A Real-Time Automatic Identification Method for Launch Vehicle Feature Events Based on Bayesian Estimation

Abstract

This invention relates to the field of real-time automatic identification of launch vehicle characteristic events, specifically a real-time automatic identification method for launch vehicle characteristic events based on Bayesian estimation, comprising the following steps: (1) preprocessing the measurement values ​​of the current data source; (2) caching the preprocessed results of the current data source; (3) performing consistency identification on all cached data sources according to the upper limit of residual error, and grouping them; (4) determining the occurrence time of the characteristic event based on the mutual verification conditions between data sources; (5) repeating steps (1) to (4) when a new data source arrives, until a judgment result is obtained. Compared with existing methods, the method proposed in this invention fully considers the mutual verification conditions between different data sources and the inherent prior probability distribution model of the characteristic event itself, ensuring the real-time automatic identification of characteristic events while having a higher identification accuracy.

Description

Technical Field

[0001] This invention relates to the field of real-time automatic identification of characteristic events of launch vehicles, specifically a real-time automatic identification method for characteristic events of launch vehicles based on Bayesian estimation. Background Technology

[0002] Characteristic events represent critical actions during a launch vehicle's flight and are crucial indicators for determining whether the flight is normal. The timing of characteristic events can be determined through optical images or changes in the vehicle's flight trajectory, but determining the timing based on telemetry command parameters remains the primary method, especially for automatic characteristic event identification. Since telemetry data inevitably contains errors during launch vehicle flight, especially during critical actions, multiple command parameters characterizing the occurrence of characteristic events are typically selected as the basis for judgment to ensure reliable identification. Furthermore, to ensure reliable transmission of telemetry data, the launch vehicle may buffer important parameters (such as command parameters) for a certain period before transmission, and the ground system may relay telemetry data through multiple telemetry devices. Therefore, the data sources used for characteristic event identification include different command parameters, different device sources, real-time sources, and delayed sources. Accurately determining the occurrence time of characteristic events from these numerous data sources is a pressing problem that needs to be solved. Here, device source refers to the data source forwarded through different devices, real-time source refers to the real-time measured value of the parameter, and delayed source refers to the delayed measured value of the parameter.

[0003] Currently, real-time automatic identification of characteristic events based on telemetry command parameters mainly determines the occurrence time of a characteristic event by the number of data sources with identical current measurements. When two or more data sources show identical measurements, the occurrence of a characteristic event is immediately determined. This method has strong real-time performance and can effectively identify the occurrence time of characteristic events when the data quality is good. However, it ignores the conditions for mutual verification between data sources and the influence of subsequent arrivals from data sources. Especially when multiple sets of measurements have the same number of data sources, judging solely based on the order of arrival of data sources has a high probability of misjudgment. In fact, characteristic events themselves have inherent probability distribution models. Fully utilizing the prior distribution information of characteristic events will help improve the accuracy of characteristic event occurrence time determination. Summary of the Invention

[0004] The purpose of this invention is to address the issue that existing methods for real-time automatic identification of feature events using telemetry command parameters, which rely solely on measurements from the current data source, may result in a certain probability of misjudgment. To further improve the accuracy of real-time automatic identification of feature events based on telemetry command parameters, this invention provides a method for real-time automatic identification of launch vehicle feature events based on Bayesian estimation.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a real-time automatic identification method for characteristic events of launch vehicles based on Bayesian estimation, specifically including the following steps:

[0006] (1) Preprocess the measured values ​​from the current data source using the method: y = x + Δt t +Δt d Where y is the result after preprocessing the data source, x is the measured value of the data source, and Δt t This represents the theoretical time deviation between the instruction represented by the data source and the characteristic event. When there is no theoretical time deviation between the instruction and the characteristic event, this value is zero. When the theoretical time of the instruction represented by the data source is less than the theoretical time of the characteristic event, this value is positive; otherwise, it is negative. Δt d This is the inherent delay bias of the delay source; obviously, when the data source is a real-time source, this value is zero.

