A coal mine safety knowledge interaction system based on natural language model

Abstract

The invention discloses a coal mine safety knowledge interaction system based on a natural language model, and belongs to the field of human-computer interaction technology. The coal mine safety knowledge interaction system of the invention includes a safety database construction module, a voice interaction module, a voice analysis module, a real-time scene establishment module, a similarity calculation module, a program formulation module and a display module; the safety database construction module is used to classify the coal mine safety knowledge in history and construct a coal mine safety knowledge database; the voice interaction module is used to conduct voice questions and answers on safety knowledge and voice broadcasts of risk warnings with users, analyze user needs, and guide users to provide the four-element feature data of human, machine, environment and management safety of the operation process required by the system; the similarity calculation module is used to calculate the similarity between the real-time virtual environment and each environmental data in the database, and judge the real-time environmental risk category.

Description

Technical Field

[0001] The present invention relates to the technical field of human-computer interaction, and in particular to a coal mine safety knowledge interaction system based on a natural language model. Background Art

[0002] In recent years, with the application of modern technologies such as information technology and unmanned technology, coal mine safety operations have entered a new stage. In the new stage, due to the wide range of coal mine safety training knowledge involved, some personnel are unable to effectively remember the relevant knowledge for a long time, resulting in some work being unable to be carried out or wrong operations being implemented. In view of this, a human-computer interactive system is designed. When the staff is unsure of the operating procedures or safety precautions, they can obtain relevant information through a wrist-worn device; however, the general interactive system only mechanically searches for the required content, which has inaccurate searches, imperfect human-computer interaction, and cannot solve the actual problems of users. Therefore, it is crucial to increase the reliability of human-computer interaction and the practicality of the knowledge recommended by the system. Summary of the invention

[0003] The purpose of the present invention is to provide a coal mine safety knowledge interaction system based on a natural language model to solve the problems raised in the above background technology.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0005] A coal mine safety knowledge interaction system based on a natural language model, the coal mine safety knowledge interaction system includes a safety database construction module, a voice interaction module, a voice analysis module, a real-time scene establishment module, a similarity calculation module, a plan formulation module and a display module;

[0006] The safety database construction module is used to collect and summarize historical coal mine safety knowledge, and construct a coal mine safety knowledge database after classification;

[0007] The security database construction module includes a data collection unit and a database construction unit;

[0008] The data collection unit is used to collect coal mine safety knowledge in the network. Suppose the collected coal mine safety knowledge is , represents the 1st, 2nd, 3rd, ..., nth coal mine safety knowledge collected, where n is a positive integer; extract the application scenario vocabulary in the collected coal mine safety knowledge, and classify the collected coal mine safety knowledge in the first dimension according to different application scenario vocabulary, and obtain the first dimension category of coal mine safety knowledge as , represents the coal mine safety knowledge of the first, second, third, ..., m application scenarios, where m is a positive integer;

[0009] After classifying the coal mine safety knowledge in the first dimension, the environmental phrases describing the environmental data in the coal mine safety knowledge in each application scenario of the first dimension are extracted, and the number of each environmental phrase is counted. , represents the number of the 1st, 2nd, 3rd, ..., jth environmental phrases in coal mine safety knowledge, where j is a positive integer. The proportion of each environmental phrase in the total environmental phrases is calculated using the formula:

[0010]

[0011] In the formula, a represents the proportion of each environmental phrase in the total environmental phrases. represents the number of each environmental phrase, i∈(1,j); j represents the number of types of environmental phrases; the proportion of j types of environmental phrases is , represents the proportion of the 1st, 2nd, 3rd, ..., jth environmental phrases; the maximum proportion of the jth environmental phrase is selected as the environmental feature of the corresponding coal mine safety knowledge; the coal mine safety knowledge is classified into the second dimension according to the environmental feature of each coal mine safety knowledge, and the second dimension category of the coal mine safety knowledge is obtained as follows: , Represents the 1st, 2nd, 3rd, ..., hth type of coal mine safety knowledge in the second dimension, where h is a positive integer.

