Probability-based fusion weight processing method and device for multiple monitoring source targets

Abstract

The embodiment of the invention discloses a probability-based fusion weight processing method and device for multiple monitoring source targets, and the method comprises the steps: carrying out the processing and confidence evaluation of each single monitoring source data, so as to obtain the confidence rate of a corresponding single monitoring source; according to the measurement result of each single monitoring source, further transverse comparison is carried out, and the consistency probability between the single monitoring sources is calculated, so that the consistency evaluation of the multiple monitoring sources is realized; combining the confidence rate of the single monitoring source with the consistency probability, calculating a comprehensive probability, and carrying out normalization processing to obtain a fusion weight so as to ensure that the weights of the data of different monitoring sources are reasonably distributed during fusion, and carrying out position fusion; the method has the beneficial effects that the fusion weight is dynamically adjusted by combining the confidence coefficient of a single monitoring source and the consistency evaluation among multiple monitoring sources, and the accuracy and the robustness of a fusion result are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of air traffic control automation monitoring, and particularly relates to a method and device for processing fusion weights of multi-surveillance source targets based on probability. Background Art

[0002] In the modern civil aviation field, Air Traffic Management (ATM) relies on an air traffic control automation system (ATCAS) to ensure the safety and efficiency of airspace operations. The air traffic control automation system is the core information processing system of air traffic control. It can receive, process, and fuse various surveillance data, provide real-time air traffic situation information for air traffic controllers, and assist in decision-making and command and dispatch. This system is widely used in key functions such as flight scheduling, conflict detection and warning, and flight trajectory prediction, and is an important technical support for ensuring aviation safety and improving operational efficiency.

[0003] In the air traffic control automation system, multi-sensor surveillance data fusion is a key technology for improving the accuracy and stability of target surveillance. Multi-surveillance sources include various detection means such as Primary Surveillance Radar (PSR), Secondary Surveillance Radar (SSR), Automatic Dependent Surveillance-Broadcast (ADS-B), and Wide Area Multilateration (WAM). These surveillance devices have their own advantages and disadvantages: The primary radar does not rely on on-board equipment but is greatly affected by weather; the secondary radar relies on on-board transponders but is affected by interference; ADS-B provides high-precision position information but the data quality may be affected by external factors; while WAM measures the time difference through multiple receiving stations for positioning and is suitable for precise surveillance in complex airspaces.

[0004] Due to problems such as observation errors and data loss in a single surveillance source, how to effectively fuse target information from different surveillance sources to improve tracking accuracy, reduce errors, and enhance the robustness of the system is the focus of current research and application of air traffic control automation systems. Therefore, the optimization of fusion methods, especially the calculation of multi-surveillance source target fusion weights based on probability, has become an important means to improve the performance of air traffic control surveillance systems.

[0005] Currently, most traditional technologies adopt technical solutions of static weight methods or dynamic weighting methods. However, traditional technologies have the following disadvantages:

[0006] 1. The static weight method is simple to calculate and has a relatively low implementation cost, making it suitable for scenarios where the monitoring environment is relatively stable. However, it has the problem of poor adaptability and is difficult to cope with a real-time changing measurement environment. For example, when the performance of a certain monitoring source changes due to factors such as weather, occlusion, or equipment failure, the fixed weight may lead to an increase in errors. In addition, the static weight method has poor robustness to outliers and is easily affected by individual measurement errors, thereby reducing the accuracy of track fusion.

[0007] 2. The dynamic weighted average method lacks accuracy, and the weighting coefficient cannot fully reflect the confidence level of the measurement signal of a single monitoring source. The fusion process mainly depends on the quality of the measurement values of a single monitoring source, while ignoring the consistency evaluation among the measurements of multiple monitoring sources, resulting in poor stability of the fused data of the monitoring sources. SUMMARY OF THE INVENTION

[0008] In view of the technical deficiencies of the prior art, the purpose of the embodiments of the present invention is to provide a method and device for processing fusion weights of multi-monitoring source targets based on probability.

[0009] To achieve the above purpose, in a first aspect, the embodiments of the present invention provide a method for processing fusion weights of multi-monitoring source targets based on probability, the method comprising:

[0010] Processing and confidence evaluation are performed on each single monitoring source data to obtain the confidence rate of the corresponding single monitoring source;

[0011] According to the measurement results of each single monitoring source, further horizontal comparison is performed to calculate the consistency probability between them, so as to achieve the consistency evaluation of multiple monitoring sources;

[0012] The single monitoring source confidence rate and the consistency probability are combined to calculate the comprehensive probability, and normalization processing is performed to obtain the fusion weight, so as to ensure reasonable distribution of weights for the data of different monitoring sources during fusion, and perform position fusion based on this.

