Flight simulation teaching and training risk analysis system and method based on big data

Abstract

The present invention discloses a flight simulation teaching and training risk analysis system and method based on big data, and relates to the field of data analysis technology. The system of the present invention includes: a data acquisition and sorting module, a risk point identification and analysis module, a risk propagation path analysis module, a real-time risk monitoring module and anomaly detection module, and a training effect evaluation and feedback module; the data acquisition and sorting module extracts historical flight data and flight simulation teaching data, analyzes the two to establish a mapping relationship, and constructs a historical flight database; the risk point identification and analysis module is used to identify risk points, extract features to construct a risk point data set; the risk propagation path analysis module identifies the propagation path and calculates the path weight; the real-time risk monitoring module acquires real-time data, identifies real-time risk points, and extracts features to construct a propagation path; the training effect evaluation and feedback module counts the number of real-time risk points, evaluates the training effect, and outputs prompt information.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a flight simulation teaching and training risk analysis system and method based on big data. Background Art

[0002] With the rapid development of the aviation industry, global air traffic has increased significantly, and the skill requirements for pilots are increasing day by day. In this context, pilot training and education are particularly important, especially in improving flight safety and reducing accident risks. Although traditional flight training methods guarantee the basic skills of pilots to a certain extent, they are limited by weather, geographical location and training resources, making it difficult to fully simulate complex flight environments and the ability to respond to emergencies. Therefore, finding a more efficient, safe and economical flight training method has become an urgent need in the industry.

[0003] In recent years, the rapid development of flight simulation technology has provided new solutions for pilot training. Flight simulators can simulate a variety of flight scenarios in a safe environment, allowing pilots to train in the face of different flight conditions and emergency situations, thereby improving their operational skills and decision-making capabilities. However, the existing flight simulation system still has shortcomings in risk analysis. Specifically, although the existing flight simulation system can quantitatively evaluate risk points by analyzing safety risk information in flight training, it lacks in-depth analysis of the possible transmission paths of risk points, which makes it impossible for pilots to truly understand the root causes of potential risks, thereby affecting the effectiveness of flight simulation teaching. Summary of the invention

[0004] The purpose of the present invention is to provide a flight simulation teaching and training risk analysis system and method based on big data to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] A flight simulation teaching and training risk analysis method based on big data, the method comprises the following steps:

[0007] Step S100. Obtain historical flight data and flight simulation teaching data from a database, analyze the historical flight data and the flight simulation teaching data, thereby obtaining a mapping relationship between the two; and organize the historical flight data based on the mapping relationship, thereby constructing a historical flight database;

[0008] Step S200. According to the historical flight database, the historical flight data in the historical flight database is compared and analyzed with the flight simulation teaching data, so as to identify the risk points in the historical flight database, and form a risk point data set; the risk point data set is analyzed, the relevant features of the risk points are extracted, and the feature vector of the risk points is generated;

[0009] Step S300. Based on the feature vectors of the risk points, the risk points in the risk point data set are analyzed accordingly, so as to obtain the correlation between different risk points; according to the correlation between different risk points, the propagation path of each risk point is obtained;

[0010] Step S400. Acquire the pilot's flight simulation training data in real time, analyze the deviation relationship between the flight simulation training data and the flight simulation teaching data, and identify the real-time risk point based on the deviation relationship; analyze the flight simulation training data to obtain the real-time feature vector of the real-time risk point, and obtain the corresponding real-time propagation path according to the real-time feature vector;

[0011] Step S500. For each pilot, obtain the flight simulation training data of the last two consecutive time periods, evaluate the pilot's flight simulation training effect according to the corresponding real-time risk points and real-time propagation paths, and output corresponding prompt information according to the flight simulation training effect.

[0012] Furthermore, step S100 includes:

[0013] S101. Obtain historical flight data and flight simulation teaching data from a database, and divide the historical flight data and flight simulation teaching data according to a preset flight mission start mark and a flight mission end mark, so that the historical flight data and the flight simulation teaching data correspond to a number of flight missions, respectively, thereby forming a historical flight data set Dh and a flight simulation teaching data set Ds, and Dh={h1,h2,...,hn}, Ds={s1,s2,...,sk}, wherein h1 represents the historical flight data corresponding to the first historical flight mission, h2 represents the historical flight data corresponding to the second historical flight mission, and so on, hn represents the historical flight data corresponding to the nth historical flight mission; similarly, s1 represents The flight simulation teaching data corresponding to the first flight simulation teaching mission, s2 represents the flight simulation teaching data corresponding to the second flight simulation teaching mission, sk represents the flight simulation teaching data corresponding to the kth flight simulation teaching mission; n represents the total number of historical flight missions, k represents the total number of flight missions of flight simulation teaching, and n>k; the historical flight data refers to various flight parameters and information collected during the actual flight process, and these data generally include the speed, altitude, heading, fuel consumption, meteorological conditions, etc. of the aircraft; the flight simulation teaching data refers to data generated or used in a flight simulator, and these data specifically include the flight parameters and information of the simulated aircraft when performing a flight mission;

[0014] S102. Extract preset key features from the historical flight data set Dh and the flight simulation teaching data set Ds, and common features include: aircraft speed V, flight altitude H, heading D, fuel consumption F, and meteorological conditions M, etc.; thereby, for each flight mission in the historical flight data set Dh and the flight simulation teaching data set Ds, corresponding key feature vectors Xh_i and Xs_j are respectively constructed, where i is a positive integer from 1 to n, and j is a positive integer from 1 to k; use a linear regression model to analyze the mapping relationship between Xh_i and Xs_j, and the linear regression model expression is:

[0015] Xs_j=A·Xh_i+b,

[0016] Where A represents the weight matrix, and b represents the bias term. Based on the linear regression model, the mapping relationship between each flight mission in the flight simulation teaching data set Ds and the flight mission in the historical flight data set Dh is obtained. According to the mapping relationship between the two, the historical flight data is sorted to construct a historical flight database.