[0007] (2) Cache the preprocessed results of the current data source;

[0008] (3) Based on the upper limit of residual error δy, perform consistency judgment on all data sources in the cache and group them. The upper limit of residual error refers to the maximum value of the total error, including the calculation error caused by different instruction calculation methods and the delay error caused by different transmission channels of different devices. The method is as follows:

[0009] (3.1) Arrange the preprocessed results of all data sources in ascending order to obtain the dataset D = {y1, y2, ... y k …y g …y n};

[0010] (3.2) Starting from y1, connect each of y2 to y1 one by one. n Comparison, when y2~y n When the difference between y1 and y2 is less than δy, it is determined that y1 is consistent with y1, that is, the measured values ​​are the same, and it is grouped with y1 to obtain the first dataset D1.

[0011] (3.3) Then from y k At the beginning, y k This represents the minimum value in dataset D that was not assigned to the first dataset D1 along with y1, and is calculated by comparing each value with y1. k+1 ~y n Comparison, when y k+1 ~y n With y k When the difference is less than δy, determine its relationship with y. k Consistent, meaning the measured values ​​are the same, is compared with y k They are grouped together to obtain the second dataset D2.

[0012] (3.4) Then from y g At the beginning, y g This indicates that data in dataset D does not contain data that is not related to y. k The minimum value assigned to the second dataset D2 is compared with y one by one. g+1 ~y n Comparison, when y g+1 ~y n With y g When the difference is less than δy, determine its relationship with y. g Consistent, meaning the measured values ​​are the same, is compared with y g They are grouped together to obtain the third dataset, D3.

[0013] This process continues until all data sources are grouped, resulting in a grouped dataset D = {D1, D2, ..., D...}. m}, where m represents the number of groups in the dataset.

[0014] (4) Determine the time of occurrence of the feature event based on the mutual corroboration conditions between data sources.

[0015] (4.1) The conditions for mutual corroboration are:

[0016] 1) When the real-time source and the delayed source measurement values ​​are consistent and different instruction parameters are forwarded by different devices in the data source, it indicates that both the information source and the transmission channel are reliable;

[0017] 2) When real-time source measurements with different instruction parameters forwarded by different devices are consistent, it indicates that both the information source and the transmission channel are reliable;

[0018] 3) When the delay measurement values ​​of different instruction parameters forwarded by different devices in the data source are consistent, it indicates that both the information source and the transmission channel are reliable;

[0019] 4) When real-time and delayed source measurements with different instruction parameters forwarded by the same device are consistent, it indicates that both the information source and the transmission channel are reliable.

[0020] 5) When real-time source measurements with different instruction parameters forwarded by the same device are consistent, it indicates that the information source is trustworthy;

[0021] 6) When different instruction parameters in the data source are forwarded by the same device and the measured delay values ​​are consistent, it indicates that the information source is trustworthy;

[0022] 7) When the real-time source and the delayed source in the data source have the same instruction parameter forwarded by the same device and their measurement values ​​are consistent, it indicates that the transmission channel is reliable;

[0023] 8) When real-time source measurements with the same instruction parameter forwarded by different devices are consistent in the data source, it indicates that the transmission channel is reliable;

[0024] 9) When the delay measurement values ​​of the same instruction parameter forwarded by different devices in the data source are consistent, it indicates that the transmission channel is reliable;

[0025] 10) When the real-time source and the delayed source measurements of the same instruction parameter forwarded by different devices in the data source are consistent, it indicates that the transmission channel is reliable;

[0026] (4.2) The specific method is as follows:

[0027] (4.2.1) When the data source satisfies the mutual verification conditions 1) to 4), it indicates that both the information source and the transmission channel are reliable, and the occurrence of the characteristic event can be determined immediately. The occurrence time of the characteristic event is the minimum measurement value when the data source is consistent.