[0012] Due to the complexity of coal mine safety knowledge, coal mine safety knowledge is classified into two dimensions. When the corresponding coal mine safety knowledge needs to be found, it can be quickly found according to the two dimensions, which greatly reduces the search time when searching for coal mine safety knowledge and increases work efficiency;

[0013] The voice interaction module is used to conduct safety knowledge Q&A and risk warning voice broadcast with the user, analyze user needs, and guide the user to provide the four-factor characteristic data of human, machine, environment, and management safety of the operation process required by the system;

[0014] The voice interaction module is used to collect the voice data sent by the user, and use the voice recognition system to convert the voice data sent by the user into text data. Suppose the text data converted from the user's voice data is , represents the 1st, 2nd, 3rd, ..., gth text data after the user's voice data is converted, where g is a positive integer;

[0015] The converted text data is analyzed, scene words in the text data are extracted, and the extracted scene words are compared with the scene characteristics of the first dimension in the coal mine safety knowledge database. When the scene Y_s described by the user is found in the coal mine safety knowledge database, the system issues a voice prompt to ask the user for real-time environmental data on site;

[0016] After analyzing the converted text data, if no scenario vocabulary is extracted or no real-time scenario vocabulary is found in the coal mine safety knowledge database, the system displays the classified scenario vocabulary of the first dimension of coal mine safety knowledge in the database for the user to choose, and finally obtains the real-time scenario required by the user.

[0017] Due to the diversity of users, it is not possible to guarantee that every user can clearly express their needs and ensure that the system can clearly identify them. In order to avoid this situation, it is necessary to use the voice interaction module to actively ask users. Based on user options, it can better help the system understand users' problems and needs faster.

[0018] The speech analysis module is used to analyze the user speech signals collected by the system, extract keywords, generate user questions and real-time on-site features;

[0019] The speech analysis module includes a question generation unit and a scene feature determination unit;

[0020] The question generation unit is used to convert the voice information provided by the user into text, classify the converted text data, process the text data provided by the user using the Transformer model, and perform part-of-speech tagging on each text data. Suppose the tag of each text data is In the mark, A represents each text data, and c represents the part of speech corresponding to the text data; the part of speech set of c is In the set, in represents interrogative words, ad represents adjectives, ve represents verbs, and no represents nouns. After marking the part of speech of each text data, Arrange the text data in the order of to form user questions;

[0021] The scene feature determination unit is used to collect the environment description text in the text data to generate scene environment data, and use the voice interaction module to collect the data value of the environment data provided by the user; the data value corresponding to each type of scene environment data is set to , represents the data value of the 1st, 2nd, 3rd, ..., fth environmental data on site, where f is a positive integer; according to the real-time scene provided by the user, the data value of each environmental data in the second dimension of the h characteristic environmental data in the corresponding scene is extracted from the database. , Represents the data values ​​of the 1st, 2nd, 3rd, ..., fth environmental data in the corresponding scene extracted from the database, and calculates the difference between the data values ​​of each environmental data in the field and in the database. The formula is:

[0022]

[0023] In the formula, C_h represents the difference between the data value of each environmental data on site and in the database, S_h represents the data value corresponding to each environmental data on site, and Z_h represents the data value of each environmental data in the corresponding scene extracted from the database; let the data difference of each environmental data finally calculated be , represents the calculated data difference of the 1st, 2nd, 3rd, ..., fth environmental data;

[0024] The environmental data is screened according to the calculated difference of each environmental data, the formula is:

[0025]

[0026] In the inequality, T represents the filtered environmental data, C_h represents the difference between the data value of each environmental data on site and in the database, f represents the number of types of environmental data, and h represents the number of characteristic environmental types in the second dimension. When the difference between the data value of environmental data on site and in the database is greater than the average value, the corresponding type of environmental data is filtered, and the final filtered environmental data types are , It is represented as the 1st, 2nd, 3rd, ..., dth environmental data after screening, and the dth environmental data after screening are used as the field characteristics.

[0027] The real-time scene building module is used to build a real-time virtual environment using real-time scene features;

[0028] The real-time scene establishment module is used to analyze the environmental data provided by the user, extract the specific data of each type of environmental data in the environmental data provided by the user, correspond them with each other in the system according to the environmental data recorded in the database, use the environmental data in the database to supplement the environmental data not provided by the user, and use special site features to build a three-dimensional virtual environment model.

[0029] The similarity calculation module is used to calculate the similarity between the real-time virtual environment and each environmental data in the database, and determine the risk category of the real-time environment;

[0030] The similarity calculation module is used to collect the environmental data provided by the user, select the environmental category of each type of coal mine safety knowledge in the second dimension under the real-time scene type of the first dimension in the database, extract the environmental data of each type of coal mine safety knowledge in the second dimension from the database, and extract the number of environmental data of the same type between the real-time scene and the environmental data of each type of coal mine safety knowledge in the database. , It represents the number of environmental data of the same type as the environmental data of the first, second, third, ..., h types of coal mine safety knowledge in the real-time on-site environmental data, and extracts the data values ​​of the same type of environmental data, and calculates the similarity between the real-time on-site environment and the environment of each type of coal mine safety knowledge in the second dimension. The formula is:

[0031]