[0017] Historical flight data and flight simulation teaching data are extracted from the database, and the relationship between the two is analyzed to build a historical flight database. The purpose of this step is to reduce the deviation between actual flight data and simulation data to improve the accuracy of training.

[0018] Further, step S200 includes:

[0019] S201. For each historical flight mission in the historical flight database, the corresponding historical flight data is obtained, and the similarity calculation is performed on the historical flight data and the flight simulation teaching data corresponding to the flight mission in the flight simulation teaching data set in turn, and the obtained similarity result S is compared with the corresponding similarity threshold S0; the flight simulation teaching data satisfying S≥S0 is used as the template data; the deviation value D between the template data Mb and the historical flight data Hf is calculated, and the calculation formula of the deviation value D is:

[0020] D={Σc∈[1,C],|Mb_c-Hf|} / C,

[0021] Wherein, Mb_c represents the cth template data, and C represents the number of flight missions in the flight simulation teaching data set corresponding to the template data;

[0022] S202. When the deviation value D ≥ the deviation threshold D0, the historical flight data and the template data are subjected to trend analysis to obtain a historical flight data trend curve A and a template data trend curve B respectively; the historical flight data trend curve A and the template data trend curve B are expressed in the same plane rectangular coordinate system, the difference T at the same time point is calculated in turn, all the difference T are summarized and the average value Tμ is calculated, and the time point when the difference T ≥ Tμ is marked as the risk point of the historical flight data; for each historical flight mission, the risk points in all historical flight data are summarized to form a risk point data set F, and F = {f1, f2, ...., ft}, where f1 represents the first risk point, f2 represents the second risk point, and so on, ft represents the tth risk point; the historical flight data corresponding to the risk point is obtained, and the preset key features are extracted from the historical flight data to generate a feature vector of the risk point.

[0023] Furthermore, step S300 includes:

[0024] S301. For each risk point dataset F of historical flight missions, extract the feature vector corresponding to the risk point, calculate the cosine similarity S1 between different risk points in turn, and construct a similarity matrix Q, and the element q in the similarity matrix Q uvrepresents the cosine similarity between risk points fu and fv; for each risk point similarity matrix Q, select the elements greater than or equal to the threshold S2, and construct the corresponding directed graph G(V,E), where the node V represents the risk point, the edge E represents the propagation relationship, and the weight w of the edge is represented by the value of the element greater than or equal to the threshold S2 selected;

[0025] S302. Based on the constructed directed graph G(V,E), all propagation paths from the risk point fu to fv are identified through the graph traversal algorithm, and the corresponding path weight W(P_uv) is calculated, and W(P_uv) is equal to the product of all node edges in the propagation path. The paths are sorted in descending order according to the path weight W(P_uv), and the path with the largest path weight W(P_uv) is selected as the propagation path for the corresponding risk point.

[0026] Furthermore, step S400 includes:

[0027] S401. Acquire the pilot's flight simulation training data in real time, calculate the similarity between the flight simulation training data and the flight simulation teaching data corresponding to the flight mission in the flight simulation teaching data set, and compare the obtained similarity result S' with the corresponding similarity threshold S0;

[0028] The flight simulation teaching data satisfying S'≥S0 is used as the first control data; the deviation value D' between the first control data and the flight simulation training data is calculated, and the calculation formula of the deviation value D' refers to the calculation formula of the deviation value D; and the flight simulation training data with the deviation value D'≥deviation threshold D0 and the first control data are subjected to trend analysis, and the trend analysis process of historical flight data and template data is referred to, so as to identify the real-time risk points in the flight simulation training data, and obtain the real-time feature vector and the corresponding real-time propagation path of the real-time risk points according to the corresponding flight simulation training data;

[0029] S402. Compare the similarity result S' with the corresponding similarity threshold S0. If the similarity result S' of the flight simulation training data of the corresponding real-time flight simulation training task is less than the similarity threshold S0, then perform similarity calculation on the flight simulation training data and the historical flight data in the historical flight database, select the historical flight data with the greatest similarity as the second control data, sort the flight simulation training data according to the mapping relationship between the second control data and the flight simulation teaching data, and perform step S401 analysis on the sorted flight simulation training data and the flight simulation teaching data, so as to identify the real-time risk points in the flight simulation training data, as well as the real-time feature vectors of the real-time risk points and the corresponding real-time propagation paths.

[0030] Furthermore, step S500 includes:

[0031] S501. For each pilot, obtain the flight simulation training data of the last two consecutive time periods, and obtain the corresponding real-time risk point data set F1 and real-time risk point data set F2 according to the flight simulation training data; count the number m1 and m2 of real-time risk points in the real-time risk point data set F1 and the real-time risk point data set F2 respectively, and construct a corresponding directed graph for each real-time risk point, calculate the path weight W of the real-time propagation path of the real-time risk point according to the directed graph, evaluate the flight simulation training effect of the pilot according to the above data, and calculate the corresponding training effect index R, the specific calculation formula is:

[0032] R={Σe1∈[1,m1],We1} / {Σe2∈[1,m2],We2},

[0033] Wherein, e1 represents the number of the real-time risk point in the real-time risk point data set F1, e2 represents the number of the real-time risk point in the real-time risk point data set F2, We1 represents the path weight of the real-time propagation path of the real-time risk point in the real-time risk point data set F1, and We2 represents the path weight of the real-time propagation path of the real-time risk point in the real-time risk point data set F2;

[0034] S502. If the training effect index R>1, it means that the pilot's training effect is good, and the prompt message "Congratulations! Your training effect is better than last time, please keep it up" is output; if the training effect index R=1, and m1>m2, it means that the pilot's training effect is average, and the prompt message "New risk points have been added, but the handling effect is equivalent, please consider strengthening the training of new risk points" is output; if the training effect index R<1, it means that the pilot's training effect is not good, and the prompt message "The recent training effect is not good, please evaluate the training content as soon as possible and formulate improvement measures" is output; for other values ​​of the training effect index R, the prompt message "Please re-evaluate the training strategy" is output.