[0028] (4.2.2) When all data sources for the feature event have solved or reached the maximum theoretical deviation limit of the event occurrence time, and the data sources satisfy the mutual verification conditions 5) to 10), it indicates that either the information source is reliable or the transmission channel is reliable, and the feature event can be determined to have occurred. The occurrence time of the feature event is determined by the feature event discrimination algorithm based on Bayesian maximum a posteriori estimation. The specific method is as follows:

[0029] (4.2.2.1) Calculate the data source as y i The prior probability density value p(y) i The formula is:

[0030] Where μ is the theoretical time for the occurrence of the characteristic event, and 3σ represents the maximum theoretical deviation of the occurrence time of the characteristic event.

[0031] (4.2.2.2) Calculate the time when the characteristic event occurs is y. i At that time, each y in dataset D j probability density value p(y) j |y i The formula is:

[0032]

[0033] (4.2.2.3) Calculate the occurrence time of the characteristic event as y i The probability density value p(D|y) of dataset D. i The formula is:

[0034]

[0035] (4.2.2.4) Calculate the occurrence time of the characteristic event as y.i The relative probability density value p(y) at time i |D), the formula is:

[0036] p(y i |D)=p(y i )p(D|y i )

[0037] (4.2.2.5) Following steps (4.2.2.1) to (4.2.2.4), calculate the relative probability density value of each data source in dataset D.

[0038] (4.2.2.6) Calculate the y of each data source in dataset D. i The relative probability P(y) of the characteristic event i |D), the formula is:

[0039]

[0040] (4.2.2.7) Calculate the value of each data source D in dataset D. l The relative probability P(D) of the characteristic event l |D), the formula is:

[0041] Among them, D l Indicates data source y s ~y e The dataset is grouped into the l-th dataset.

[0042] (4.2.2.8) From {P(D1|D), P(D2|D), ..., P(D l |D)…,P(D m The dataset with the highest probability value in |D)} is selected as the valid dataset, and the occurrence time of the feature event is the minimum measurement value in the valid dataset.

[0043] (4.2.3) When all data sources for the characteristic event have solved or reached the maximum theoretical deviation limit of the occurrence time of the characteristic event, if the data sources do not meet the verification conditions, it is determined that the characteristic event has not occurred.

[0044] (5) When a new data source arrives, repeat steps (1) to (4) until a decision is reached.

[0045] The beneficial effects of this invention are as follows: Addressing the problem of automatic identification of feature events based on telemetry command parameters during launch vehicle flight, this invention proposes a real-time automatic identification method for feature events based on Bayesian estimation. Compared with existing methods, this method fully considers the conditions for mutual verification between different data sources. It not only utilizes the current data source measurements but also fully leverages the inherent prior probability distribution model of the feature events themselves. While ensuring the real-time nature of automatic feature event identification, it further improves the accuracy of feature event occurrence time identification. Attached Figure Description

[0046] Figure 1 This is a flowchart of a real-time automatic discrimination method for feature events based on Bayesian estimation.

[0047] Figure 2 To evaluate the probability distribution of feature event 2 in the case.

[0048] Figure 3 To evaluate the probability distribution of feature event 3 in the case. Detailed Implementation

[0049] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0050] like Figure 1 As shown, a real-time automatic identification method for characteristic events of launch vehicles based on Bayesian estimation specifically includes the following steps:

[0051] (1) Preprocess the measured values ​​from the current data source using the following method:

[0052] y = x + Δt t +Δt d Where y is the result after preprocessing the data source, x is the measured value of the data source, and Δt t This represents the theoretical time deviation between the instruction represented by the data source and the characteristic event. When there is no theoretical time deviation between the instruction and the characteristic event, this value is zero. When the theoretical time deviation of the instruction represented by the data source is less than the theoretical time deviation of the characteristic event, this value is positive; otherwise, it is negative. Δt d This is the inherent delay bias of the delay source; obviously, when the data source is a real-time source, this value is zero.

