CN115470796A - Text instruction generation method and equipment for air traffic control simulation training - Google Patents
Text instruction generation method and equipment for air traffic control simulation training Download PDFInfo
- Publication number
- CN115470796A CN115470796A CN202211116595.8A CN202211116595A CN115470796A CN 115470796 A CN115470796 A CN 115470796A CN 202211116595 A CN202211116595 A CN 202211116595A CN 115470796 A CN115470796 A CN 115470796A
- Authority
- CN
- China
- Prior art keywords
- instruction
- text
- control
- intention
- air traffic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
- G06F16/355—Class or cluster creation or modification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
-
- G—PHYSICS
- G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
- G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
- G09B9/00—Simulators for teaching or training purposes
- G09B9/02—Simulators for teaching or training purposes for teaching control of vehicles or other craft
- G09B9/08—Simulators for teaching or training purposes for teaching control of vehicles or other craft for teaching control of aircraft, e.g. Link trainer
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Aviation & Aerospace Engineering (AREA)
- Business, Economics & Management (AREA)
- Educational Administration (AREA)
- Educational Technology (AREA)
- Data Mining & Analysis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a text instruction generation method and equipment for air traffic control simulation training, belonging to the technical field of civil air traffic control simulation training and comprising the following steps: inputting an instruction text of a controller of the air traffic control conversation, and preprocessing the instruction text; inputting the preprocessed instruction text into an empty management instruction analysis module, and extracting a semantic tag sequence and an intention of the preprocessed instruction text; classifying the control instructions in the preprocessed instruction text according to the intention; based on the instruction repeating generation module, the semantic tag sequence and the classified control instruction are respectively generated into corresponding instruction parameters and instruction repeating texts for operation simulation flight, so that automatic generation of dialogue and instructions in the control simulation training process is realized, the training cost is reduced, and meanwhile, the efficiency of control simulation training is improved.
Description
Technical Field
The invention relates to the technical field of civil aviation management simulation training, in particular to a text instruction generation method and equipment for air traffic management simulation training.
Background
In the civil aviation control scene, a controller carries out instruction interaction with a pilot through radio equipment to command the aircraft to fly, the controller sends a control instruction, a captain confirms that the instruction is correct through a mode of repeating the control instruction, the aircraft is controlled to fly according to the control instruction, and the controller sends the control instruction. The existing control simulation training system depends on a manual mode to serve as a role of a captain, and manually inputs a control command to control the airplane flight of the simulated flight software in the training process, so that the training cost of a controller is higher, and the mode of manually inputting the control command is low in efficiency and easy to make mistakes.
Therefore, a text instruction generation method and a text instruction generation device for air traffic control simulation training are needed to solve the problems of manual conversation and manual instruction input, so that the efficiency of control simulation training is improved.
Disclosure of Invention
The invention aims to overcome the defects that a control simulation training system in the prior art depends on a manual mode to serve as a role of a captain, and manually inputs a control instruction to control the airplane of simulated flight software to fly in the training process, so that the training cost of a controller is higher, the efficiency of the manual control instruction input mode is low, and errors are easy to occur, and provides a text instruction generation method and equipment for air traffic control simulation training.
In order to achieve the above purpose, the invention provides the following technical scheme:
a text instruction generation method for air traffic control simulation training comprises the following steps:
s1: inputting an instruction text of a controller of the air traffic control conversation, and preprocessing the instruction text;
s2: inputting the preprocessed instruction text into an empty management instruction analysis module, and extracting a semantic tag sequence and an intention of the preprocessed instruction text;
the empty pipe instruction analysis module comprises: a word segmentation model, an intention identification model and an entity naming model;
the word segmentation model consists of a bidirectional LSTM layer and a CRF layer, wherein the bidirectional LSTM layer is used for extracting features, and the CRF layer is used for outputting word segmentation labels;
the intention recognition model consists of a bidirectional LSTM layer and two fully-connected layers, and averaging calculation on the time dimension of the text sequence is added between the two fully-connected layers to unify characteristic dimensions;
the entity naming model consists of a bidirectional LSTM layer and a CRF layer, wherein the bidirectional LSTM layer is used for extracting features, and the CRF layer is used for outputting semantic labels;
s3: classifying the control instructions in the preprocessed instruction text according to the intention;
s4: and respectively generating corresponding instruction parameters and instruction repeating texts of the operation simulation flight according to the semantic tag sequence and the classified control instruction based on an instruction repeating generation module.