[0035] The flight simulation teaching and training risk analysis system based on big data includes: data acquisition and sorting module, risk point identification and analysis module, risk propagation path analysis module, real-time risk monitoring module and training effect evaluation and feedback module;

[0036] The data acquisition and arrangement module extracts historical flight data and flight simulation teaching data from the database, divides the historical flight data and the flight simulation teaching data to form a historical flight data set and a flight simulation teaching data set, analyzes the historical flight data set and the flight simulation teaching data set to obtain a mapping relationship between the historical flight data and the flight simulation teaching data, and arranges the historical flight data to construct a historical flight database;

[0037] The risk point identification and analysis module calculates the similarity between each historical flight mission in the historical flight database and the flight simulation teaching data, and compares it with the similarity threshold to filter out the template data; calculates the deviation value between the template data and the historical flight data, and performs trend analysis to identify risk points; extracts relevant features from the identified risk points, constructs risk point feature vectors, and forms a risk point data set;

[0038] The risk propagation path analysis module calculates the cosine similarity between risk points based on the risk point feature vectors, constructs a similarity matrix, screens risk points based on the similarity matrix, and constructs a directed graph; uses a graph traversal algorithm to identify the propagation paths between risk points, calculates the path weights, and selects the path with the largest weight as the propagation path for the corresponding risk point;

[0039] The real-time risk monitoring module obtains the pilot's flight simulation training data in real time, and calculates the similarity with the flight simulation teaching data. It identifies the real-time risk points based on the similarity calculation results, extracts the feature vectors of the real-time risk points, and constructs the real-time propagation path.

[0040] The training effect evaluation and feedback module counts the number of risk points in the real-time risk point data set for each pilot and calculates the corresponding propagation path weight; calculates the training effect index according to the path weight of the real-time risk point, and outputs corresponding prompt information according to different values ​​of the training effect index.

[0041] Furthermore, the data acquisition and arrangement module includes a data acquisition unit, a mapping relationship establishment unit and a data arrangement unit;

[0042] The data acquisition unit extracts historical flight data and flight simulation teaching data from the database, and divides the historical flight data and flight simulation teaching data into multiple flight missions according to the preset flight mission start and end marks to form historical flight data sets and flight simulation teaching data sets; the mapping relationship establishment unit extracts key features and constructs key feature vectors, and uses a linear regression model to analyze the mapping relationship between historical flight data and flight simulation teaching data; the data sorting unit sorts the historical flight data based on the mapping relationship between the historical flight data and the flight simulation teaching data to construct a historical flight database.

[0043] Furthermore, the risk point identification and analysis module includes a similarity calculation unit, a deviation analysis unit, and a risk point feature extraction unit;

[0044] The similarity calculation unit calculates the similarity with the flight simulation teaching data for each historical flight mission in the historical flight database, and compares it with the similarity threshold to screen out the template data; the deviation analysis unit calculates the deviation value between the template data and the historical flight data, and performs trend analysis to identify risk points; the risk point feature extraction unit extracts relevant features from the identified risk points, constructs risk point feature vectors, and forms a risk point data set;

[0045] The risk propagation path analysis module includes a similarity matrix construction unit, a propagation relationship identification unit, and a propagation path calculation unit;

[0046] The similarity matrix construction unit calculates the cosine similarity between risk points based on the risk point feature vectors and constructs a similarity matrix; the propagation relationship identification unit screens out risk points with high similarity based on the similarity matrix and constructs a directed graph; the propagation path calculation unit uses a graph traversal algorithm to identify the propagation path between risk points, calculates the path weight and selects the path with the largest weight as the corresponding propagation path.

[0047] Further, the real-time risk monitoring module includes a real-time data acquisition unit, a real-time deviation analysis unit, and a real-time feature vector generation unit;

[0048] The real-time data acquisition unit acquires the pilot's flight simulation training data in real time and performs similarity calculation with the flight simulation teaching data; the real-time deviation analysis unit analyzes the corresponding deviation value according to the similarity calculation result, and performs trend analysis to identify the real-time risk point; the real-time feature vector generation unit extracts the feature vector of the real-time risk point and constructs the real-time propagation path;

[0049] The training effect evaluation and feedback module includes a real-time risk point statistics unit, a training effect index calculation unit, and a feedback prompt information unit;

[0050] The real-time risk point statistics unit counts the number of real-time risk points in the real-time risk point data set for each pilot, and calculates the corresponding real-time propagation path weight; the training effect index calculation unit calculates the training effect index according to the path weight of the real-time risk point; the feedback prompt information unit outputs corresponding prompt information according to different values ​​of the training effect index.

[0051] Compared with the prior art, the present invention has the following beneficial effects: by analyzing the mapping relationship between historical flight data and flight simulation teaching data, not only can potential risk points be identified, but also the characteristic vectors of risk points can be extracted, and the correlation between different risk points can be analyzed. This comprehensive risk identification mechanism enables pilots to have a deeper understanding and cope with potential risks during training; the existing flight simulation system often only stays at the quantitative evaluation of risk points, while the present invention can deeply analyze the possible propagation paths of risk points by constructing directed graphs and propagation paths between risk points. This mechanism enables pilots to better understand the root causes and propagation methods of risks, thereby improving their decision-making ability in complex situations; by acquiring the pilot's training data in real time and analyzing its deviation from historical data, the present invention can identify risk points and propagation paths in real time. This function ensures that pilots can obtain feedback in time during training, adjust training strategies, and thus improve their ability to cope with emergencies; the present invention designs a training effect index evaluation mechanism for each pilot. By comparing the real-time risk point data sets of different time periods, training feedback can be tailored for each pilot, and specific improvement suggestions can be provided, thereby improving the pertinence and effectiveness of training. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] 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:

[0053] Figure 1 It is a schematic diagram of the module of the flight simulation teaching and training risk analysis system based on big data of the present invention. DETAILED DESCRIPTION

[0054] 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.