[0032] In the formula, sim represents the similarity between the real-time on-site environment and the environment of each type of coal mine safety knowledge in the second dimension, F represents the number of environmental data of the same type in the real-time on-site and the environmental data of each type of coal mine safety knowledge in the database, S_h represents the real-time on-site data value of the same type of environmental data, and Z_h represents the data value of the same type of environmental data in the second dimension;

[0033] After calculating the similarities of h kinds of coal mine safety knowledge with different environmental characteristics in the same scenario under the first dimension, the h similarities are obtained as follows: , The similarity of coal mine safety knowledge of the 1st, 2nd, 3rd, ..., hth environmental features is represented, and the environmental feature with the greatest similarity is selected as the real-time environmental risk category.

[0034] The plan formulation module is used to search for corresponding emergency response plans in the security database and optimize them according to user problems and real-time environmental risk categories;

[0035] The program formulation module is used to extract the coal mine safety knowledge corresponding to the real-time environmental risk category in the second dimension that solves the user problem according to the user problem judged by the system, the first dimension scenario and the real-time environmental risk category in the second dimension, analyze the extracted coal mine safety knowledge, and extract the change value G_h of the environmental data caused by the operation steps in the coal mine safety knowledge; optimize the coal mine safety knowledge according to the calculated field characteristics, and the formula is:

[0036]

[0037] In the formula, G_op represents the change value of the operating steps in the coal mine safety knowledge for the environmental data after optimization, G_h represents the change value of the operating steps in the coal mine safety knowledge for the environmental data before optimization, Q_h represents the original data value of the environmental data type changed by the operating steps in the coal mine safety knowledge before optimization, and T_h represents the environmental data value of the field characteristics;

[0038] According to the site characteristics, the change values ​​of the types of environmental data corresponding to the d types of site characteristics in the coal mine safety knowledge are optimized and adjusted, and finally the optimized change values ​​of the types of environmental data corresponding to the d types of site characteristics are obtained as follows: , It represents the optimized change value of the environmental data type corresponding to the 1st, 2nd, 3rd, ..., dth field characteristics; and the extracted coal mine safety knowledge is optimized using the optimized change value.

[0039] Because the real-time environment is unlikely to be exactly the same, in order to be closer to the real-time environmental data and ensure the availability of coal mine safety knowledge, the coal mine safety knowledge is optimized according to the actual environmental data to avoid the occurrence of dangerous accidents caused by inadequate processing; this increases the extensive use of coal mine safety knowledge;

[0040] The display module is used to visually display and voice output the solution finally obtained to solve the user's problem.

[0041] The display module is used to visualize the optimized coal mine safety knowledge and mark and display the changed values ​​of each environmental data after optimization; and to voice output the optimized coal mine safety knowledge to guide users to complete operations in sequence according to the prompts of the coal mine safety knowledge and solve user problems; when the system finishes voice output of the coal mine safety knowledge, it uses the voice interaction module to ask the user whether the problem has been solved. The system makes a judgment based on the user's answer. When the user answers "yes", the task is completed. When the user answers "no", the steps are continued to screen the coal mine safety knowledge.

[0042] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0043] 1. The present invention classifies coal mine safety knowledge in two dimensions according to application scenarios and environmental characteristics; after detailed classification, when searching for coal mine safety knowledge, the search for corresponding coal mine safety knowledge can be completed quickly and conveniently, thereby increasing the working efficiency of the system.

[0044] 2. The present invention calculates the similarity between the real-time environment and the environmental features in the coal mine safety knowledge in the database; searches for the category of the real-time environment according to the environmental features; and then combines the user's questions to quickly and accurately obtain the coal mine safety knowledge the user needs, thereby helping the user solve the problem more quickly.

[0045] 3. The present invention calculates the environmental change values ​​and real-time environmental data in the coal mine safety knowledge, optimizes and adjusts the environmental change values ​​in the coal mine safety knowledge according to the actual situation, and is more in line with the user's real-time problems and environment, thereby ensuring the availability of coal mine safety knowledge and increasing the extensive use of coal mine safety knowledge. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0047] Figure 1 It is a module distribution diagram of a coal mine safety knowledge interaction system based on a natural language model of the present invention;

[0048] Figure 2 It is a flow chart of a coal mine safety knowledge interaction system based on a natural language model of the present invention. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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 creative work are within the scope of protection of the present invention.