[0053] (2) Cache the preprocessed results of the current data source;

[0054] (3) Based on the upper limit of residual error δy, perform consistency judgment on all data sources in the cache and group them. The upper limit of residual error refers to the maximum value of the total error, including the calculation error caused by different instruction calculation methods and the delay error caused by different transmission channels of different devices. The method is as follows:

[0055] (3.1) Arrange the preprocessed results of all data sources in ascending order to obtain the dataset D = {y1, y2, ... y k …y g …y n};

[0056] (3.2) Starting from y1, connect each of y2 to y1 one by one. n Comparison, when y2~y n When the difference between y1 and y2 is less than δy, it is determined that y1 is consistent with y1, that is, the measured values ​​are the same, and it is grouped with y1 to obtain the first dataset D1.

[0057] (3.3) Then from y k At the beginning, y k This represents the minimum value in dataset D that was not assigned to the first dataset D1 along with y1, and is calculated by comparing each value with y1. k+1 ~y n Comparison, when y k+1 ~y n With y k When the difference is less than δy, determine its relationship with y. k Consistent, meaning the measured values ​​are the same, is compared with y k They are grouped together to obtain the second dataset D2.

[0058] (3.4) Then from y g At the beginning, y g This indicates that data in dataset D does not contain data that is not related to y. k The minimum value assigned to the second dataset D2 is compared with y one by one. g+1 ~y n Comparison, when y g+1 ~y n With y g When the difference is less than δy, determine its relationship with y. g Consistent, meaning the measured values ​​are the same, is compared with y g They are grouped together to obtain the third dataset, D3.

[0059] This process continues until all data sources are grouped, resulting in a grouped dataset D = {D1, D2, ..., D...}. m}, where m represents the number of groups in the dataset.

[0060] (4) Determine the time of occurrence of the feature event based on the mutual corroboration conditions between data sources.

[0061] (4.1) The conditions for mutual corroboration are:

[0062] 1) When the real-time source and the delayed source measurement values ​​are consistent and different instruction parameters are forwarded by different devices in the data source, it indicates that both the information source and the transmission channel are reliable;

[0063] 2) When real-time source measurements with different instruction parameters forwarded by different devices are consistent, it indicates that both the information source and the transmission channel are reliable;

[0064] 3) When the delay measurement values ​​of different instruction parameters forwarded by different devices in the data source are consistent, it indicates that both the information source and the transmission channel are reliable;

[0065] 4) When real-time and delayed source measurements with different instruction parameters forwarded by the same device are consistent, it indicates that both the information source and the transmission channel are reliable.

[0066] 5) When real-time source measurements with different instruction parameters forwarded by the same device are consistent, it indicates that the information source is trustworthy;

[0067] 6) When different instruction parameters in the data source are forwarded by the same device and the measured delay values ​​are consistent, it indicates that the information source is trustworthy;

[0068] 7) When the real-time source and the delayed source in the data source have the same instruction parameter forwarded by the same device and their measurement values ​​are consistent, it indicates that the transmission channel is reliable;

[0069] 8) When real-time source measurements with the same instruction parameter forwarded by different devices are consistent in the data source, it indicates that the transmission channel is reliable;

[0070] 9) When the delay measurement values ​​of the same instruction parameter forwarded by different devices in the data source are consistent, it indicates that the transmission channel is reliable;

[0071] 10) When the real-time source and the delayed source measurements of the same instruction parameter forwarded by different devices in the data source are consistent, it indicates that the transmission channel is reliable;

[0072] (4.2) The specific method is as follows:

[0073] (4.2.1) When the data source satisfies the mutual verification conditions 1) to 4), it indicates that both the information source and the transmission channel are reliable, and the occurrence of the characteristic event can be determined immediately. The occurrence time of the characteristic event is the minimum measurement value when the data source is consistent.

[0074] (4.2.2) When all data sources for the feature event have solved or reached the maximum theoretical deviation limit of the event occurrence time, and the data sources satisfy the mutual verification conditions 5) to 10), it indicates that either the information source is reliable or the transmission channel is reliable, and the feature event can be determined to have occurred. The occurrence time of the feature event is determined by the feature event discrimination algorithm based on Bayesian maximum a posteriori estimation. The specific method is as follows:

[0075] (4.2.2.1) Calculate the data source as y i The prior probability density value p(y)i The formula is: Where μ is the theoretical time for the occurrence of the characteristic event, and 3σ represents the maximum theoretical deviation of the occurrence time of the characteristic event.