[0055] See also Figure 1 , the present invention provides a technical solution:

[0056] The flight simulation teaching and training risk analysis system based on big data includes: data acquisition and sorting module, risk point identification and analysis module, risk propagation path analysis module, real-time risk monitoring module and training effect evaluation and feedback module;

[0057] The data acquisition and arrangement module extracts historical flight data and flight simulation teaching data from the database, divides the historical flight data and the flight simulation teaching data to form a historical flight data set and a flight simulation teaching data set, analyzes the historical flight data set and the flight simulation teaching data set to obtain a mapping relationship between the historical flight data and the flight simulation teaching data, and arranges the historical flight data to construct a historical flight database;

[0058] The risk point identification and analysis module calculates the similarity between each historical flight mission in the historical flight database and the flight simulation teaching data, and compares it with the similarity threshold to filter out the template data; calculates the deviation value between the template data and the historical flight data, and performs trend analysis to identify risk points; extracts relevant features from the identified risk points, constructs risk point feature vectors, and forms a risk point data set;

[0059] The risk propagation path analysis module calculates the cosine similarity between risk points based on the risk point feature vectors, constructs a similarity matrix, screens risk points based on the similarity matrix, and constructs a directed graph; uses a graph traversal algorithm to identify the propagation paths between risk points, calculates the path weights, and selects the path with the largest weight as the propagation path for the corresponding risk point;

[0060] The real-time risk monitoring module obtains the pilot's flight simulation training data in real time, and calculates the similarity with the flight simulation teaching data. It identifies the real-time risk points based on the similarity calculation results, extracts the feature vectors of the real-time risk points, and constructs the real-time propagation path.

[0061] The training effect evaluation and feedback module counts the number of risk points in the real-time risk point data set for each pilot and calculates the corresponding propagation path weight; calculates the training effect index according to the path weight of the real-time risk point, and outputs corresponding prompt information according to different values ​​of the training effect index.

[0062] The data acquisition and sorting module includes a data acquisition unit, a mapping relationship establishment unit and a data sorting unit;

[0063] The data acquisition unit extracts historical flight data and flight simulation teaching data from the database, and divides the historical flight data and flight simulation teaching data into multiple flight missions according to the preset flight mission start and end marks to form historical flight data sets and flight simulation teaching data sets; the mapping relationship establishment unit extracts key features and constructs key feature vectors, and uses a linear regression model to analyze the mapping relationship between historical flight data and flight simulation teaching data; the data sorting unit sorts the historical flight data based on the mapping relationship between the historical flight data and the flight simulation teaching data to construct a historical flight database.

[0064] The risk point identification and analysis module includes a similarity calculation unit, a deviation analysis unit, and a risk point feature extraction unit;

[0065] The similarity calculation unit calculates the similarity with the flight simulation teaching data for each historical flight mission in the historical flight database, and compares it with the similarity threshold to screen out the template data; the deviation analysis unit calculates the deviation value between the template data and the historical flight data, and performs trend analysis to identify risk points; the risk point feature extraction unit extracts relevant features from the identified risk points, constructs risk point feature vectors, and forms a risk point data set;

[0066] The risk propagation path analysis module includes a similarity matrix construction unit, a propagation relationship identification unit, and a propagation path calculation unit;

[0067] The similarity matrix construction unit calculates the cosine similarity between risk points based on the risk point feature vectors and constructs a similarity matrix; the propagation relationship identification unit screens out risk points with high similarity based on the similarity matrix and constructs a directed graph; the propagation path calculation unit uses a graph traversal algorithm to identify the propagation path between risk points, calculates the path weight and selects the path with the largest weight as the corresponding propagation path.

[0068] The real-time risk monitoring module includes a real-time data acquisition unit, a real-time deviation analysis unit, and a real-time feature vector generation unit;

[0069] The real-time data acquisition unit acquires the pilot's flight simulation training data in real time and performs similarity calculation with the flight simulation teaching data; the real-time deviation analysis unit analyzes the corresponding deviation value according to the similarity calculation result, and performs trend analysis to identify the real-time risk point; the real-time feature vector generation unit extracts the feature vector of the real-time risk point and constructs the real-time propagation path;

[0070] The training effect evaluation and feedback module includes a real-time risk point statistics unit, a training effect index calculation unit, and a feedback prompt information unit;

[0071] The real-time risk point statistics unit counts the number of real-time risk points in the real-time risk point data set for each pilot, and calculates the corresponding real-time propagation path weight; the training effect index calculation unit calculates the training effect index according to the path weight of the real-time risk point; the feedback prompt information unit outputs corresponding prompt information according to different values ​​of the training effect index.

[0072] A flight simulation teaching and training risk analysis method based on big data, the method comprises the following steps:

[0073] Step S100. Obtain historical flight data and flight simulation teaching data from a database, analyze the historical flight data and the flight simulation teaching data, thereby obtaining a mapping relationship between the two; and organize the historical flight data based on the mapping relationship, thereby constructing a historical flight database;

[0074] Step S200. According to the historical flight database, the historical flight data in the historical flight database is compared and analyzed with the flight simulation teaching data, so as to identify the risk points in the historical flight database, and form a risk point data set; the risk point data set is analyzed, the relevant features of the risk points are extracted, and the feature vector of the risk points is generated;

[0075] Step S300. Based on the feature vectors of the risk points, the risk points in the risk point data set are analyzed accordingly, so as to obtain the correlation between different risk points; according to the correlation between different risk points, the propagation path of each risk point is obtained;

[0076] Step S400. Acquire the pilot's flight simulation training data in real time, analyze the deviation relationship between the flight simulation training data and the flight simulation teaching data, and identify the real-time risk point based on the deviation relationship; analyze the flight simulation training data to obtain the real-time feature vector of the real-time risk point, and obtain the corresponding real-time propagation path according to the real-time feature vector;

[0077] Step S500. For each pilot, obtain the flight simulation training data of the last two consecutive time periods, evaluate the pilot's flight simulation training effect according to the corresponding real-time risk points and real-time propagation paths, and output corresponding prompt information according to the flight simulation training effect.