[0013] As a preferred implementation manner of the present application, the method further comprises: performing filtering and smoothing processing on the fused position, and optimizing the speed of the target to make the track information more stable and reliable.

[0014] As a specific implementation manner of the present application, the processing and confidence evaluation of each single monitoring source data specifically comprise:

[0015] First, interactive multi-model filtering is performed on the measurement value of each independent monitoring source to improve the accuracy of target state estimation;

[0016] Then, by calculating the Mahalanobis distance and chi-square distribution probability of the measurement residual, the credibility of the measurement value of this monitoring source is evaluated, and the confidence rate of the corresponding single monitoring source is obtained.

[0017] As a preferred implementation manner of the present application, before the consistency evaluation, linear interpolation is also used for time alignment.

[0018] As a specific implementation manner of the present application, the consistency evaluation of multiple monitoring sources specifically includes:

[0019] Analyze the measurement deviations of different monitoring sources for the same target, and combine statistical methods to calculate the consistency probability between multiple monitoring sources, and evaluate whether the data of each monitoring source is consistent within a preset error range.

[0020] As a specific implementation manner of the present application, the position fusion specifically includes:

[0021] Combine the confidence rate of a single monitoring source and the consistency probability to obtain the comprehensive probability of each monitoring source;

[0022] Then normalize the comprehensive probability to obtain the fusion weight;

[0023] Finally, weight the position information of each monitoring source according to the normalized weight, perform position fusion, and obtain the fused position.

[0024] As a specific implementation manner of the present application, the fused track is smoothed by Kalman filtering, and the speed of the target is optimized by using the moving average smoothing algorithm.

[0025] In a second aspect, an embodiment of the present invention further provides a fusion weight processing device for a multi-monitoring source target based on probability. The device includes:

[0026] A confidence module, configured to process and evaluate the confidence of each single monitoring source data to obtain the confidence rate of the corresponding single monitoring source;

[0027] An evaluation module, configured to further compare horizontally according to the measurement results of each single monitoring source, and calculate the consistency probability between them to achieve the consistency evaluation of multiple monitoring sources;

[0028] A fusion module, configured to combine the confidence rate of a single monitoring source and the consistency probability, calculate the comprehensive probability, and perform normalization processing to obtain the fusion weight, so as to ensure reasonable distribution of weights when fusing the data of different monitoring sources, and perform position fusion based on this.

[0029] As a preferred implementation manner of the present application, it further includes an optimization module, configured to perform filtering and smoothing processing on the fused position, and optimize the speed of the target to make the track information more stable and reliable.

[0030] As a preferred implementation manner of the present application, it further includes an alignment module, and the alignment module is used to perform time alignment by using linear interpolation before the consistency evaluation.

[0031] The technical solution provided by the embodiments of the present invention processes and evaluates the confidence of each single monitoring source data first, and then dynamically adjusts the fusion weight by combining the single monitoring source measurement confidence and the consistency evaluation among multiple monitoring sources, improving the accuracy and robustness of the fusion result; at the same time, optimizing the target tracking improves the continuity and stability of the estimated value of the fusion track speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art.

[0033] Figure 1 is a flowchart of a method for processing the fusion weight of a multi-monitoring source target based on probability provided by the embodiments of the present invention;

[0034] Figure 2 is a schematic diagram of the process of a method for processing the fusion weight of a multi-monitoring source target based on probability provided by the embodiments of the present invention;

[0035] Figure 3 is a schematic structural diagram of a device for processing the fusion weight of a multi-monitoring source target based on probability provided by the embodiments of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0037] It should be understood that when used in this specification and the appended claims, the terms "comprises" and "comprising" indicate the presence of the described features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or their groups.

[0038] Throughout the specification, the reference to "an embodiment", "embodiment", "an example" or "example" means that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present invention. Thus, the phrases "in an embodiment", "in the embodiment", "an example" or "example" appearing throughout the specification do not necessarily all refer to the same embodiment or example. Additionally, the particular features, structures, or characteristics may be combined in any suitable combination and / or sub-combination in one or more embodiments or examples.

[0039] It should be noted that, unless otherwise specified, the technical terms in this embodiment have the usual meanings understood in the technical field to which they belong.

[0040] Please refer to Figure 1 , a method for processing the fusion weight of a multi-surveillance source target based on probability provided by an embodiment of the present invention, the method includes:

[0041] S101, process and evaluate the confidence of each single surveillance source data to obtain the confidence rate of the corresponding single surveillance source.