Adopt above-mentioned technical scheme, carry out the label of intention analysis and word segmentation label to the instruction text of input through empty pipe instruction analysis module, again according to empty pipe repeating rule and empty pipe simulation training instruction rule respectively to handling different types of control command, generate corresponding intention parameter and repeating instruction text, and simultaneously, special circumstances deals with the module and is used for the aircraft to meet with output captain text instruction under the special circumstances, thereby realized dialogue generation and control aircraft flight, realized control simulation training in-process automatic generation dialogue and instruction, the training cost has been reduced, the efficiency of control simulation training has been improved, the richness of control simulation training has also been improved simultaneously.
As a preferable scheme of the present invention, the step S1 of preprocessing the instruction text includes the following steps:
s11: removing punctuations and special characters in the instruction text, and filtering meaningless miscellaneous words in the instruction text, wherein the meaningless miscellaneous words comprise tone words and repeated words;
s12: and combining the special words of the combination type appearing in the instruction text.
As a preferable embodiment of the present invention, the step S2 includes:
s21: inputting the preprocessed instruction text, and classifying Chinese and English firstly;
s22: inputting Chinese characters into the word segmentation model to perform Chinese instruction word segmentation, and outputting word segmentation labels;
s23: comparing the word group in the word segmentation label with an empty management special vocabulary, and correcting the wrong special vocabulary segmentation in the word segmentation label to obtain a corrected word segmentation label, wherein the data in the empty management special vocabulary at least comprises a place, an instruction, an airline name and a place name;
s24: inputting English texts or the corrected word segmentation labels into the intention recognition model, outputting probability vectors of all intention classifications, wherein the length of the vectors is the same as the number of intention categories, the value of each position in the vectors is a prediction probability value containing the intention, meanwhile, setting a threshold value, screening the prediction probability values to obtain the prediction probability value with the reliability higher than the threshold value, checking whether keywords corresponding to the prediction probability value are contained in the preprocessed instruction text, and if so, obtaining the final intention;
s25: inputting the English text or the corrected participle label into the entity naming model, outputting a semantic label sequence corresponding to the preprocessed instruction text, removing the instruction semantics needing to be corrected in the semantic label sequence of the instruction text containing more corrected intentions in the final intentions, and only keeping the corrected instruction semantics as the final semantic label sequence.
As a preferable embodiment of the present invention, the step S2 further includes: and carrying out corresponding post-processing on the intention and the semantic tag sequence, replacing fixed professional vocabularies with English abbreviations, and replacing Chinese and English numbers with Arabic numbers.
As a preferred aspect of the present invention, the intention includes: declarative regulatory intents, regulatory recognition intents, query regulatory intents.
As a preferable embodiment of the present invention, in step S3, the regulation instruction is divided into: declarative regulatory instructions, regulatory identification instructions, query-like instructions;
the declarative control instruction is a traffic intervention instruction issued to a pilot through a declarative statement, and the instruction form is an AB type of a flight number plus instruction;
the control identification instruction is a controller instruction after the aircraft actively contacts with a controller for the first time when the aircraft enters a control sector;
the inquiry class instruction is divided into: the method comprises the steps of repeating confirmation instructions and negotiation instructions, wherein the repetition instructions are repeated confirmation of pilot instruction requests by controllers, and the negotiation instructions are feasibility of issuing control instructions to the pilot inquiry.