[0050] See also Figure 1-Figure 2 , the present invention provides a technical solution:

[0051] A coal mine safety knowledge interaction system based on a natural language model, the coal mine safety knowledge interaction system includes a safety database construction module, a voice interaction module, a voice analysis module, a real-time scene establishment module, a similarity calculation module, a plan formulation module and a display module;

[0052] The safety database construction module is used to collect and summarize historical coal mine safety knowledge, and construct a coal mine safety knowledge database after classification;

[0053] The security database construction module includes a data collection unit and a database construction unit;

[0054] The data collection unit is used to collect coal mine safety knowledge in the network. Suppose the collected coal mine safety knowledge is , represents the 1st, 2nd, 3rd, ..., nth coal mine safety knowledge collected, where n is a positive integer; extract the application scenario vocabulary in the collected coal mine safety knowledge, and classify the collected coal mine safety knowledge in the first dimension according to different application scenario vocabulary, and obtain the first dimension category of coal mine safety knowledge as , represents the coal mine safety knowledge of the first, second, third, ..., m application scenarios, where m is a positive integer;

[0055] After classifying the coal mine safety knowledge in the first dimension, the environmental phrases describing the environmental data in the coal mine safety knowledge in each application scenario of the first dimension are extracted, and the number of each environmental phrase is counted. , represents the number of the 1st, 2nd, 3rd, ..., jth environmental phrases in coal mine safety knowledge, where j is a positive integer. The proportion of each environmental phrase in the total environmental phrases is calculated using the formula:

[0056]

[0057] In the formula, a represents the proportion of each environmental phrase in the total environmental phrases. represents the number of each environmental phrase, i∈(1,j); j represents the number of types of environmental phrases; the proportion of j types of environmental phrases is , represents the proportion of the 1st, 2nd, 3rd, ..., jth environmental phrases; the maximum proportion of the jth environmental phrase is selected as the environmental feature of the corresponding coal mine safety knowledge; the coal mine safety knowledge is classified into the second dimension according to the environmental feature of each coal mine safety knowledge, and the second dimension category of the coal mine safety knowledge is obtained as follows: , It represents the 1st, 2nd, 3rd, ..., hth coal mine safety knowledge in the second dimension, where h is a positive integer. The environmental phrase represents the descriptive words about the environment in coal mine safety knowledge;

[0058] Due to the complexity of coal mine safety knowledge, coal mine safety knowledge is classified into two dimensions. When the corresponding coal mine safety knowledge needs to be found, it can be quickly found according to the two dimensions, which greatly reduces the search time when searching for coal mine safety knowledge and increases work efficiency;

[0059] The voice interaction module is used to conduct safety knowledge Q&A and risk warning voice broadcast with the user, analyze user needs, and guide the user to provide the four-factor characteristic data of human, machine, environment, and management safety of the operation process required by the system;

[0060] The voice interaction module is used to collect the voice data sent by the user, and use the voice recognition system to convert the voice data sent by the user into text data. Suppose the text data converted from the user's voice data is , represents the 1st, 2nd, 3rd, ..., gth text data after the user's voice data is converted, where g is a positive integer;

[0061] The converted text data is analyzed, scene words in the text data are extracted, and the extracted scene words are compared with the scene characteristics of the first dimension in the coal mine safety knowledge database. When the scene Y_s described by the user is found in the coal mine safety knowledge database, the system issues a voice prompt to ask the user for real-time environmental data on site;

[0062] After analyzing the converted text data, if no scenario vocabulary is extracted or no real-time scenario vocabulary is found in the coal mine safety knowledge database, the system displays the classified scenario vocabulary of the first dimension of coal mine safety knowledge in the database for the user to choose, and finally obtains the real-time scenario required by the user.

[0063] Due to the diversity of users, it is not possible to guarantee that every user can clearly express their needs and ensure that the system can clearly identify them. In order to avoid this situation, it is necessary to use the voice interaction module to actively ask users. Based on user options, it can better help the system understand users' problems and needs faster.

[0064] The speech analysis module is used to analyze the user speech signals collected by the system, extract keywords, generate user questions and real-time on-site features;

[0065] The speech analysis module includes a question generation unit and a scene feature determination unit;

[0066] The question generation unit is used to convert the voice information provided by the user into text, classify the converted text data, process the text data provided by the user using the Transformer model, and perform part-of-speech tagging on each text data. Suppose the tag of each text data is In the mark, A represents each text data, and c represents the part of speech corresponding to the text data; the part of speech set of c is In the set, in represents interrogative words, ad represents adjectives, ve represents verbs, and no represents nouns. After marking the part of speech of each text data, Arrange the text data in the order of to form user questions;