[0076] (4.2.2.2) Calculate the time when the characteristic event occurs is y. i At that time, each y in dataset D j probability density value p(y) j |y i The formula is:

[0077]

[0078] (4.2.2.3) Calculate the occurrence time of the characteristic event as y i The probability density value p(D|y) of dataset D. i The formula is:

[0079]

[0080] (4.2.2.4) Calculate the occurrence time of the characteristic event as y. i The relative probability density value p(y) at time i |D), the formula is:

[0081] p(y i |D)=p(y i )p(D|y i )

[0082] (4.2.2.5) Following steps (4.2.2.1) to (4.2.2.4), calculate the relative probability density value of each data source in dataset D.

[0083] (4.2.2.6) Calculate the y of each data source in dataset D. i The relative probability P(y) of the characteristic event i |D), the formula is:

[0084]

[0085] (4.2.2.7) Calculate the value of each data source D in dataset D. l The relative probability P(D) of the characteristic event l |D), the formula is:

[0086] Among them, D l Indicates data source y s ~y e The dataset is grouped into the l-th dataset.

[0087] (4.2.2.8) From {P(D1|D), P(D2|D), ..., P(D m The dataset with the highest probability value in |D)} is selected as the valid dataset, and the occurrence time of the feature event is the minimum measurement value in the valid dataset.

[0088] (4.2.3) When all data sources for the characteristic event have solved or reached the maximum theoretical deviation limit of the occurrence time of the characteristic event, if the data sources do not meet the verification conditions, it is determined that the characteristic event has not occurred.

[0089] (5) When a new data source arrives, repeat steps (1) to (4) until a decision is reached.

[0090] To evaluate the effectiveness of the method proposed in this invention, a typical data source of characteristic events during aircraft flight is used as an example to conduct a detailed test and analysis of the method's practicality and effectiveness. During the experiment, the upper limit of residual error was set to 0.3 seconds, the maximum theoretical deviation of characteristic events was set to 30 seconds, and the delay time of the delay source was 10 seconds.

[0091]

[0092] Table 1. Measurement values ​​of characteristic event data sources

[0093] Table 1 shows the measurement values ​​of the data source for characteristic events during the flight of the aircraft. As can be seen from characteristic event 1 in Table 1, due to a theoretical deviation of 1 second between command parameter 3 and characteristic event 1, consistency correction is needed before judgment. When the flight time is 116.696 seconds, although the measurement values ​​of the data source for command parameter 2 and command parameter 3 are consistent at 116.696 seconds and 116.701 seconds (115.701 seconds + 1 second), the existing method can immediately determine that the characteristic event has occurred at 116.696 seconds. However, according to the judgment method proposed in this invention, since both are real-time sources forwarded by the same device, the judgment condition of immediate occurrence of the characteristic event is not met. Therefore, this measurement value cannot be used as the occurrence time of the characteristic event. Until the flight time is 131.259 seconds, since multiple command parameters have consistent measurement values ​​from data sources forwarded by different devices, the occurrence of characteristic event 1 can be immediately determined, and the occurrence time is 131.259 seconds. Clearly, analysis of all data sources shows that the occurrence time of the characteristic event is 131.259 seconds, which is consistent with the discrimination result of the method proposed in this invention.

