, and I0 is the feedback vector; The preset remote sensing mission vector library is updated according to the adjusted target remote sensing mission vector.

7. A remote sensing control system based on a natural language model, characterized in that: The system comprises: A user request acquisition module is used to acquire user requests, identify the user requests based on a pre-trained natural language model, and obtain a user intent vector representation based on the recognition result; A vector matching module is used to filter out a target remote sensing task vector that is closest to the user intention vector from a preset remote sensing task vector library, obtain the remote sensing task corresponding to the target remote sensing task vector, obtain the task type of the remote sensing task, and select the corresponding remote sensing algorithm based on the task type; A control parameter generation module is used to generate a control parameter set based on the remote sensing algorithm and the state parameters of the current remote sensing device, and send the control parameter set to the control system of the remote sensing device within a first preset time, so that the control system can remotely control the remote sensing device according to the control parameter set.

8. A storage medium, characterized in that: The storage medium stores one or more programs, which, when executed by a processor, implement the remote sensing control method based on a natural language model as described in any one of claims 1 to 6.

9. An electronic device comprising a memory and a processor, wherein: The memory is used to store computer programs; When the processor is used to execute the computer program stored in the memory, it implements the remote sensing control method based on the natural language model as described in any one of claims 1 to 6.

Images