[0042] In this embodiment, the process of processing and evaluating the confidence of each single surveillance source data specifically includes:

[0043] First, perform interacting multiple model filtering on each independent surveillance source measurement value to improve the accuracy of target state estimation;

[0044] Then, evaluate the credibility of the surveillance source measurement value by calculating the Mahalanobis distance and chi-square distribution probability of the measurement residual, and obtain the confidence rate of the corresponding single surveillance source; among them, the higher the confidence rate, the higher the credibility of the surveillance source data.

[0045] Specifically, interacting multiple model filtering (IMM, Interacting Multiple Model), IMM filtering is composed of multiple Kalman filter models, including a constant velocity straight line model CV, a constant acceleration straight line model CA, a constant velocity left turn model CT1, and a constant velocity right turn model CT2;

[0046] Mahalanobis Distance is an index for measuring the similarity between multi-dimensional data. Different from the Euclidean distance, it takes into account the correlation and distribution of the data and can more effectively distinguish outliers. In this solution, the Mahalanobis distance is used to evaluate the deviation between the surveillance source measurement value and the true position of the target, and the larger the distance, the greater the measurement error.

[0047] The chi-square distribution is a statistical distribution, which is often used in hypothesis testing. In this solution, the chi-square probability is used to calculate the reasonableness of the measurement error. If the measurement value of a certain surveillance source has a large deviation, its corresponding chi-square probability will be low, thereby reducing its weight in the fusion calculation.

[0048] Confidence rate calculation

[0049] The difference between each group of measurement values and predicted values can reflect the reliability of the measurement data, which is represented by the confidence rate. Confidence rate calculation based on filtering residuals:

[0050] The residual y (the difference between the predicted value and the measured value) and covariance P after IMM filtering y , calculate the squared Mahalanobis distance

[0051]

[0052] Squared Mahalanobis distance Conforms to the chi-square distribution, and the confidence level P can be calculated accordingly a :

[0053]

[0054] Result comparison: If the confidence level is less than the threshold (default is 5%, configurable), it is considered an outlier and is removed; by removing the measured values with low confidence rates, the influence of outliers on the fusion result is reduced.

[0055] S102. According to the measurement results of each single monitoring source, further compare horizontally and calculate the consistency probability between them to achieve the consistency evaluation of multiple monitoring sources.

[0056] Furthermore, referring to Figure 2 , during implementation, before the consistency evaluation, linear interpolation is also used for time alignment;

[0057] It should be noted that the fused track is output at a fixed period. Each time of fusion only selects the latest received data of each single monitoring source, and the time difference does not exceed T + 2 (T is the fusion period); the data with an expired time does not participate in the fusion; time alignment can improve the accuracy and real-time performance during fusion.

[0058] In this embodiment, the consistency evaluation of the multiple monitoring sources specifically includes:

[0059] Analyze the measurement deviations of different monitoring sources for the same target, and combine statistical methods to calculate the consistency probability between multiple monitoring sources to evaluate whether the data of each monitoring source is consistent within a preset error range.

[0060] That is, the obtained consistency probability is the credibility calculated by horizontally comparing the measurement results of different monitoring sources for the same target;

[0061] The consistency evaluation makes a horizontal comparison of the measured values of multiple monitoring sources for the same track at the same time point. The farther the measured position of the track is from the mean position, the lower the credibility of the measured value, and vice versa.

[0062] Based on the calculation of the consistency probability of multiple monitoring sources:

[0063] After the multi-monitoring source data is separately filtered, the respective state estimations are obtained. These estimations are compared horizontally, and the consistency probability between the multi-monitoring sources is calculated.

[0064] Calculate the mean X of the multi-monitoring sources m And the covariance Cov, where N represents the number of monitoring sources, and X i Is the state estimation value of each monitoring source. Assume X i Complies with the multi-dimensional normal distribution, then the probability of each group of state values can be estimated through the difference between the mean and the state value and the covariance.

[0065] Mean:

[0066]

[0067] Covariance:

[0068]

[0069] Square of the Mahalanobis distance

[0070]

[0071] Chi-square distribution probability P b :

[0072]

[0073] In S103, combine the single-monitoring source confidence rate and the consistency probability, calculate the comprehensive probability, and perform normalization processing to obtain the fusion weight, so as to ensure reasonable distribution of the weights of the data of different monitoring sources during fusion, and perform position fusion based on this.