[0067] The scene feature determination unit is used to collect the environment description text in the text data to generate scene environment data, and use the voice interaction module to collect the data value of the environment data provided by the user; the data value corresponding to each type of scene environment data is set to , represents the data value of the 1st, 2nd, 3rd, ..., fth environmental data on site, where f is a positive integer; according to the real-time scene provided by the user, the data value of each environmental data in the second dimension of the h characteristic environmental data in the corresponding scene is extracted from the database. , Represents the data values ​​of the 1st, 2nd, 3rd, ..., fth environmental data in the corresponding scene extracted from the database, and calculates the difference between the data values ​​of each environmental data in the field and in the database. The formula is:

[0068]

[0069] In the formula, C_h represents the difference between the data value of each environmental data on site and in the database, S_h represents the data value corresponding to each environmental data on site, and Z_h represents the data value of each environmental data in the corresponding scene extracted from the database; let the data difference of each environmental data finally calculated be , represents the calculated data difference of the 1st, 2nd, 3rd, ..., fth environmental data;

[0070] The environmental data is screened according to the calculated difference of each environmental data, the formula is:

[0071]

[0072] In the inequality, T represents the filtered environmental data, C_h represents the difference between the data value of each environmental data on site and in the database, f represents the number of types of environmental data, and h represents the number of characteristic environmental types in the second dimension. When the difference between the data value of environmental data on site and in the database is greater than the average value, the corresponding type of environmental data is filtered, and the final filtered environmental data types are , It is represented as the 1st, 2nd, 3rd, ..., dth environmental data after screening, and the dth environmental data after screening are used as the field characteristics.

[0073] The real-time scene building module is used to build a real-time virtual environment using real-time scene features;

[0074] The real-time scene establishment module is used to analyze the environmental data provided by the user, extract the specific data of each type of environmental data in the environmental data provided by the user, correspond them with each other in the system according to the environmental data recorded in the database, use the environmental data in the database to supplement the environmental data not provided by the user, and use special site features to build a three-dimensional virtual environment model.

[0075] The similarity calculation module is used to calculate the similarity between the real-time virtual environment and each environmental data in the database, and determine the risk category of the real-time environment;

[0076] The similarity calculation module is used to collect the environmental data provided by the user, select the environmental category of each type of coal mine safety knowledge in the second dimension under the real-time scene type of the first dimension in the database, extract the environmental data of each type of coal mine safety knowledge in the second dimension from the database, and extract the number of environmental data of the same type between the real-time scene and the environmental data of each type of coal mine safety knowledge in the database. , It represents the number of environmental data of the same type as the environmental data of the first, second, third, ..., h types of coal mine safety knowledge in the real-time on-site environmental data, and extracts the data values ​​of the same type of environmental data, and calculates the similarity between the real-time on-site environment and the environment of each type of coal mine safety knowledge in the second dimension. The formula is:

[0077]

[0078] In the formula, sim represents the similarity between the real-time on-site environment and the environment of each type of coal mine safety knowledge in the second dimension, F represents the number of environmental data of the same type in the real-time on-site and the environmental data of each type of coal mine safety knowledge in the database, S_h represents the real-time on-site data value of the same type of environmental data, and Z_h represents the data value of the same type of environmental data in the second dimension;

[0079] After calculating the similarities of h kinds of coal mine safety knowledge with different environmental characteristics in the same scenario under the first dimension, the h similarities are obtained as follows: , The similarity of coal mine safety knowledge of the 1st, 2nd, 3rd, ..., hth environmental features is represented, and the environmental feature with the greatest similarity is selected as the real-time environmental risk category.

[0080] The plan formulation module is used to search for corresponding emergency response plans in the security database and optimize them according to user problems and real-time environmental risk categories;

[0081] The program formulation module is used to extract the coal mine safety knowledge corresponding to the real-time environmental risk category in the second dimension that solves the user problem according to the user problem judged by the system, the first dimension scenario and the real-time environmental risk category in the second dimension, analyze the extracted coal mine safety knowledge, and extract the change value G_h of the environmental data caused by the operation steps in the coal mine safety knowledge; optimize the coal mine safety knowledge according to the calculated field characteristics, and the formula is:

[0082]

[0083] In the formula, G_op represents the change value of the operating steps in the coal mine safety knowledge for the environmental data after optimization, G_h represents the change value of the operating steps in the coal mine safety knowledge for the environmental data before optimization, Q_h represents the original data value of the environmental data type changed by the operating steps in the coal mine safety knowledge before optimization, and T_h represents the environmental data value of the field characteristics;

[0084] According to the site characteristics, the change values ​​of the types of environmental data corresponding to the d types of site characteristics in the coal mine safety knowledge are optimized and adjusted, and finally the optimized change values ​​of the types of environmental data corresponding to the d types of site characteristics are obtained as follows: , It represents the optimized change value of the environmental data type corresponding to the 1st, 2nd, 3rd, ..., dth field characteristics; and the extracted coal mine safety knowledge is optimized using the optimized change value.