[0078] Step S100 includes:

[0079] S101. Obtain historical flight data and flight simulation teaching data from a database, and divide the historical flight data and flight simulation teaching data according to a preset flight mission start mark and a flight mission end mark, so that the historical flight data and the flight simulation teaching data correspond to a number of flight missions, respectively, thereby forming a historical flight data set Dh and a flight simulation teaching data set Ds, and Dh={h1,h2,...,hn}, Ds={s1,s2,...,sk}, wherein h1 represents the historical flight data corresponding to the first historical flight mission, h2 represents the historical flight data corresponding to the second historical flight mission, and so on, hn represents the historical flight data corresponding to the nth historical flight mission; similarly, s1 represents The flight simulation teaching data corresponding to the first flight simulation teaching mission, s2 represents the flight simulation teaching data corresponding to the second flight simulation teaching mission, sk represents the flight simulation teaching data corresponding to the kth flight simulation teaching mission; n represents the total number of historical flight missions, k represents the total number of flight missions of flight simulation teaching, and n>k; the historical flight data refers to various flight parameters and information collected during the actual flight process, and these data generally include the speed, altitude, heading, fuel consumption, meteorological conditions, etc. of the aircraft; the flight simulation teaching data refers to data generated or used in a flight simulator, and these data specifically include the flight parameters and information of the simulated aircraft when performing a flight mission;

[0080] S102. Extract preset key features from the historical flight data set Dh and the flight simulation teaching data set Ds, and common features include: aircraft speed V, flight altitude H, heading D, fuel consumption F, and meteorological conditions M, etc.; thereby, for each flight mission in the historical flight data set Dh and the flight simulation teaching data set Ds, corresponding key feature vectors Xh_i and Xs_j are respectively constructed, where i is a positive integer from 1 to n, and j is a positive integer from 1 to k; use a linear regression model to analyze the mapping relationship between Xh_i and Xs_j, and the linear regression model expression is:

[0081] Xs_j=A·Xh_i+b,

[0082] Where A represents the weight matrix, and b represents the bias term. Based on the linear regression model, the mapping relationship between each flight mission in the flight simulation teaching data set Ds and the flight mission in the historical flight data set Dh is obtained. According to the mapping relationship between the two, the historical flight data is sorted to construct a historical flight database.

[0083] Historical flight data and flight simulation teaching data are extracted from the database, and the relationship between the two is analyzed to build a historical flight database. The purpose of this step is to reduce the deviation between actual flight data and simulation data to improve the accuracy of training.

[0084] In this embodiment, the parameters A and b in the linear regression model are solved by the least squares method, and the corresponding objective function is:

[0085] Minimize J(A,b)=(1 / k)Σj∈[1,k],[Xs_j-(A·Xh_i+b)]^2,

[0086] Among them, J(A,b) represents the loss function of the model, and the goal is to minimize the difference between the actual flight simulation teaching data Xs_j and the predicted value; by solving the derivative of the above objective function and setting it to zero, an analytical solution can be obtained, which is:

[0087] A=(XhT·Xh)-1·XhT·Xs,

[0088] b=(1 / k)Σj∈[1,k],(Xs_j-A·Xh_i),

[0089] Among them, Xh represents the feature matrix of all flight missions in the historical flight dataset Dh, and Xs represents the feature matrix of all flight missions in the flight simulation teaching dataset Ds.

[0090] Step S200 includes:

[0091] S201. For each historical flight mission in the historical flight database, the corresponding historical flight data is obtained, and the similarity calculation is performed on the historical flight data and the flight simulation teaching data corresponding to the flight mission in the flight simulation teaching data set in turn, and the obtained similarity result S is compared with the corresponding similarity threshold S0; the flight simulation teaching data satisfying S≥S0 is used as the template data; the deviation value D between the template data Mb and the historical flight data Hf is calculated, and the calculation formula of the deviation value D is:

[0092] D={Σc∈[1,C],|Mb_c-Hf|} / C,

[0093] Wherein, Mb_c represents the cth template data, and C represents the number of flight missions in the flight simulation teaching data set corresponding to the template data;

[0094] S202. When the deviation value D ≥ the deviation threshold D0, the historical flight data and the template data are subjected to trend analysis to obtain a historical flight data trend curve A and a template data trend curve B respectively; the historical flight data trend curve A and the template data trend curve B are expressed in the same plane rectangular coordinate system, the difference T at the same time point is calculated in turn, all the difference T are summarized and the average value Tμ is calculated, and the time point when the difference T ≥ Tμ is marked as the risk point of the historical flight data; for each historical flight mission, the risk points in all historical flight data are summarized to form a risk point data set F, and F = {f1, f2, ...., ft}, where f1 represents the first risk point, f2 represents the second risk point, and so on, ft represents the tth risk point; the historical flight data corresponding to the risk point is obtained, and the preset key features are extracted from the historical flight data to generate a feature vector of the risk point.

[0095] Step S300 includes:

[0096] S301. For each risk point dataset F of historical flight missions, extract the feature vector corresponding to the risk point, calculate the cosine similarity S1 between different risk points in turn, and construct a similarity matrix Q, and the element q in the similarity matrix Q uv represents the cosine similarity between risk points fu and fv; for each risk point similarity matrix Q, select the elements greater than or equal to the threshold S2, and construct the corresponding directed graph G(V,E), where the node V represents the risk point, the edge E represents the propagation relationship, and the weight w of the edge is represented by the value of the element greater than or equal to the threshold S2 selected;

[0097] S302. Based on the constructed directed graph G(V,E), all propagation paths from the risk point fu to fv are identified through the graph traversal algorithm, and the corresponding path weight W(P_uv) is calculated, and W(P_uv) is equal to the product of all node edges in the propagation path. The paths are sorted in descending order according to the path weight W(P_uv), and the path with the largest path weight W(P_uv) is selected as the propagation path for the corresponding risk point.