[0074] In this embodiment, the position fusion specifically includes:

[0075] Combine the single-monitoring source confidence rate and the consistency probability to obtain the comprehensive probability of each monitoring source;

[0076] Then normalize the comprehensive probability to obtain the fusion weight;

[0077] Finally, weight the position information of each monitoring source according to the normalized weight, perform position fusion, and obtain the fusion position.

[0078] Specifically, the position fusion combines the confidence rate P a Of the measurement data and the consistency evaluation probability P b Calculate the weight (P i =P a *P b Of the position fusion, and then perform normalization calculation on P i , and perform position fusion based on this.

[0079] Probability-based fusion scheme:

[0080] Combining the confidence rate of a single monitoring source and the consistency probability to obtain the comprehensive probability of each monitoring source:

[0081] P i = P a * P b

[0082] Normalize P i to obtain the final fusion weight:

[0083]

[0084] Weight the position information of each monitoring source according to the normalized weight to obtain the fused position:

[0085]

[0086] Adjust the fusion weight by combining the consistency probability to improve the reliability of data fusion;

[0087] Combine the confidence rate of a single monitoring source and the consistency probability to dynamically calculate the fusion weight; and

[0088] Adopt a normalization method to ensure reasonable weight allocation, optimize the final fusion result, and achieve adaptive and high-precision fusion weight adjustment.

[0089] Furthermore, the method further includes:

[0090] Perform filtering and smoothing on the fused position, and optimize the speed of the target to improve the continuity and stability of the estimated value of the fused track speed, making the track information more stable and reliable.

[0091] It should be noted that a track is a continuous tracking model of the real-time dynamics of an aircraft by the air traffic control system, integrating position, speed, and predicted path to provide a uniquely identified flight track;

[0092] The fused positions at all time points generated by the foregoing steps constitute the position track information of the track.

[0093] In this embodiment, the fused track (which can also be understood as the fused position) is smoothed by Kalman filtering, and the speed of the target is optimized using the Moving Average Smoothing algorithm.

[0094] It should be noted that when performing smoothing, the sliding window calculates the average value of the nearest n points.

[0095] In the above solution, by first processing and evaluating the confidence of each single monitoring source data, and then combining the measurement confidence of the single monitoring source and the consistency evaluation among multiple monitoring sources, the fusion weight is dynamically adjusted to improve the accuracy and robustness of the fusion result; at the same time, the target tracking is optimized, and the continuity and stability of the estimated value of the fusion track speed are improved.

[0096] Based on the same inventive concept, an embodiment of the present invention further provides a device for processing the fusion weight of a multi-monitoring source target based on probability. Referring to Figure 3 , the device includes:

[0097] A confidence module for processing and evaluating the confidence of each single monitoring source data to obtain the confidence rate of the corresponding single monitoring source;

[0098] An evaluation module for further comparing horizontally according to the measurement results of each single monitoring source and calculating the consistency probability between them to achieve the consistency evaluation of multiple monitoring sources;

[0099] A fusion module for combining the single monitoring source confidence rate and the consistency probability, calculating the comprehensive probability, and performing normalization processing to obtain the fusion weight to ensure reasonable distribution of the weights of data from different monitoring sources during fusion, and performing position fusion based on this.

[0100] Based on the above technical solution, the device for processing the fusion weight of a multi-monitoring source target based on probability further includes an optimization module for performing filtering and smoothing processing on the fused position and optimizing the speed of the target to improve the continuity and stability of the estimated value of the fusion track speed, making the track information more stable and reliable.

[0101] Further, the device for processing the fusion weight of a multi-monitoring source target based on probability further includes an alignment module, and the alignment module is used to perform time alignment by linear interpolation before the consistency evaluation.

[0102] In this embodiment, the processing and confidence evaluation of each single monitoring source data specifically include:

[0103] First, perform interactive multi-model filtering on each independent monitoring source measurement value to improve the accuracy of target state estimation;

[0104] Then, evaluate the credibility of the monitoring source measurement value by calculating the Mahalanobis distance and chi-square distribution probability of the measurement residual, and obtain the confidence rate of the corresponding single monitoring source.

[0105] The consistency evaluation of the multiple monitoring sources specifically includes:

[0106] Analyze the measurement deviations of the same target from different monitoring sources, calculate the consistency probability among multiple monitoring sources by combining statistical methods, and evaluate whether the data of each monitoring source is consistent within the preset error range.

[0107] That is, the confidence level is the probability calculated based on the residual value of the measurement result and the prediction result of a single monitoring source; the consistency probability obtained by the evaluation module is the credibility calculated by horizontally comparing the measurement results of different monitoring sources for the same target; and finally, the fused weight is obtained by integrating two independent and uncorrelated probabilities.