[0085] Because the real-time environment is unlikely to be exactly the same, in order to be closer to the real-time environmental data and ensure the availability of coal mine safety knowledge, the coal mine safety knowledge is optimized according to the actual environmental data to avoid the occurrence of dangerous accidents caused by inadequate processing; this increases the extensive use of coal mine safety knowledge;

[0086] The display module is used to visually display and voice output the solution finally obtained to solve the user's problem.

[0087] The display module is used to visualize the optimized coal mine safety knowledge and mark and display the changed values ​​of each environmental data after optimization; and to voice output the optimized coal mine safety knowledge to guide users to complete operations in sequence according to the prompts of the coal mine safety knowledge and solve user problems; when the system finishes voice output of the coal mine safety knowledge, it uses the voice interaction module to ask the user whether the problem has been solved. The system makes a judgment based on the user's answer. When the user answers "yes", the task is completed. When the user answers "no", the steps are continued to screen the coal mine safety knowledge.

[0088] Example 1

[0089] The coal mine safety knowledge is collected through the Internet, and classified into disaster and accident handling, risk prevention, and safe operation according to different usage scenarios.

[0090] Collect environmental data words of coal mine safety knowledge in each scenario, calculate the proportion of all environmental data words, and classify the coal mine safety knowledge in each scenario in the second dimension according to environmental characteristics. Disaster accident handling categories are divided into: gas accidents, roof accidents, electromechanical accidents, blasting accidents, flood accidents, fire accidents, transportation accidents, and other accidents; similarly, classify the coal mine safety knowledge in the other two scenarios in the second dimension according to environmental characteristics; use the coal mine safety knowledge classification to build a database;

[0091] The voice interaction module is used to collect user questions. Based on the analysis of the user's voice, the user's question is "How to safely handle gas leakage?" After analysis, it belongs to the accident handling category in the first dimension. Then, the on-site environmental information is collected from the user, and the real-time on-site characteristics are calculated. The formula is:

[0092]

[0093]

[0094] After comparing and calculating with the environmental data of coal mine safety knowledge in the database, the on-site characteristics are obtained as temperature and air velocity;

[0095] Collect the real-time environmental data provided by users, and calculate the similarity between the real-time environmental data and the coal mine safety knowledge environment category in the second dimension. The formula is:

[0096]

[0097] After calculation, the real-time environmental risk category is the gas accident in the second dimension;

[0098] The change in environmental data caused by the operation steps in the coal mine safety knowledge collection is -10 degrees in temperature and 1 in air velocity, and the original temperature in the coal mine safety knowledge is 25 and the original air velocity is 0.7; the coal mine safety knowledge is optimized, and the temperature in the field characteristics is 35 degrees and the air velocity is 0.01; the formula is:

[0099]

[0100] The optimized temperature change value is calculated to be -20; the optimized air flow rate is 1.69;

[0101] The optimized coal mine safety knowledge is displayed and voice output is performed.