As a preferable embodiment of the present invention, the step S4 includes: receiving the stated regulation instruction, sequentially adjusting and complementing a flight number and an instruction according to the empty management repeating rule, wherein the stated regulation instruction is changed into a BA type instruction of adding the flight number to the instruction, the BA type instruction is a repeating instruction text of the stated regulation instruction, a stated regulation intention parameter is extracted from the stated regulation intention corresponding to the stated regulation instruction, and the stated regulation intention parameter are converted into instruction parameters which can be identified by a simulator;
receiving the control identification instruction, repeating an aircraft call sign by a pilot according to the air traffic control repeating rule, analyzing a semantic tag sequence corresponding to the control identification instruction, extracting the aircraft call sign, extracting control identification intention parameters from the control identification intention corresponding to the control identification instruction, and converting the control identification intention and the control identification intention parameters into instruction parameters which can be identified by a simulator;
receiving the inquiry type command, judging the type of the inquiry type command, if judging that the inquiry type command is a repeated confirmation type command, repeating the statement command by a pilot according to the empty management repeating rule, dividing an aircraft call sign and other commands in the repeated confirmation type command according to a corresponding semantic tag sequence, reordering the commands, and finishing the aircraft call sign to obtain a command repeating text of the inquiry type command; if the judgment result is a negotiation instruction, extracting an inquiry control intention parameter from the inquiry control intention corresponding to the negotiation instruction according to the empty management repeating rule, judging the inquiry control intention and the inquiry control intention parameter by the captain according to the flight state, repeating the instruction if the inquiry control intention and the inquiry control intention parameter are legal, converting the control identification intention and the control identification intention parameter into an instruction parameter which can be identified by the simulator, and starting the next round of negotiation if the judgment result is not the rule.
As a preferable embodiment of the present invention, the step S1 further includes: and constructing a special duty handling module, processing special situation scenes encountered in the flight process, and generating and outputting a captain text instruction.
As a preferred solution of the present invention, the special case scenario in step S1 includes: aircraft faults, crew flare conditions, external environmental disturbances.
In another aspect, a text instruction generating device for air traffic control simulation training is provided, which includes at least one processor, and a memory communicatively connected to the at least one processor; the memory stores instructions executable by the at least one processor to cause the at least one processor to perform the method of any one of the above.
Compared with the prior art, the invention has the beneficial effects that: carry out the mark of intention analysis and word segmentation label to the instruction text of input through empty pipe instruction analysis module, again handle different types of control command respectively according to empty pipe repeating rule and empty pipe simulation training instruction rule, produce corresponding intention parameter and repeating instruction text, and simultaneously, special circumstances handles the module and is used for the aircraft to meet with output captain text instruction under the special circumstances, the analytic rate of instruction text has been improved, the suitability is strong, make the analysis of instruction text more intelligent, thereby conversation generation and control aircraft flight have been realized, automatic generation conversation and instruction in the control simulation training process have been realized, the training cost has been reduced, the efficiency of control simulation training has been improved, the richness of control simulation training has also been improved simultaneously.
Drawings
Fig. 1 is a flowchart of a text instruction generation method for air traffic control simulation training according to embodiment 1 of the present invention;
fig. 2 is an instruction parsing flowchart of a text instruction generation method for air traffic control simulation training according to embodiment 1 of the present invention;
fig. 3 is a word segmentation model structure diagram of a text instruction generation method for air traffic control simulation training according to embodiment 1 of the present invention;
fig. 4 is an intention prediction and word segmentation label diagram of the text instruction generation method for air traffic control simulation training according to embodiment 1 of the present invention;
FIG. 5 is a block diagram of a text instruction generating device for air traffic control simulation training according to embodiment 2 of the present invention
Detailed Description
The present invention will be described in further detail with reference to test examples and specific embodiments. It should be understood that the scope of the above-described subject matter of the present invention is not limited to the following examples, and any technique realized based on the contents of the present invention is within the scope of the present invention.