[0108] Further, the position fusion specifically includes:

[0109] Combine the single-monitoring-source confidence rate and the consistency probability to obtain the comprehensive probability of each monitoring source;

[0110] Then normalize the comprehensive probability to obtain the fused weight;

[0111] Finally, weight the position information of each monitoring source according to the normalized weight, perform position fusion, and obtain the fused position.

[0112] It should be noted that for a more specific description of the working process of the device embodiment, please refer to the foregoing method embodiment section and will not be elaborated here.

[0113] The entire solution improves the accuracy and stability of track fusion by adopting a combination of the single-monitoring-source confidence rate and multi-monitoring-source consistency evaluation;

[0114] At the same time, by combining the Kalman filter and the smoothing algorithm, the continuity and stability of the estimated value of the fused track speed are improved; the track speed smoothness is improved, and target tracking is optimized; it overcomes the defect that the traditional solution often lacks targeted optimization in track smoothing, resulting in jitter in speed estimation.

[0115] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for processing the fusion weights of multi-surveillance source targets based on probability, characterized in that, The method includes: Processing and confidence evaluation are performed on each single monitoring source data to obtain the confidence rate of the corresponding single monitoring source; According to the measurement results of each single monitoring source, further horizontal comparison is carried out to calculate the consistency probability between them, so as to realize the consistency evaluation of multiple monitoring sources; The confidence rate of the single monitoring source and the consistency probability are combined to calculate the comprehensive probability, and normalization processing is performed to obtain the fusion weight, so as to ensure the reasonable distribution of weights when data from different monitoring sources are fused, and position fusion is carried out based on this.

2. The method for processing the fusion weight of a multi-surveillance source target based on probability according to claim 1, wherein The method further includes: Filtering and smoothing the fused position, and optimizing the speed of the target to make the track information more stable and reliable.

3. The method for processing the fusion weight of a multi-surveillance-source target based on probability according to claim 1, wherein The processing and confidence evaluation of each single monitoring source data specifically include: First, interactive multi-model filtering is performed on the measurement values of each independent monitoring source to improve the accuracy of target state estimation; Then, by calculating the Mahalanobis distance and chi-square distribution probability of the measurement residual, the credibility of the measurement value of this monitoring source is evaluated to obtain the confidence rate of the corresponding single monitoring source.

4. A method for processing fusion weights of a multi-surveillance-source target based on probability, as claimed in claim 1 or 2, wherein Before the consistency evaluation, linear interpolation is also used for time alignment.

5. The method for processing the fusion weight of a multi-surveillance source target based on probability according to claim 4, wherein The consistency evaluation of the multiple monitoring sources specifically includes: Analyze the measurement deviation of different monitoring sources for the same target, and combine statistical methods to calculate the consistency probability between multiple monitoring sources, and evaluate whether the data of each monitoring source is consistent within the preset error range.

6. The method for processing the fusion weight of a multi-surveillance source target based on probability according to claim 1, wherein, The position fusion specifically includes: Combining the confidence rate of the single monitoring source and the consistency probability to obtain the comprehensive probability of each monitoring source; Then normalize the comprehensive probability to obtain the fusion weight; Finally, weight the position information of each monitoring source according to the normalized weight, perform position fusion, and obtain the fused position.

7. The method for processing the fusion weight of a multi-surveillance source target based on probability according to claim 2, wherein, The fused track is smoothed by Kalman filtering, and the speed of the target is optimized by using the moving average smoothing algorithm.

8. A fusion weight processing device for multi-surveillance source targets based on probability, characterized in that, The device includes: A confidence module, which is used for processing and confidence evaluation of each single monitoring source data to obtain the confidence rate of the corresponding single monitoring source; An evaluation module, which is used for further horizontal comparison according to the measurement results of each single monitoring source, calculating the consistency probability between them, so as to realize the consistency evaluation of multiple monitoring sources; A fusion module, which is used for combining the confidence rate of the single monitoring source and the consistency probability, calculating the comprehensive probability, and performing normalization processing to obtain the fusion weight, so as to ensure the reasonable distribution of weights when data from different monitoring sources are fused, and performing position fusion based on this.

9. The fusion weight processing device for a multi-surveillance source target based on probability according to claim 8, wherein, It further includes an optimization module, which is used for filtering and smoothing the fused position, and optimizing the speed of the target to make the track information more stable and reliable.

10. A fusion weight processing device for a multi-surveillance source target based on probability according to claim 8 or 9, characterized in that It further includes an alignment module, and the alignment module is used for performing time alignment by using linear interpolation before the consistency evaluation.

Images