[0102] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0103] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A coal mine safety knowledge interaction system based on a natural language model, characterized by: The coal mine safety knowledge interactive system includes a safety database building module, a voice interaction module, a voice analysis module, a real-time scenario building module, a similarity calculation module, a plan formulation module and a display module; The safety database construction module is used to collect and summarize historical coal mine safety knowledge, and construct a coal mine safety knowledge database after classification; The security database construction module includes a data collection unit and a database construction unit; The data collection unit is used to collect coal mine safety knowledge in the network. Suppose the collected coal mine safety knowledge is {Z1, Z2, Z3, ..., Z n }, Z1, Z2, Z3, ..., Z n represents the 1st, 2nd, 3rd, ..., nth coal mine safety knowledge collected, where n is a positive integer; extract application scenario vocabulary from the collected coal mine safety knowledge, and classify the collected coal mine safety knowledge in the first dimension according to different application scenario vocabulary, and obtain the first dimension category of coal mine safety knowledge as {Y1, Y2, Y3, ..., Y m }, Y1, Y2, Y3, ..., Y m represents the coal mine safety knowledge of the first, second, third, ..., m application scenarios, where m is a positive integer; After classifying the coal mine safety knowledge in the first dimension, the environmental phrases describing the environmental data in the coal mine safety knowledge in each application scenario of the first dimension are extracted, and the number of each environmental phrase is counted as {S1, S2, S3, ..., S j }, S1, S2, S3, ..., S j represents the number of the 1st, 2nd, 3rd, ..., jth environmental phrases in coal mine safety knowledge, where j is a positive integer. The proportion of each environmental phrase in the total environmental phrases is calculated using the formula: In the formula, a represents the proportion of each environmental phrase in the total environmental phrases, S i represents the number of each environmental phrase, i∈(1,j); j represents the number of types of environmental phrases; the proportion of j types of environmental phrases is {a1, a2, a3, ..., a j }, a1, a2, a3, ..., a j represents the proportion of the 1st, 2nd, 3rd, ..., jth environmental phrases; the maximum proportion of the jth environmental phrase is selected as the environmental feature of the corresponding coal mine safety knowledge; the coal mine safety knowledge is classified into the second dimension according to the environmental feature of each coal mine safety knowledge, and the second dimension categories of coal mine safety knowledge are obtained as {H1, H2, H3, ..., H h }, H1, H2, H3, ..., H h represents the 1st, 2nd, 3rd, ..., hth type of coal mine safety knowledge in the second dimension, where h is a positive integer; The voice interaction module is used to conduct safety knowledge Q&A and risk warning voice broadcast with the user, analyze user needs, and guide the user to provide the four-factor characteristic data of human, machine, environment, and management safety of the operation process required by the system; The speech analysis module is used to analyze the user speech signals collected by the system, extract keywords, generate user questions and real-time on-site features; The speech analysis module includes a question generation unit and a scene feature determination unit; The question generation unit is used to convert the voice information provided by the user into text, classify the converted text data, process the text data provided by the user using the Transformer model, and perform part-of-speech tagging on each text data. Suppose the tag of each text data is A c In the tag, A represents each text data, and c represents the part of speech corresponding to the text data; the part of speech set of c is {in, ad, ve, no}, in which in represents a question word, ad represents an adjective, ve represents a verb, and no represents a noun; after marking the part of speech of each text data, the text data are arranged in the order of in→ad→ve→no to form a user question; The scene feature determination unit is used to collect the environment description text in the text data to generate scene environment data, and use the voice interaction module to collect the data value of the environment data provided by the user; suppose the data value corresponding to each type of scene environment data is {S_h1, S_h2, S_h3, ..., S_h f }, S_h1, S_h2, S_h3,..., S_h f represents the data value of the 1st, 2nd, 3rd, ..., fth environmental data on site, where f is a positive integer; according to the real-time scene provided by the user, the data value of each environmental data in the second dimension of the h characteristic environmental data in the corresponding scene is extracted from the database {Z_h1, Z_h2, Z_h3, ..., Z_h f }, Z_h1, Z_h2, Z_h3,..., Z_h f Represents the data values ​​of the 1st, 2nd, 3rd, ..., fth environmental data in the corresponding scene extracted from the database, and calculates the difference between the data values ​​of each environmental data in the field and in the database. The formula is: C_h=|S_h-Z_h| In the formula, C_h represents the difference between the data value of each environmental data on site and in the database, S_h represents the data value corresponding to each environmental data on site, and Z_h represents the data value of each environmental data in the corresponding scene extracted from the database; suppose the data difference of each environmental data finally calculated is {C_h1, C_h2, C_h3, ..., C_h f }, C_h1, C_h2, C_h3,..., C_h f represents the calculated data difference of the 1st, 2nd, 3rd, ..., fth environmental data; The environmental data is screened according to the calculated difference of each environmental data, the formula is: In the inequality, T represents the filtered environmental data, C_h represents the difference between the data value of each environmental data on site and in the database, f represents the number of types of environmental data, and h is the number of characteristic environmental types in the second dimension; when the difference between the data value of the environmental data on site and in the database is greater than the average value, the corresponding type of environmental data is filtered, and the final filtered environmental data types are {T1, T2, T3, ..., T d }, T1, T2, T3, ..., T d It is represented as the first, second, third, ..., d types of environmental data after screening, and the d types of environmental data after screening are used as the field characteristics; The real-time scene building module is used to build a real-time virtual environment using real-time scene features; The similarity calculation module is used to calculate the similarity between the real-time virtual environment and each environmental data in the database, and determine the risk category of the real-time environment; The similarity calculation module is used to collect the environmental data provided by the user, select the environmental category of each type of coal mine safety knowledge in the second dimension under the real-time scene type of the first dimension in the database, extract the environmental data of each type of coal mine safety knowledge in the second dimension from the database, and extract the number of environmental data of the same type between the real-time scene and the environmental data of each type of coal mine safety knowledge in the database as {F1, F2, F3, ..., F h }, F1, F2, F3, ..., F h It represents the number of environmental data of the same type as the environmental data of the first, second, third, ..., h types of coal mine safety knowledge in the real-time on-site environmental data, and extracts the data values ​​of the same type of environmental data, and calculates the similarity between the real-time on-site environment and the environment of each type of coal mine safety knowledge in the second dimension. The formula is: sim=F×(S_h-Z_h) -1 In the formula, sim represents the similarity between the real-time on-site environment and the environment of each type of coal mine safety knowledge in the second dimension, F represents the number of environmental data of the same type in the real-time on-site and the environmental data of each type of coal mine safety knowledge in the database, S_h represents the real-time on-site data value of the same type of environmental data, and Z_h represents the data value of the same type of environmental data in the second dimension; After calculating the similarities of h kinds of coal mine safety knowledge with different environmental characteristics in the same scene under the first dimension, we get h similarities {sim1, sim2, sim3, ..., sim h }, sim1, sim2, sim3,..., sim h Represent the similarity of coal mine safety knowledge of the 1st, 2nd, 3rd, ..., hth environmental features, and select the environmental feature with the greatest similarity as the real-time environmental risk category; The plan formulation module is used to search for corresponding emergency response plans in the security database and optimize them according to user problems and real-time environmental risk categories; The program formulation module is used to extract the coal mine safety knowledge corresponding to the real-time environmental risk category in the second dimension that solves the user problem according to the user problem determined by the system, the first dimension scenario and the real-time environmental risk category in the second dimension, analyze the extracted coal mine safety knowledge, and extract the change value G_h of the environmental data caused by the operation steps in the coal mine safety knowledge; optimize the coal mine safety knowledge according to the calculated field characteristics, and the formula is: G_op=(G_h+Q_h)−T_h In the formula, G_op represents the change value of the operating steps in the coal mine safety knowledge for the environmental data after optimization, G_h represents the change value of the operating steps in the coal mine safety knowledge for the environmental data before optimization, Q_h represents the original data value of the environmental data type changed by the operating steps in the coal mine safety knowledge before optimization, and T_h represents the environmental data value of the field characteristics; According to the on-site characteristics, the change values ​​of the types of environmental data corresponding to the d on-site characteristics in the coal mine safety knowledge are optimized and adjusted, and finally the optimized change values ​​of the types of environmental data corresponding to the d on-site characteristics are obtained as {G_op1, G_op2, G_op3, ..., G_op d }, G_op1, G_op2, G_op3,..., G_op d Indicates the optimized change values ​​of the first, second, third, ..., d types of field characteristics corresponding to the types of environmental data; and optimizes the extracted coal mine safety knowledge using the optimized change values; The display module is used to visually display and voice output the solution finally obtained to solve the user's problem.