Example 1
A text instruction generation method for air traffic control simulation training is disclosed, as shown in FIG. 1, and comprises the following steps:
s1: inputting an instruction text of a controller of the air traffic control conversation, and preprocessing the instruction text;
the step S1 of preprocessing the instruction text comprises the following steps:
s11: removing punctuations and special characters in the instruction text, and filtering meaningless miscellaneous words in the instruction text, wherein the meaningless miscellaneous words comprise tone words and repeated words;
specifically, punctuation marks and special characters in the instruction text are removed, for example, as shown in table 1:
TABLE 1
Specifically, the meaningless miscellaneous words in the instruction text are filtered, including the tone words and the repeated words, so as to avoid negative interference on subsequent semantic analysis, for example, as shown in table 2:
TABLE 2
S12: combining special words and phrases appearing in the instruction text, and combining words and phrases appearing in the water sentences, such as English airline company names;
the step S1 further includes: constructing a special duty handling module, processing special situation scenes encountered in the flight process, and generating and outputting a captain text instruction, wherein the captain text instruction is in the form of: the special situation of the calling number of the aircraft can be added into the operation requested by the captain, such as the request of forced landing, the request of flying around and the height adjustment, and the special situation scene comprises the following steps: aircraft faults, unit emergency conditions, external environmental disturbances;
specifically, aircraft failure: in this scenario, the captain actively reports a condition to the controller, requests to perform some operations or waits for a controller command, as shown in table 3, for a mechanical failure of the aircraft itself, such as a failure or damage to the engine, other equipment of the aircraft, or the like:
TABLE 3
The burst condition of the unit: the flight crew encounters abnormal situations, such as personal injury, hijack, attack or lost direction, under this scenario, the captain actively reports the situation to the controller, requests to perform some operations or waits for the controller to command instructions, as shown in table 4:
TABLE 4
External environment interference: in the case of extreme weather, such as rain clouds, thunderstorms, hail, or other environmental factors affecting the flight, the captain actively reports the situation to the controller, requests to perform some operations or waits for the controller to command instructions, as shown in table 5:
TABLE 5
S2: inputting the preprocessed instruction text into an empty management instruction analysis module, and extracting a semantic tag sequence and an intention of the preprocessed instruction text;
as shown in fig. 2, the step S2 includes:
s21: inputting the preprocessed instruction text, classifying Chinese and English firstly, detecting whether the preprocessed instruction text contains Chinese characters, if so, determining the preprocessed instruction text to be Chinese, and if not, determining the preprocessed instruction text to be English;
s22: as shown in fig. 3, inputting a chinese character into a word segmentation model for chinese instruction word segmentation, wherein the word segmentation model is composed of a bidirectional LSTM layer and a CRF layer, the bidirectional LSTM layer is used for extracting features, and the CRF layer is used for outputting word segmentation labels;
s23: comparing the word groups in the word segmentation labels with an empty-managed special vocabulary, and correcting the error special vocabulary segmentation in the word segmentation labels to obtain corrected word segmentation labels, so that the accuracy of word segmentation is ensured, wherein the data in the empty-managed special vocabulary comprises a vocabulary formed by fixed word groups such as places, instructions, names of airlines, place names and the like;
s24: constructing an intention recognition model, as shown in fig. 4, wherein the intention recognition model consists of a bidirectional LSTM layer and two fully-connected layers, an averaging calculation on a text sequence time dimension is added between the two fully-connected layers to unify feature dimensions, an english text or the modified word segmentation label is input, probability vectors of all intention classifications are output, the vector length is the same as the number of intention categories, the value of each position in the vectors is a prediction probability value for judging whether the intention is included, meanwhile, a threshold is set, the prediction probability values are screened, a prediction probability value with the reliability higher than the threshold is screened out, whether a keyword corresponding to the prediction probability value is included in the preprocessed instruction text is checked, and if yes, the final intention is obtained;
s25: and constructing an entity naming model, as shown in fig. 4, wherein the entity naming model consists of a bidirectional LSTM layer and a CRF layer, inputting the english text or the modified participle tag, outputting a semantic tag sequence corresponding to each value in the preprocessed instruction text, removing instruction semantics needing to be corrected in the semantic tag sequence of the instruction text containing a more corrected intention in the final intention, and only keeping the corrected instruction semantics as the final semantic tag sequence.
The step S2 further includes: performing corresponding post-processing on the intention and the semantic tag sequence, replacing fixed professional vocabularies with English abbreviations, replacing Chinese and English numbers with Arabic numbers, and replacing a flight number 'two-in-three-east-orientation corner' with 'CES 2307'; the value in the command is normalized to an aviation standard value, such as the aviation altitude "Yao hole Turn" becomes the standard value "10700".
The intent includes: declarative regulatory intents, regulatory recognition intents, query regulatory intents.
