Control sector safety risk index prediction method

Abstract

The invention discloses a control sector security risk index prediction method. The method comprises the following steps: defining a current security risk index and a future security risk index of a control sector; acquiring situation characteristics and real values of future security risk indexes of the controlled sectors from historical operation data of the controlled sectors, and establishing a database; establishing and training a prediction model by using the database to realize mapping from the current state of the control sector to a future safety risk index; and receiving real-time trajectory data of the aircraft, and predicting the future safety risk index of the controlled sector in real time by using the trained prediction model. According to the method, the concept of the security risk is popularized to the sector security risk from the conflict risk of every two aircrafts, and the future security risk index of the sector can be predicted according to the current situation information of the sector, so that the situation awareness capability of managers to the managed sector is improved.

Description

Technical Field

[0001] This invention belongs to the field of air traffic management technology, specifically relating to a method for predicting safety risk indicators in controlled sectors. Background Technology

[0002] In the current field of air traffic management, sector safety risks are a major concern for managers. However, there is currently no quantitative indicator to characterize sector safety risks, which leads to a disconnect between the needs of managers and the functions of the system. It is urgent to address this shortcoming in the current air traffic control system.

[0003] Currently, some researchers have proposed quantitative characterization methods for aircraft accident symptoms between pairs of aircraft and published standards. However, the focus of managers differs from that of command personnel. Managers tend to be aware of more macro-level safety risk indicators rather than the safety risks between pairs of aircraft. Therefore, how to transform the safety risk indicators between aircraft into overall sector safety risk indicators is a current research challenge. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the purpose of this invention is to provide a method for predicting security risk indicators of controlled sectors, so as to solve the problem that the prior art lacks the ability to predict security risk indicators of controlled sectors in the next 15 minutes.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is as follows:

[0006] The present invention provides a method for predicting security risk indicators of controlled sectors, comprising the following steps:

[0007] 1) Define the current and future security risk indicators for the controlled sectors;

[0008] 2) Obtain the true values ​​of situation characteristics and future security risk indicators of the controlled sectors from historical operational data, and establish a database;

[0009] 3) Utilize the database to establish and train a prediction model to map the current state of the controlled sector to future security risk indicators;

[0010] 4) Receive real-time aircraft trajectory data and use the trained prediction model to predict future safety risk indicators of the control sector in real time.

[0011] Further, step 1) specifically includes:

[0012] 11) The definition of the current safety risk index of the control sector is: the risk level score of the current control sector obtained by statistical analysis of the accident symptom risk index of each pair of aircraft in the current control sector, with a score range of 0-100;

[0013] 12) The definition of the future security risk index of the controlled sector is: the statistical result of the current security risk index of the controlled sector from the current moment to a certain period of time in the future. It is used to describe the peak information of the security risk of the controlled sector in the future period of time, and the score range is 0-100.

[0014] Furthermore, step 2) specifically includes:

[0015] 21) Obtain historical operational data of control sectors stored in the form of trackpoints in the flow system / air traffic automation system;

[0016] 22) Integrate the trajectory data according to the actual spatial range of the sector to obtain the overall situational characteristics data of the target sector;

[0017] 23) Calculate the current security risk index of the controlled sector at each moment in the historical data of the trajectory data, and obtain the true value of the future security risk index of the controlled sector at each moment based on this. Then, store it in the database together with the overall situational characteristic data of the target sector, with time information as the primary key.

[0018] Furthermore, the overall situational characteristic data of the target sector in step 22) includes: the current number of flights in the sector, the proportion of level flights in the sector, and the number of eccentric flights in the sector.

[0019] Furthermore, the calculation method for the current safety risk index of the control sector in step 23) is as follows: the maximum value of the aircraft accident symptom value between any two aircraft in the control sector, the average of the first three valid values ​​of the aircraft accident symptom value between any two aircraft in the control sector, or the weighted average of the first three valid values ​​of the aircraft accident symptom value between any two aircraft in the control sector.

[0020] Furthermore, step 3) specifically includes:

[0021] 31) Construct a regression model based on a two-layer neural network using a database. The model consists of an input layer, two hidden layers, and an output layer. The input layer has 3 nodes and receives three data points as input: the current number of flights in the sector, the percentage of level flights in the sector, and the number of skewed flights in the sector. The two hidden layers each contain 30 nodes, and the activation function is sigmoid. The output layer contains 1 node and outputs the predicted future security risk index of the sector. The nodes in each layer are connected using a fully connected method.

[0022] 32) Using the overall situational characteristics data of the target sector as input and the true value of the sector's future security risk indicators as output, the model is trained using stochastic gradient descent based on backpropagation. The learning rate and number of iterations are adjusted to train the model to a stable state, resulting in a trained prediction model. The overall structure of the trained model remains unchanged from that before training, and the training process only focuses on the activation coefficients between nodes. The trained model can predict the sector's future security risk indicators based on three data points: the current number of flights in the sector, the proportion of level flights in the sector, and the number of eccentric flights in the sector.