[0094] As can be seen from Feature Event 2 in Table 1, when the flight time is 208.789 seconds, although the data source measurements forwarded by Device 4 and Device 5 for Command Parameter 2 are consistent (both are 208.789 seconds), the existing method can immediately determine that the feature event has occurred, and the occurrence time is 208.789 seconds. However, according to the discrimination method proposed in this invention, since both are real-time sources forwarded by different devices for the same command parameter, their similarity only indicates that the feature event has occurred, but does not meet the discrimination condition that the feature event occurs immediately. Therefore, this measurement value cannot be immediately taken as the occurrence time of the feature event. Similarly, when the flight time is 218.669 seconds (208.669 seconds + 10 seconds), although there are multiple consistent measurements in the data source, they are either real-time sources forwarded by different devices for the same command parameter, or real-time and delayed sources forwarded by the same device for the same command parameter. Neither of these satisfies the discrimination condition that the feature event occurs immediately. By the time the flight duration reached 222.789 seconds (212.789 seconds + 10 seconds), all data sources for this feature event had been resolved, and the Bayesian estimation algorithm yielded the following result: Figure 2 The probability distribution diagram of feature event 2 shown is obtained by... Figure 2 It can be seen that since the probability of the characteristic event occurring at 208.669 seconds is 30.93%, which is the measurement value with the highest probability of the characteristic event occurring among all data sources, the occurrence time of characteristic event 2 is 208.669 seconds. Clearly, the discrimination result of the method proposed in this invention for characteristic event 2 is consistent with that of existing methods.

[0095] As can be seen from feature event 3 in Table 1, the information source of this feature event contains only one instruction parameter, which does not meet the discrimination condition that the feature event occurs immediately. Therefore, the feature event can only be discriminated using a Bayesian estimation algorithm. Table 1 also shows that there are two sets of data source measurements that are consistent in this feature event (563.088 seconds and 567.656 seconds), and the number of consistent data sources is the same. According to the existing method, when the flight time is 563.089 seconds, the feature event can be immediately determined to have occurred, and the occurrence time of the feature event is 563.088 seconds. However, through… Figure 3 The probability distribution chart for feature event 3 shows that the probability of the feature event occurring at a measurement value of 563.088 seconds is 39.46%, while the probability at a measurement value of 570.302 seconds is 40.68%, both higher than the probability at 563.088 seconds and the highest among all data sources. Therefore, the occurrence time of feature event 3 is 570.302 seconds. Actual results confirm that feature event 3 occurred at 570.302 seconds, consistent with the determination result of the method proposed in this invention.

[0096] The analysis of the three typical examples above shows that the method proposed in this invention has a higher discrimination accuracy compared with existing methods.