S3: classifying the control instructions in the preprocessed instruction text according to the intention;
in step S3, the control instruction is divided into: declarative control instructions, control identification instructions, query class instructions;
the stated control command is a traffic intervention command issued to the pilot by a stated statement, and the command is AB type of flight number plus command, and the intention of the command includes a stated control intention;
specifically, as shown in table 6:
| a control instruction: the four-season eight-crutch and eight-crutch rise to the Yao-hole crutch |
TABLE 6
The control identification instruction is a controller instruction after the airplane actively contacts a controller for the first time when entering a control sector, and the intention comprises a control identification intention;
specifically, as shown in table 7:
| a control instruction: the radar for turning two or three holes in east is already identified |
TABLE 7
The inquiry class instruction is divided into: determining whether the intentions are inquiry instructions according to the fact whether the intentions comprise flavor-seeking control intentions, if yes, determining the inquiry instructions, dividing the inquiry instructions into repeated confirmation instructions and negotiation instructions according to keywords in the instruction text, wherein the repeated confirmation instructions are repeated confirmation of a controller to a pilot instruction request, and the instructions comprise keyword 'confirmation';
s4: respectively generating corresponding instruction parameters and instruction repeating texts of the operation simulation flight according to the semantic tag sequence and the classified control instruction based on an instruction repeating generation module;
specifically, the instruction type is obtained by analyzing a large amount of real empty management data, and a system configuration file is provided for adding or modifying the instruction type, so that the method has the characteristics of comprehensive instruction processing and flexible and configurable system;
the step S4 includes: receiving the stated regulation instruction, sequentially adjusting and complementing a flight number and an instruction according to the empty management repeating rule, wherein the stated regulation instruction is changed into a BA type instruction of adding the flight number to the instruction, the BA type instruction is a repeating instruction text of the stated regulation instruction, a stated regulation intention parameter is extracted from the stated regulation intention corresponding to the stated regulation instruction, and the stated regulation intention parameter are converted into instruction parameters which can be identified by a simulator;
specifically, inputting the instruction in table 6 into the empty pipe instruction parsing module to obtain the semantic parsing result of the instruction, including: the word segmentation labels, declarative regulatory intents, and declarative regulatory intention parameters, as shown in Table 8:
TABLE 8
The instruction repeat generation module sequentially adjusts and replys the flight number and the instruction key elements according to the semantic parsing result described in table 8 and the empty management repeat rule, and the instruction form becomes a BA type instruction of the instruction plus the flight number, that is, as shown in table 9:
| repeating the instructions: go up to the Yao hole to turn a Sichuan eight-turn |
TABLE 9
At the same time, the conventional declarative regulatory intention and the conventional declarative regulatory intention parameters are used to operate a specified aircraft to perform an instructional operation, the conventional declarative regulatory intention and the conventional declarative regulatory intention parameters in the instructional instance are converted to as shown in table 10 according to an air traffic simulation training instructional rule:
| ACID：CSC8878；CMD：LVL1070 |
TABLE 10
Receiving the control identification instruction, repeating an aircraft call sign by a pilot according to the air traffic control repeating rule, analyzing a semantic tag sequence corresponding to the control identification instruction, extracting the aircraft call sign, extracting control identification intention parameters from the control identification intention corresponding to the control identification instruction, and converting the control identification intention and the control identification intention parameters into instruction parameters which can be identified by a simulator;
receiving the query type command, determining the type of the query type command, if determining that the command is a repeat confirmation type command, repeating the statement command by the pilot according to the empty management repeating rule, dividing the aircraft call number and other commands in the repeat confirmation type command according to the corresponding semantic tag sequence, reordering the aircraft call number, and ending with the aircraft call number to obtain a command repeating text of the query type command, for example, as shown in table 11:
| a controller: do you have a good idea of flying into both the east and the three-hole crutch |
| The mechanical length repeats: two-three hole crutch for eastern flying adult |
TABLE 11
If the negotiation instruction is judged, extracting an inquiry control intention parameter from the inquiry control intention corresponding to the negotiation instruction according to the air traffic control repeating rule, judging the inquiry control intention and the inquiry control intention parameter by the captain according to the flight state, repeating the instruction if the inquiry control intention and the inquiry control intention parameter are legal, starting the next round of negotiation if the inquiry is not legal, and repeating the text of the legal negotiation instruction as shown in a table 12:
| a controller: do you can turn in two hundred directions in two or three holes at east |
| The mechanical length repeats: walking stick with two hundred-degree of orientation, two-degree of east and two-degree of hole |
TABLE 12
By adopting the technical scheme, the input instruction text is subjected to intention analysis and word segmentation label labeling through the empty management instruction analysis module, different types of control instructions are respectively processed according to the empty management repeating rule and the empty management simulation training instruction rule, corresponding intention parameters and repeating instruction texts are generated, meanwhile, the special situation handling module is used for outputting a captain text instruction under the condition that the airplane encounters special conditions, the accuracy rate of instruction text analysis is improved, the applicability is high, the analysis of the instruction text is more intelligent, the dialogue generation and control of airplane flight are realized, the automatic dialogue and instruction generation in the control simulation training process is realized, the training cost is reduced, the efficiency of control simulation training is improved, and the richness of the control simulation training is also improved.