[0023] Furthermore, step 4) specifically includes:

[0024] 41) Receive real-time flight track data of aircraft, statistically analyze the overall situational characteristics of the target sector, and obtain three statistical values: the current number of flights in the sector, the proportion of level flights in the sector, and the number of eccentric flights in the sector. These are the input data required for the trained prediction model.

[0025] 42) Using the input data and the trained prediction model, the system passes through the input layer, the first hidden layer, the second hidden layer and the output layer in one forward propagation, and calculates the value output in the final output layer to obtain the prediction result of the sector security risk index.

[0026] This invention uses historical flight operation data to obtain sector status information at each time point, and combines it with the real values ​​of future sector security indicators for a period of time after the time point to train a regression model composed of neural networks, thereby obtaining the mapping relationship between the current sector status information and the future sector security indicators. Thus, in actual operation, the future sector security indicators at the current moment can also be predicted based on the current sector status information.

[0027] The beneficial effects of this invention are:

[0028] This invention extends the concept of security risk from the risk of conflict between two aircraft to the security risk of a sector, and can predict future security risk indicators of a sector based on the current situation information of the sector, thereby improving the situational awareness of managers over the sectors under their jurisdiction. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the method of the present invention. Detailed Implementation

[0030] In order to facilitate understanding by those skilled in the art, the present invention will be further described below with reference to embodiments and drawings. The contents mentioned in the embodiments are not intended to limit the present invention.

[0031] Reference Figure 1 As shown, the present invention provides a method for predicting security risk indicators for controlled sectors, comprising the following steps:

[0032] 1) Define the current and future security risk indicators for the controlled sector; specifically including:

[0033] 11) The definition of the current safety risk index of the control sector is: the risk level score of the current control sector obtained by statistical analysis of the accident symptom risk index of each pair of aircraft in the current control sector, with a score range of 0-100;

[0034] 12) The definition of the future security risk index of the controlled sector is: the statistical result of the current security risk index of the controlled sector from the current moment to a certain period of time in the future. It is used to describe the peak information of the security risk of the controlled sector in the future period of time, and the score range is 0-100.

[0035] 2) Obtain the true values ​​of situational characteristics and future security risk indicators of the controlled sectors from historical operational data, and establish a database; specifically including:

[0036] 21) Obtain historical operational data of control sectors stored in the form of trackpoints in the flow system / air traffic automation system;

[0037] 22) Integrate the trajectory data according to the actual spatial range of the sector to obtain the overall situational characteristics data of the target sector;

[0038] 23) Calculate the current security risk index of the controlled sector at each moment in the historical data of the trajectory data, and obtain the true value of the future security risk index of the controlled sector at each moment based on this. Then, store it in the database together with the overall situational characteristic data of the target sector, with time information as the primary key.

[0039] Among them, the overall situational characteristic data of the target sector in step 22) includes: the current number of flights in the sector, the proportion of level flights in the sector, and the number of eccentric flights in the sector.

[0040] The calculation method for the current safety risk index of the control sector in step 23) is as follows: the maximum value of the aircraft accident symptom value between any two aircraft in the control sector, the average of the first three valid values ​​of the aircraft accident symptom value between any two aircraft in the control sector, or the weighted average of the first three valid values ​​of the aircraft accident symptom value between any two aircraft in the control sector.

[0041] 3) Utilize databases to establish and train predictive models to map the current state of controlled sectors to future security risk indicators; specifically including:

[0042] 31) Construct a regression model based on a two-layer neural network using a database. The model consists of an input layer, two hidden layers, and an output layer. The input layer has 3 nodes and receives three data points as input: the current number of flights in the sector, the percentage of level flights in the sector, and the number of skewed flights in the sector. The two hidden layers each contain 30 nodes, and the activation function is sigmoid. The output layer contains 1 node and outputs the predicted future security risk index of the sector. The nodes in each layer are connected using a fully connected method.

[0043] 32) Using the overall situational characteristics data of the target sector as input and the true value of the sector's future security risk indicators as output, the model is trained using stochastic gradient descent based on backpropagation. The learning rate and number of iterations are adjusted to train the model to a stable state, resulting in a trained prediction model. The overall structure of the trained model remains unchanged from that before training, and the training process only focuses on the activation coefficients between nodes. The trained model can predict the sector's future security risk indicators based on three data points: the current number of flights in the sector, the proportion of level flights in the sector, and the number of eccentric flights in the sector.