Claims

1. A real-time automatic identification method for characteristic events of launch vehicles based on Bayesian estimation, characterized in that: Includes the following steps: (1) Preprocess the measured values ​​from the current data source using the method: y = x + Δt t +Δt d Where y is the result after preprocessing the data source, x is the measured value of the data source, and Δt t Δt represents the theoretical time deviation between the instructions represented by the data source and the characteristic events. d This is due to the inherent delay bias of the delay source; (2) Cache the preprocessed results of the current data source; (3) Based on the upper limit of residual error δy, perform consistency judgment on all data sources in the cache and group them. The upper limit of residual error refers to the maximum value of the total error caused by different instruction calculation methods and different delay errors caused by different transmission channels of different devices. The method is as follows: (3.1) Arrange the preprocessed results of all data sources in ascending order to obtain the dataset D = {y1, y2, ... y k …y g …y n }; (3.2) Starting from y1, connect each of y2 to y1 one by one. n Comparison, when y2~y n When the difference between y1 and y2 is less than δy, it is determined that it is consistent with y1, that is, the measurement values ​​are the same, and it is grouped with y1 to obtain the first dataset D1. (3.3) Then from y k At the beginning, y k This represents the minimum value in dataset D that was not assigned to the first dataset D1 along with y1, and is calculated by comparing each value with y1. k+1 ~y n Comparison, when y k+1 ~y n With y k When the difference is less than δy, determine its relationship with y. k Consistent, meaning the measured values ​​are the same, is compared with y. k They are grouped together to obtain the second dataset D2; (3.4) Then from y g At the beginning, y g This indicates that data in dataset D does not contain data that is not related to y. k The minimum value assigned to the second dataset D2 is compared with y one by one. g+1 ~y n Comparison, when y g+1 ~y n With y g When the difference is less than δy, determine its relationship with y. g Consistent, meaning the measured values ​​are the same, is compared with y. g They are grouped together to obtain the third dataset D3; This process continues until all data sources are grouped, resulting in a grouped dataset D = {D1, D2, ..., D...}. m }, where m represents the number of groups in the dataset; (4) Determine the occurrence time of the feature event based on the mutual corroboration conditions between data sources; (4.1) The conditions for mutual corroboration are: 1) When the real-time source and the delayed source measurement values ​​are consistent and different instruction parameters are forwarded by different devices in the data source, it indicates that both the information source and the transmission channel are reliable; 2) When real-time source measurements with different instruction parameters forwarded by different devices are consistent, it indicates that both the information source and the transmission channel are reliable; 3) When the delay measurement values ​​of different instruction parameters forwarded by different devices in the data source are consistent, it indicates that both the information source and the transmission channel are reliable; 4) When real-time and delayed source measurements with different instruction parameters forwarded by the same device are consistent, it indicates that both the information source and the transmission channel are reliable. 5) When real-time source measurements with different instruction parameters forwarded by the same device are consistent, it indicates that the information source is trustworthy; 6) When different instruction parameters in the data source are forwarded by the same device and the measured delay values ​​are consistent, it indicates that the information source is trustworthy; 7) When the real-time source and the delayed source in the data source have the same instruction parameter forwarded by the same device and their measurement values ​​are consistent, it indicates that the transmission channel is reliable; 8) When the real-time source measurement values ​​of the same instruction parameter forwarded by different devices in the data source are consistent, it indicates that the transmission channel is reliable; 9) When the delay measurement values ​​of the same instruction parameter forwarded by different devices in the data source are consistent, it indicates that the transmission channel is reliable; 10) When the real-time source and the delayed source measurements of the same instruction parameter forwarded by different devices in the data source are consistent, it indicates that the transmission channel is reliable; (4.2) The specific method is as follows: (4.2.1) When the data source satisfies the mutual verification conditions 1) to 4), it indicates that both the information source and the transmission channel are reliable, and the occurrence of the characteristic event can be determined immediately. The occurrence time of the characteristic event is the minimum measurement value when the data source is consistent. (4.2.2) When all data sources for the feature event have solved or reached the maximum theoretical deviation limit of the event occurrence time, and the data sources satisfy the mutual verification conditions 5) to 10), it indicates that either the information source is reliable or the transmission channel is reliable, and the feature event can be determined to have occurred. The occurrence time of the feature event is determined by the feature event discrimination algorithm based on Bayesian maximum a posteriori estimation. The specific method is as follows: (4.2.2.1) Calculate the data source as y i The prior probability density value p(y) i The formula is: Where μ is the theoretical time of occurrence of the characteristic event, and 3σ represents the maximum theoretical deviation of the occurrence time of the characteristic event; (4.2.2.2) Calculate the time when the characteristic event occurs is y. i At that time, each y in dataset D j probability density value p(y) j |y i The formula is: (4.2.2.3) Calculate the occurrence time of the characteristic event as y i The probability density value p(D|y) of dataset D. i The formula is: (4.2.2.4) Calculate the occurrence time of the characteristic event as y. i The relative probability density value p(y) at time i |D), the formula is: p(y i |D)=p(y i )p(D|y i ) (4.2.2.5) Following steps (4.2.2.1) to (4.2.2.4), calculate the relative probability density value of each data source in dataset D; (4.2.2.6) Calculate the y of each data source in dataset D. i The relative probability P(y) of the characteristic event i |D), the formula is: (4.2.2.7) Calculate the value of each data source D in dataset D. l The relative probability P(D) of the characteristic event l |D), the formula is: Among them, D l Indicates data source y s ~y e The dataset is grouped into the l-th dataset; (4.2.2.8) From {P(d1|d), P(D2|D), ..., P(d l |D)…,P(D m |d)} Select the dataset with the highest probability value as the valid dataset, and the occurrence time of the feature event is the minimum measurement value in the valid dataset; (4.2.3) When all data sources for the characteristic event have solved or reached the maximum theoretical deviation limit of the occurrence time of the characteristic event, if the data sources do not meet the verification conditions, it is determined that the characteristic event has not occurred. (5) When a new data source arrives, repeat steps (1) to (4) until a decision is reached.

Images