Example 2
As shown in fig. 5, a text instruction generating device for air traffic control simulation training comprises at least one processor, and a memory communicatively connected to the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method and the apparatus for generating text instructions for air traffic control simulation training according to the above embodiments. The input and output interface can comprise a display, a keyboard, a mouse and a USB interface and is used for inputting and outputting data; the power supply is used for supplying electric energy to the electronic equipment.
Those skilled in the art will understand that: all or part of the steps for realizing the method embodiments can be completed by hardware related to program instructions, the program can be stored in a computer readable storage medium, and the program executes the steps comprising the method embodiments when executed; and the aforementioned storage medium includes: various media that can store program codes, such as a removable Memory device, a Read Only Memory (ROM), a magnetic disk, or an optical disk.
When the integrated unit of the present invention is implemented in the form of a software functional unit and sold or used as a separate product, it may also be stored in a computer-readable storage medium. Based on such understanding, the technical solutions of the embodiments of the present invention may be essentially implemented or a part contributing to the prior art may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a removable storage device, a ROM, a magnetic or optical disk, or other various media that can store program code.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (10)
1. A text instruction generation method for air traffic control simulation training is characterized by comprising the following steps:
s1: inputting an instruction text of a controller of the air traffic control conversation, and preprocessing the instruction text;
s2: inputting the preprocessed instruction text into an empty management instruction analysis module, and extracting a semantic tag sequence and an intention of the preprocessed instruction text;
the empty pipe instruction analysis module comprises: a word segmentation model, an intention identification model and an entity naming model;
the word segmentation model consists of a bidirectional LSTM layer and a CRF layer, wherein the bidirectional LSTM layer is used for extracting features, and the CRF layer is used for outputting word segmentation labels;
the intention recognition model consists of a bidirectional LSTM layer and two full-connection layers, and averaging calculation on the time dimension of the text sequence is added between the two full-connection layers to unify the feature dimension;
the entity naming model consists of a bidirectional LSTM layer and a CRF layer, wherein the bidirectional LSTM layer is used for extracting features, and the CRF layer is used for outputting semantic labels;
s3: classifying the control instructions in the preprocessed instruction text according to the intention;
s4: and respectively generating corresponding instruction parameters and instruction repeat texts of the operation simulation flight according to the semantic tag sequence and the classified control instruction based on an instruction repeat generation module.
2. The method for generating the text instruction for the air traffic control simulation training as claimed in claim 1, wherein the step S1 of preprocessing the instruction text comprises the steps of:
s11: removing punctuations and special characters in the instruction text, and filtering meaningless miscellaneous words in the instruction text, wherein the meaningless miscellaneous words comprise tone words and repeated words;
s12: and merging the combined special words appearing in the instruction text.