[0044] 4) Receive real-time aircraft trajectory data and use the trained prediction model to predict future safety risk indicators for the controlled sector in real time; specifically including:

[0045] 41) Receive real-time flight track data of aircraft, statistically analyze the overall situational characteristics of the target sector, and obtain three statistical values: the current number of flights in the sector, the proportion of level flights in the sector, and the number of eccentric flights in the sector. These are the input data required for the trained prediction model.

[0046] 42) Using the input data and the trained prediction model, the system passes through the input layer, the first hidden layer, the second hidden layer and the output layer in one forward propagation, and calculates the value output in the final output layer to obtain the prediction result of the sector security risk index.

[0047] This invention has many specific applications. The above description is only a preferred embodiment of this invention. It should be noted that for those skilled in the art, several improvements can be made without departing from the principle of this invention, and these improvements should also be considered within the scope of protection of this invention.

Claims

1. A method for predicting security risk indicators of controlled sectors, characterized in that, Here are the steps: 1) Define the current and future security risk indicators for the controlled sectors; 2) Obtain the true values ​​of situation characteristics and future security risk indicators of the controlled sectors from historical operational data, and establish a database; 3) Utilize the database to establish and train a prediction model to map the current state of the controlled sector to future security risk indicators; 4) Receive real-time aircraft trajectory data and use the trained prediction model to predict future safety risk indicators of the control sector in real time.

2. The method for predicting security risk indicators of controlled sectors according to claim 1, characterized in that, Step 1) specifically includes: 11) The definition of the current safety risk index of the control sector is: the risk level score of the current control sector obtained by statistical analysis of the accident symptom risk index of each pair of aircraft in the current control sector, with a score range of 0-100; 12) The definition of the future security risk index of the controlled sector is: the statistical result of the current security risk index of the controlled sector from the current moment to a certain period of time in the future. It is used to describe the peak information of the security risk of the controlled sector in the future period of time, and the score range is 0-100.

3. The method for predicting security risk indicators of controlled sectors according to claim 1, characterized in that, Step 2) specifically includes: 21) Obtain historical operational data of control sectors stored in the form of trackpoints in the flow system / air traffic automation system; 22) Integrate the trajectory data according to the actual spatial range of the sector to obtain the overall situational characteristics data of the target sector; 23) Calculate the current security risk index of the controlled sector at each moment in the historical data of the trajectory data, and obtain the true value of the future security risk index of the controlled sector at each moment based on this. Then, store it in the database together with the overall situational characteristic data of the target sector, with time information as the primary key.

4. The method for predicting security risk indicators of controlled sectors according to claim 1, characterized in that, Step 3) specifically includes: 31) Construct a regression model based on a two-layer neural network using a database. The model consists of an input layer, two hidden layers, and an output layer. The input layer has 3 nodes and receives three data points as input: the current number of flights in the sector, the percentage of level flights in the sector, and the number of skewed flights in the sector. The two hidden layers each contain 30 nodes, and the activation function is sigmoid. The output layer contains 1 node and outputs the predicted future security risk index of the sector. The nodes in each layer are connected using a fully connected method. 32) Using the overall situational characteristics data of the target sector as input and the true value of the sector's future security risk indicators as output, the model is trained using stochastic gradient descent based on backpropagation. The learning rate and number of iterations are adjusted to train the model to a stable state, resulting in a trained prediction model. The overall structure of the trained model remains unchanged from that before training, and the training process only focuses on the activation coefficients between nodes. The trained model can predict the sector's future security risk indicators based on three data points: the current number of flights in the sector, the proportion of level flights in the sector, and the number of eccentric flights in the sector.

5. The method for predicting security risk indicators of controlled sectors according to claim 4, characterized in that, The overall situational characteristic data of the target sector in step 22) includes: the current number of flights in the sector, the proportion of level flights in the sector, and the number of eccentric flights in the sector.

6. The method for predicting security risk indicators of controlled sectors according to claim 4, characterized in that, The calculation method for the current safety risk index of the control sector in step 23) is as follows: the maximum value of the aircraft accident symptom value between any two aircraft in the control sector, the average of the first three valid values ​​of the aircraft accident symptom value between any two aircraft in the control sector, or the weighted average of the first three valid values ​​of the aircraft accident symptom value between any two aircraft in the control sector.

7. The method for predicting security risk indicators of controlled sectors according to claim 4, characterized in that, Step 4) specifically includes: 41) Receive real-time flight track data of aircraft, statistically analyze the overall situational characteristics of the target sector, and obtain three statistical values: the current number of flights in the sector, the proportion of level flights in the sector, and the number of eccentric flights in the sector. These are the input data required for the trained prediction model. 42) Using the input data and the trained prediction model, the system passes through the input layer, the first hidden layer, the second hidden layer and the output layer in one forward propagation, and calculates the value output in the final output layer to obtain the prediction result of the sector security risk index.

Images