[0098] Step S400 includes:

[0099] S401. Acquire the pilot's flight simulation training data in real time, calculate the similarity between the flight simulation training data and the flight simulation teaching data corresponding to the flight mission in the flight simulation teaching data set, and compare the obtained similarity result S' with the corresponding similarity threshold S0;

[0100] The flight simulation teaching data satisfying S'≥S0 is used as the first control data; the deviation value D' between the first control data and the flight simulation training data is calculated, and the calculation formula of the deviation value D' refers to the calculation formula of the deviation value D; and the flight simulation training data with the deviation value D'≥deviation threshold D0 and the first control data are subjected to trend analysis, and the trend analysis process of historical flight data and template data is referred to, so as to identify the real-time risk points in the flight simulation training data, and obtain the real-time feature vector and the corresponding real-time propagation path of the real-time risk points according to the corresponding flight simulation training data;

[0101] S402. Compare the similarity result S' with the corresponding similarity threshold S0. If the similarity result S' of the flight simulation training data of the corresponding real-time flight simulation training task is less than the similarity threshold S0, then perform similarity calculation on the flight simulation training data and the historical flight data in the historical flight database, select the historical flight data with the greatest similarity as the second control data, sort the flight simulation training data according to the mapping relationship between the second control data and the flight simulation teaching data, and perform step S401 analysis on the sorted flight simulation training data and the flight simulation teaching data, so as to identify the real-time risk points in the flight simulation training data, as well as the real-time feature vectors of the real-time risk points and the corresponding real-time propagation paths.

[0102] Step S500 includes:

[0103] S501. For each pilot, obtain the flight simulation training data of the last two consecutive time periods, and obtain the corresponding real-time risk point data set F1 and real-time risk point data set F2 according to the flight simulation training data; count the number m1 and m2 of real-time risk points in the real-time risk point data set F1 and the real-time risk point data set F2 respectively, and construct a corresponding directed graph for each real-time risk point, calculate the path weight W of the real-time propagation path of the real-time risk point according to the directed graph, evaluate the flight simulation training effect of the pilot according to the above data, and calculate the corresponding training effect index R, the specific calculation formula is:

[0104] R={Σe1∈[1,m1],We1} / {Σe2∈[1,m2],We2},

[0105] Wherein, e1 represents the number of the real-time risk point in the real-time risk point data set F1, e2 represents the number of the real-time risk point in the real-time risk point data set F2, We1 represents the path weight of the real-time propagation path of the real-time risk point in the real-time risk point data set F1, and We2 represents the path weight of the real-time propagation path of the real-time risk point in the real-time risk point data set F2;

[0106] S502. If the training effect index R>1, it means that the pilot's training effect is good, and the prompt message "Congratulations! Your training effect is better than last time, please keep it up" is output; if the training effect index R=1, and m1>m2, it means that the pilot's training effect is average, and the prompt message "New risk points have been added, but the handling effect is equivalent, please consider strengthening the training of new risk points" is output; if the training effect index R<1, it means that the pilot's training effect is not good, and the prompt message "The recent training effect is not good, please evaluate the training content as soon as possible and formulate improvement measures" is output; for other values ​​of the training effect index R, the prompt message "Please re-evaluate the training strategy" is output.

[0107] 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.

[0108] 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 risk analysis method for flight simulation teaching and training based on big data, characterized by: The method comprises the following steps: Step S100. Obtain historical flight data and flight simulation teaching data from a database, analyze the historical flight data and the flight simulation teaching data, thereby obtaining a mapping relationship between the two; and organize the historical flight data based on the mapping relationship, thereby constructing a historical flight database; Step S200. According to the historical flight database, the historical flight data in the historical flight database is compared and analyzed with the flight simulation teaching data, so as to identify the risk points in the historical flight database, and form a risk point data set; the risk point data set is analyzed, the relevant features of the risk points are extracted, and the feature vector of the risk points is generated; Step S300. Based on the feature vectors of the risk points, the risk points in the risk point data set are analyzed accordingly, so as to obtain the correlation between different risk points; according to the correlation between different risk points, the propagation path of each risk point is obtained; Step S400. Acquire the pilot's flight simulation training data in real time, analyze the deviation relationship between the flight simulation training data and the flight simulation teaching data, and identify the real-time risk point based on the deviation relationship; analyze the flight simulation training data to obtain the real-time feature vector of the real-time risk point, and obtain the corresponding real-time propagation path according to the real-time feature vector; Step S500. For each pilot, obtain the flight simulation training data of the last two consecutive time periods, evaluate the flight simulation training effect of the pilot according to the corresponding real-time risk points and real-time propagation paths, and output corresponding prompt information according to the flight simulation training effect; The step S200 includes: S201. For each historical flight mission in the historical flight database, the corresponding historical flight data is obtained, and the similarity calculation is performed on the historical flight data and the flight simulation teaching data corresponding to the flight mission in the flight simulation teaching data set in turn, and the obtained similarity result S is compared with the corresponding similarity threshold S0; the flight simulation teaching data satisfying S≥S0 is used as the template data; the deviation value D between the template data Mb and the historical flight data Hf is calculated, and the calculation formula of the deviation value D is: D={Σc∈[1,C],|Mb_c-Hf|} / C, Wherein, Mb_c represents the cth template data, and C represents the number of flight missions in the flight simulation teaching data set corresponding to the template data; S202. When the deviation value D ≥ the deviation threshold value D0, the historical flight data and the template data are subjected to trend analysis, thereby obtaining a historical flight data trend curve A and a template data trend curve B respectively; the historical flight data trend curve A and the template data trend curve B are expressed in the same plane rectangular coordinate system, the difference T at the same time point is calculated in turn, all the difference Ts are summarized and the average value Tμ is calculated, and the time point when the difference T ≥ Tμ is marked as the risk point of the historical flight data; for each historical flight mission, the risk points in all historical flight data are summarized to form a risk point data set F, and F = {f1, f2, ...., ft}, where f1 represents the first risk point, f2 represents the second risk point, and so on, ft represents the tth risk point; the historical flight data corresponding to the risk point is obtained, and the preset key features are extracted from the historical flight data, thereby generating a feature vector of the risk point; The step S300 includes: S301. For each risk point dataset F of historical flight missions, extract the feature vector corresponding to the risk point, calculate the cosine similarity S1 between different risk points in turn, and construct a similarity matrix Q, and the element q in the similarity matrix Q uv represents the cosine similarity between risk points fu and fv; for each risk point similarity matrix Q, select the elements greater than or equal to the threshold S2, and construct the corresponding directed graph G(V,E), where the node V represents the risk point, the edge E represents the propagation relationship, and the weight w of the edge is represented by the value of the element greater than or equal to the threshold S2 selected; S302. Based on the constructed directed graph G(V,E), all propagation paths from the risk point fu to fv are identified through the graph traversal algorithm, and the corresponding path weight W(P_uv) is calculated, and W(P_uv) is equal to the product of all node edges in the propagation path. The paths are sorted in descending order according to the path weight W(P_uv), and the path with the largest path weight W(P_uv) is selected as the propagation path for the corresponding risk point.