3. The method for generating the text instruction for the air traffic control simulation training as claimed in claim 1, wherein the step S2 comprises:
s21: inputting the preprocessed instruction text, and classifying Chinese and English firstly;
s22: inputting Chinese characters into the word segmentation model to perform Chinese instruction word segmentation, and outputting word segmentation labels;
s23: comparing the word group in the word segmentation label with an empty management special vocabulary, and correcting the wrong special vocabulary segmentation in the word segmentation label to obtain a corrected word segmentation label, wherein the data in the empty management special vocabulary at least comprises a place, an instruction, an airline name and a place name;
s24: inputting English texts or the corrected word segmentation labels into the intention recognition model, outputting probability vectors of all intention classifications, wherein the length of the vectors is the same as the number of the intention categories, the value of each position in the vectors is a prediction probability value containing the intention, meanwhile, setting a threshold value, screening the prediction probability values to obtain the prediction probability values with the reliability higher than the threshold value, checking whether keywords corresponding to the prediction probability values are contained in the preprocessed instruction texts, and if yes, obtaining the final intention;
s25: inputting the English text or the corrected participle label into the entity naming model, outputting a semantic label sequence corresponding to the preprocessed instruction text, removing the instruction semantics needing to be corrected in the semantic label sequence of the instruction text containing more corrected intentions in the final intentions, and only keeping the corrected instruction semantics as the final semantic label sequence.
4. The method for generating the text instruction for the air traffic control simulation training as claimed in claim 3, wherein the step S2 further comprises: and carrying out corresponding post-processing on the intention and the semantic tag sequence, replacing fixed professional vocabularies with English abbreviations, and replacing Chinese and English numbers with Arabic numbers.
5. The method for generating the text instruction for the air traffic control simulation training as claimed in claim 4, wherein the intention comprises: declarative regulatory intents, regulatory recognition intents, query regulatory intents.
6. The method as claimed in claim 5, wherein in step S3, the control instruction is divided into: declarative control instructions, control identification instructions, query class instructions;
the declarative control instruction is a traffic intervention instruction issued to the pilot through a declarative statement, and the instruction is in the form of an AB type of a flight number plus an instruction;
the control identification instruction is a controller instruction after the airplane actively contacts with a controller for the first time when the airplane enters a control sector;
the inquiry class instruction is divided into: the method comprises the steps of repeating confirmation instructions and negotiation instructions, wherein the repeating confirmation instructions are repeated confirmation of pilot instruction requests by a controller, and the negotiation instructions are feasibility of issuing control instructions to the pilot.
7. The method for generating the text instruction for the air traffic control simulation training as claimed in claim 6, wherein the step S4 comprises: receiving the declarative control instruction, sequentially adjusting and complementing a flight number and an instruction according to the empty management repeating rule, wherein the declarative control instruction is changed into a BA type instruction of adding the flight number to the instruction, the BA type instruction is a repeating instruction text of the declarative control instruction, declarative control intention parameters are extracted from the declarative control intention corresponding to the declarative control instruction, and the declarative control intention parameters are converted into instruction parameters which can be identified by a simulator;
receiving the control identification instruction, repeating an aircraft call sign by a pilot according to the air traffic control repeating rule, analyzing a semantic tag sequence corresponding to the control identification instruction, extracting the aircraft call sign, extracting control identification intention parameters from the control identification intention corresponding to the control identification instruction, and converting the control identification intention and the control identification intention parameters into instruction parameters which can be identified by a simulator;
receiving the inquiry type command, judging the type of the inquiry type command, if judging that the inquiry type command is a repeated confirmation type command, repeating the statement command by a pilot according to the empty management repeating rule, dividing an aircraft call sign and other commands in the repeated confirmation type command according to a corresponding semantic tag sequence, reordering the commands, and finishing the aircraft call sign to obtain a command repeating text of the inquiry type command; if the judgment result is a negotiation instruction, extracting an inquiry control intention parameter from the inquiry control intention corresponding to the negotiation instruction according to the empty management repeating rule, judging the inquiry control intention and the inquiry control intention parameter by the captain according to the flight state, repeating the instruction if the inquiry control intention and the inquiry control intention parameter are legal, converting the control identification intention and the control identification intention parameter into an instruction parameter which can be identified by the simulator, and starting the next round of negotiation if the judgment result is not the rule.
8. The method for generating the text instruction for the air traffic control simulation training according to any one of claims 1 to 7, wherein the step S1 further comprises: and constructing a special service handling module, processing special situation scenes encountered in the flight process, and generating and outputting a captain text instruction.