2. A coal mine safety knowledge interaction system based on a natural language model according to claim 1, characterized in that: The voice interaction module is used to collect the voice data sent by the user, and convert the voice data sent by the user into text data using the voice recognition system. Suppose the text data after the user's voice data is converted is {W1, W2, W3, ..., W g }, W1, W2, W3, ..., W g represents the 1st, 2nd, 3rd, ..., gth text data after the user's voice data is converted, where g is a positive integer; The converted text data is analyzed, scene words in the text data are extracted, and the extracted scene words are compared with the scene characteristics of the first dimension in the coal mine safety knowledge database. When the scene Y_s described by the user is found in the coal mine safety knowledge database, the system issues a voice prompt to ask the user for real-time environmental data on site; After analyzing the converted text data, if no scenario vocabulary is extracted or no real-time scenario vocabulary is found in the coal mine safety knowledge database, the system displays the classified scenario vocabulary of the first dimension of coal mine safety knowledge in the database for the user to choose, and finally obtains the real-time scenario required by the user.

3. The coal mine safety knowledge interaction system based on a natural language model according to claim 1 is characterized in that: The real-time scene establishment module is used to analyze the environmental data provided by the user, extract the specific data of each type of environmental data in the environmental data provided by the user, correspond to each other in the system according to the environmental data recorded in the database, use the environmental data in the database to supplement the environmental data not provided by the user, and use the on-site characteristics to build a three-dimensional virtual environment model.

4. The coal mine safety knowledge interaction system based on a natural language model according to claim 1 is characterized in that: The display module is used to visualize the optimized coal mine safety knowledge and mark and display the changed values ​​of each optimized environmental data; The optimized coal mine safety knowledge is output in voice, guiding users to complete operations in sequence according to the coal mine safety knowledge prompts to solve user problems; After the system finishes outputting the coal mine safety knowledge voice, it uses the voice interaction module to ask the user whether the problem has been solved. The system makes a judgment based on the user's answer. When the user answers "yes", the task is completed. When the user answers "no", the steps are continued to screen the coal mine safety knowledge.

Images