2. The flight simulation teaching and training risk analysis method based on big data according to claim 1 is characterized by: The step S100 includes: S101. Obtain historical flight data and flight simulation teaching data from a database, and divide the historical flight data and flight simulation teaching data according to a preset flight mission start mark and a flight mission end mark, so that the historical flight data and the flight simulation teaching data correspond to a number of flight missions, respectively, to form a historical flight data set Dh and a flight simulation teaching data set Ds, and Dh={h1,h2,...,hn}, Ds={s1,s2,...,sk}, wherein h1 represents the historical flight data corresponding to the first historical flight mission, h2 represents the historical flight data corresponding to the second historical flight mission, and so on, and hn represents the historical flight data corresponding to the nth historical flight mission. The historical flight data corresponding to the flight mission of the first flight simulation teaching is represented by s1; similarly, s2 represents the flight simulation teaching data corresponding to the flight mission of the second flight simulation teaching is represented by s2; sk represents the flight simulation teaching data corresponding to the flight mission of the kth flight simulation teaching is represented by sk; n represents the total number of historical flight missions, k represents the total number of flight missions of the flight simulation teaching, and n>k; the historical flight data refers to various flight parameters and information collected during the actual flight process; the flight simulation teaching data refers to data generated or used in the flight simulator, and these data specifically include the flight parameters and information of the simulated aircraft when performing the flight mission; S102. Extract preset key features from the historical flight data set Dh and the flight simulation teaching data set Ds; thereby constructing corresponding key feature vectors Xh_i and Xs_j for each flight mission in the historical flight data set Dh and the flight simulation teaching data set Ds, respectively, where i is a positive integer from 1 to n, and j is a positive integer from 1 to k; use a linear regression model to analyze the mapping relationship between Xh_i and Xs_j, and the linear regression model expression is: Xs_j=A·Xh_i+b, Where A represents the weight matrix, and b represents the bias term. Based on the linear regression model, the mapping relationship between each flight mission in the flight simulation teaching data set Ds and the flight mission in the historical flight data set Dh is obtained. According to the mapping relationship between the two, the historical flight data is sorted to construct a historical flight database.

3. The flight simulation teaching and training risk analysis method based on big data according to claim 1 is characterized by: The step S400 includes: S401. Acquire the pilot's flight simulation training data in real time, calculate the similarity between the flight simulation training data and the flight simulation teaching data corresponding to the flight mission in the flight simulation teaching data set, and compare the obtained similarity result S' with the corresponding similarity threshold S0; The flight simulation teaching data satisfying S'≥S0 is used as the first control data; the deviation value D' between the first control data and the flight simulation training data is calculated, and the calculation formula of the deviation value D' refers to the calculation formula of the deviation value D; and the flight simulation training data with the deviation value D'≥deviation threshold D0 and the first control data are subjected to trend analysis, and the trend analysis process of historical flight data and template data is referred to, so as to identify the real-time risk points in the flight simulation training data, and obtain the real-time feature vector and the corresponding real-time propagation path of the real-time risk points according to the corresponding flight simulation training data; S402. Compare the similarity result S' with the corresponding similarity threshold S0. If the similarity result S' of the flight simulation training data of the corresponding real-time flight simulation training task is less than the similarity threshold S0, then perform similarity calculation on the flight simulation training data and the historical flight data in the historical flight database, select the historical flight data with the greatest similarity as the second control data, sort the flight simulation training data according to the mapping relationship between the second control data and the flight simulation teaching data, and perform step S401 analysis on the sorted flight simulation training data and the flight simulation teaching data, so as to identify the real-time risk points in the flight simulation training data, as well as the real-time feature vectors of the real-time risk points and the corresponding real-time propagation paths.

4. The flight simulation teaching and training risk analysis method based on big data according to claim 3 is characterized by: The step S500 includes: S501. For each pilot, obtain the flight simulation training data of the last two consecutive time periods, and obtain the corresponding real-time risk point data set F1 and real-time risk point data set F2 according to the flight simulation training data; count the number m1 and m2 of real-time risk points in the real-time risk point data set F1 and the real-time risk point data set F2 respectively, and construct a corresponding directed graph for each real-time risk point, calculate the path weight W of the real-time propagation path of the real-time risk point according to the directed graph, evaluate the flight simulation training effect of the pilot according to the above data, and calculate the corresponding training effect index R, the specific calculation formula is: R={Σe1∈[1,m1],We1} / {Σe2∈[1,m2],We2}, Wherein, e1 represents the number of the real-time risk point in the real-time risk point data set F1, e2 represents the number of the real-time risk point in the real-time risk point data set F2, We1 represents the path weight of the real-time propagation path of the real-time risk point in the real-time risk point data set F1, and We2 represents the path weight of the real-time propagation path of the real-time risk point in the real-time risk point data set F2; S502. If the training effect index R>1, it means that the pilot's training effect is good, and the prompt message "Congratulations! Your training effect is better than last time, please keep it up" is output; if the training effect index R=1, and m1>m2, it means that the pilot's training effect is average, and the prompt message "New risk points, but the handling effect is equivalent, please consider strengthening the training of new risk points" is output; if the training effect index R<1, it means that the pilot's training effect is not good, and the prompt message "The recent training effect is not good, please evaluate the training content as soon as possible and formulate improvement measures" is output; for other values ​​of the training effect index R, the prompt message "Please re-evaluate the training strategy" is output.

5. A flight simulation teaching and training risk analysis system based on big data, applied to the flight simulation teaching and training risk analysis method based on big data as described in any one of claims 1 to 4, characterized in that: The system includes: a data acquisition and sorting module, a risk point identification and analysis module, a risk propagation path analysis module, a real-time risk monitoring module, and a training effect evaluation and feedback module; The data acquisition and arrangement module extracts historical flight data and flight simulation teaching data from the database, divides the historical flight data and the flight simulation teaching data to form a historical flight data set and a flight simulation teaching data set, analyzes the historical flight data set and the flight simulation teaching data set to obtain a mapping relationship between the historical flight data and the flight simulation teaching data, and arranges the historical flight data to construct a historical flight database; The risk point identification and analysis module calculates the similarity between each historical flight mission in the historical flight database and the flight simulation teaching data, and compares it with the similarity threshold to screen out the template data; calculates the deviation value between the template data and the historical flight data, and performs trend analysis to identify the risk points; extracts relevant features from the identified risk points, constructs risk point feature vectors, and forms a risk point data set; The risk propagation path analysis module calculates the cosine similarity between risk points according to the risk point feature vectors, constructs a similarity matrix, screens risk points based on the similarity matrix, and constructs a directed graph; uses a graph traversal algorithm to identify the propagation path between risk points, calculates the path weight, and selects the path with the largest weight as the propagation path for the corresponding risk point; The real-time risk monitoring module acquires the pilot's flight simulation training data in real time, and performs similarity calculation with the flight simulation teaching data, identifies the real-time risk points according to the similarity calculation results, extracts the feature vectors of the real-time risk points, and constructs the real-time propagation path; The training effect evaluation and feedback module counts the number of risk points in the real-time risk point data set for each pilot and calculates the corresponding propagation path weight; calculates the training effect index according to the path weight of the real-time risk point, and outputs corresponding prompt information according to different values ​​of the training effect index; The risk point identification and analysis module includes a similarity calculation unit, a deviation analysis unit and a risk point feature extraction unit; The similarity calculation unit calculates the similarity with the flight simulation teaching data for each historical flight mission in the historical flight database, and compares it with the similarity threshold to screen out the template data; the deviation analysis unit calculates the deviation value between the template data and the historical flight data, and performs trend analysis to identify risk points; the risk point feature extraction unit extracts relevant features from the identified risk points, constructs risk point feature vectors, and forms a risk point data set; The risk propagation path analysis module includes a similarity matrix construction unit, a propagation relationship identification unit and a propagation path calculation unit; The similarity matrix construction unit calculates the cosine similarity between risk points based on the risk point feature vectors and constructs a similarity matrix; the propagation relationship identification unit screens out risk points with high similarity based on the similarity matrix and constructs a directed graph; the propagation path calculation unit uses a graph traversal algorithm to identify the propagation path between risk points, calculates the path weight and selects the path with the largest weight as the corresponding propagation path.

6. The flight simulation teaching and training risk analysis system based on big data according to claim 5 is characterized by: The data acquisition and sorting module includes a data acquisition unit, a mapping relationship establishment unit and a data sorting unit; The data acquisition unit extracts historical flight data and flight simulation teaching data from the database, and divides the historical flight data and flight simulation teaching data into multiple flight missions according to preset flight mission start and end marks to form historical flight data sets and flight simulation teaching data sets; the mapping relationship establishment unit extracts key features and constructs key feature vectors, and uses a linear regression model to analyze the mapping relationship between the historical flight data and the flight simulation teaching data; the data sorting unit sorts the historical flight data based on the mapping relationship between the historical flight data and the flight simulation teaching data to construct a historical flight database.

7. The flight simulation teaching and training risk analysis system based on big data according to claim 5 is characterized by: The real-time risk monitoring module includes a real-time data acquisition unit, a real-time deviation analysis unit and a real-time feature vector generation unit; The real-time data acquisition unit acquires the pilot's flight simulation training data in real time, and performs similarity calculation with the flight simulation teaching data; the real-time deviation analysis unit analyzes the corresponding deviation value according to the similarity calculation result, and performs trend analysis to identify the real-time risk point; the real-time feature vector generation unit extracts the feature vector of the real-time risk point and constructs the real-time propagation path; The training effect evaluation and feedback module includes a real-time risk point statistics unit, a training effect index calculation unit and a feedback prompt information unit; The real-time risk point statistics unit counts the number of real-time risk points in the real-time risk point data set for each pilot, and calculates the corresponding real-time propagation path weight; the training effect index calculation unit calculates the training effect index according to the path weight of the real-time risk point; The feedback prompt information unit outputs corresponding prompt information according to different values ​​of the training effect index.

Images