9. The method for generating text instructions for air traffic control simulation training according to claim 1, wherein the special case scenario in step S1 includes: aircraft faults, crew emergency conditions, external environmental disturbances.
10. A text instruction generating device for air traffic control simulation training, which is characterized by comprising at least one processor and a memory which is in communication connection with the at least one processor; the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211116595.8A CN115470796A (en) | 2022-09-14 | 2022-09-14 | Text instruction generation method and equipment for air traffic control simulation training |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211116595.8A CN115470796A (en) | 2022-09-14 | 2022-09-14 | Text instruction generation method and equipment for air traffic control simulation training |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115470796A true CN115470796A (en) | 2022-12-13 |
Family
ID=84332613
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211116595.8A Pending CN115470796A (en) | 2022-09-14 | 2022-09-14 | Text instruction generation method and equipment for air traffic control simulation training |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115470796A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116662555A (en) * | 2023-07-28 | 2023-08-29 | 成都赛力斯科技有限公司 | Request text processing method and device, electronic equipment and storage medium |
| CN117593924A (en) * | 2024-01-19 | 2024-02-23 | 中国民用航空飞行学院 | Scene reproduction-based air traffic controller training method and system |
-
2022
- 2022-09-14 CN CN202211116595.8A patent/CN115470796A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116662555A (en) * | 2023-07-28 | 2023-08-29 | 成都赛力斯科技有限公司 | Request text processing method and device, electronic equipment and storage medium |
| CN116662555B (en) * | 2023-07-28 | 2023-10-20 | 成都赛力斯科技有限公司 | Request text processing method and device, electronic equipment and storage medium |
| CN117593924A (en) * | 2024-01-19 | 2024-02-23 | 中国民用航空飞行学院 | Scene reproduction-based air traffic controller training method and system |
| CN117593924B (en) * | 2024-01-19 | 2024-03-26 | 中国民用航空飞行学院 | Scene reproduction-based air traffic controller training method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115470796A (en) | Text instruction generation method and equipment for air traffic control simulation training | |
| CN107688803B (en) | Method and device for verifying recognition result in character recognition | |
| CN113569545B (en) | Control information extraction method based on voice recognition error correction model | |
| CN106910376B (en) | Air traffic operation control instruction monitoring method and system | |
| CN110188345B (en) | Intelligent identification method and device for electric operation ticket | |
| CN110428830B (en) | Regular expression-based empty pipe instruction intention identification method | |
| CN112966555B (en) | Remote sensing image airplane identification method based on deep learning and component prior | |
| CN111627257B (en) | Control instruction safety rehearsal and verification method based on aircraft motion trend prejudgment | |
| CN113053367A (en) | Speech recognition method, model training method and device for speech recognition | |
| CN114218932A (en) | Aviation fault text abstract generation method and device based on fault cause and effect map | |
| CN113673228A (en) | Text error correction method, text error correction device, computer storage medium and computer program product | |
| CN113268595A (en) | Structured airport alarm processing method based on entity relationship extraction | |
| CN110069771B (en) | Control instruction information processing method based on semantic chunk | |
| CN110232121B (en) | Semantic network-based control instruction classification method | |
| US20210224307A1 (en) | Information processing device, information processing system, and computer program product | |
| CN115938347B (en) | Flight student conversation normalization scoring method and system based on voice recognition | |
| CN115359784B (en) | Civil aviation land-air voice recognition model training method and system based on transfer learning | |
| CN109460547B (en) | Structured control instruction extraction method based on natural language processing | |
| CN111652157A (en) | Dictionary entry extraction and identification method for low-resource languages and general languages | |
| Rataj et al. | AcListant with Continuous Learning: Speech Recognition in Air Traffic Control | |
| CN113627197B (en) | Text intention recognition method, device, equipment and storage medium | |
| CN116092342A (en) | Automatic response and quality assessment method and system for controller simulation training | |
| CN115689603A (en) | User feedback information collection method and device and user feedback system | |
| US20030236668A1 (en) | Process and device for devising an abridged form of any term that is used in an alarm message intended to be displayed on a screen of the cockpit of an aircraft | |
| CN114510925A (en) | Chinese text error correction method, system, terminal equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |