Target data real-time detection and smoothing method and system based on error estimation
By employing a three-level hierarchical selection and adaptive smoothing strategy, the problem of insufficient identification of continuous outliers in existing technologies is solved, achieving efficient smoothing and improved accuracy of trajectory data, and enhancing the robustness and adaptability of the system.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-18
- Publication Date
- 2026-04-10
AI Technical Summary
Existing methods struggle to effectively identify and process "spot-like" outliers that are continuous and time-dependent, which compromises the smoothness and accuracy of trajectory data. Furthermore, when measurement data experiences jumps or drastic changes, the traditional five-point linear compensation model may introduce new deviation points, further exacerbating the distortion of trajectory data.
A three-level hierarchical selection mechanism is adopted, combined with the sampling interval variation coefficient judgment, and an adaptive smoothing strategy of equal interval central smoothing and non-equal interval least squares fitting is used to identify and remove continuous outliers, and adaptive processing is performed in abnormal cases.
It significantly improves the smoothness and accuracy of trajectory data, enhances the robustness and adaptability of trajectory data processing, and can effectively cope with abnormal situations in measurement data.
Smart Images
Figure CN121834091A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of data processing, in particular to a target data real-time selection and smoothing method and system based on error estimation. BACKGROUND
[0002] In the scenarios of space monitoring, flight control, radar detection, etc., accurate and continuous acquisition of target flight trajectory data is a key prerequisite for ensuring stable operation and accurate decision-making of the system. Due to the differences in measurement accuracy, time synchronization mechanism, processing capacity and environmental conditions, various sensors often lead to abnormal phenomena such as outliers, systematic errors and random errors in target data. These outliers directly affect the authenticity and continuity of the data, especially in time-height, time-velocity and other trajectory characteristic curves, which show obvious jumps, jitter and even incorrect trends, seriously interfering with the accuracy of subsequent data fusion, state estimation and target prediction. Therefore, it is of great practical value and theoretical significance to carry out real-time error identification and smoothing processing research on measured trajectory data.
[0003] To improve the quality of target trajectory data, existing real-time data processing systems generally introduce an outlier identification and rejection module in the preprocessing stage. A typical method, such as a processing framework jointly implemented by a difference-based model selection filter and a five-point linear extrapolation selector, identifies sudden and non-smooth points as potential outliers through fourth-order difference to make a reasonable judgment on the sampling sequence, and uses a five-point linear prediction model constructed from adjacent points to replace and compensate for abnormal points. This method combines the local stationarity assumption of time series, and realizes real-time rejection and repair of isolated outliers without significantly increasing the computational burden, and has good performance in general engineering applications.
[0004] However, the existing method relies on fourth-order difference and linear extrapolation strategy, mainly designed for isolated outliers, and is difficult to effectively identify and process "spot-type" outliers with continuity and time correlation, which can easily damage the overall smoothness and accuracy of the trajectory; in addition, in the case of sudden jumps, drastic mutations or large-scale drifts in measured data, the traditional five-point linear compensation model often predicts distortion, which not only cannot effectively correct errors, but also may introduce new bias points, further exacerbating the distortion of trajectory data. SUMMARY
[0005] In order to solve the problems that the existing method relies on four-order difference and linear extrapolation strategy, mainly aims at isolated wild value design, it is difficult to effectively identify and process "spot type" wild value with continuity and time correlation, and easily leads to the damage of overall smoothness and accuracy of the trajectory; in addition, in the case that the measured data jumps, sharply mutates or widely drifts, the traditional five-point linear compensation model often predicts distortion, which not only cannot effectively correct the error, but also may introduce new deviation points, further aggravating the distortion of trajectory data, the application provides a target data real-time selection and smoothing method and system based on error estimation.
[0006] The technical scheme provided by the embodiments of the application is as follows: The first aspect is: The target data real-time selection and smoothing method based on error estimation provided by the embodiments of the application comprises the following steps: S1: acquiring flight trajectory data comprising time, position, speed and height; S2: performing trajectory abnormal value selection processing on the flight trajectory data through a three-level hierarchical selection manner to obtain target trajectory data; S3: calculating a sampling interval variation coefficient of the target trajectory data, and judging whether the sampling interval variation coefficient is less than a preset sampling interval variation coefficient; if yes, determining that the target trajectory data is equidistant sampling data, and entering S4; otherwise, determining that the target trajectory data is non-equidistant sampling data, and entering S5; S4: performing smoothing processing on the equidistant sampling target trajectory data through an equidistant center smoothing processing manner, and entering S6; S5: performing smoothing processing on the non-equidistant sampling target trajectory data through a non-equidistant least square fitting manner, and entering S6; S6: saving the smoothed target trajectory data to a trajectory processing result cache area through a multi-layer cache mechanism.
[0007] The second aspect is: The target data real-time selection and smoothing system based on error estimation provided by the embodiments of the application comprises: A processor; A memory, wherein the memory stores computer readable instructions, and the computer readable instructions are executed by the processor to realize the target data real-time selection and smoothing method based on error estimation as described in the first aspect.
[0008] The third aspect is: The computer readable storage medium provided by the embodiments of the application stores a computer program, and the program is executed by the processor to realize the target data real-time selection and smoothing method based on error estimation as described in the first aspect.
[0009] The technical scheme provided by the embodiment of the present application has at least the following beneficial effects: In the embodiment of the present application, by introducing a three-level hierarchical selection mechanism, the limitation that the traditional fourth-order difference and linear extrapolation method can only process isolated wild values is effectively overcome, and the “spot type” wild value with continuity and time correlation can be identified and removed, which significantly improves the smoothness and accuracy of the trajectory data. At the same time, combined with the judgment mechanism of the sampling interval coefficient of variation, the adaptive smoothing strategy of equal interval center smoothing and non-equal interval least square fitting is used according to the sampling characteristics of the data, which effectively deals with the distortion of the traditional model under abnormal conditions such as jump, sharp mutation or large range drift of the measurement data, and enhances the robustness and adaptability of the trajectory smoothing processing. BRIEF DESCRIPTION OF DRAWINGS
[0010] In order to more clearly illustrate the technical scheme in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.
[0011] Figure 1 A flowchart of a target data real-time selection and smoothing method based on error estimation provided by the embodiment of the present application is shown in the figure. Figure 2 A structure diagram of a target data real-time selection and smoothing system based on error estimation provided by the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION
[0012] The technical scheme in the present application will be described below with reference to the drawings.
[0013] In the embodiments of the present application, the words “example”, “for example” and the like are used to represent an example, illustration or description. Any embodiment or design scheme described as “example” in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word “example” is intended to present the concept in a specific way. In addition, in the embodiments of the present application, the meaning expressed by “and / or” can be both, or can be one of the two.
[0014] In the embodiments of the present application, “image” and “picture” can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent. “Of”, “corresponding” and “corresponding” can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent.
[0015] In the embodiments of the present application, sometimes the subscript such as W1 may be written in the form of non-subscript such as W1, and the meanings expressed thereby are consistent when the difference is not emphasized.
[0016] In order to make the technical problems, technical solutions and advantages to be solved by the present application more clear, the following will be described in detail with reference to the drawings and specific embodiments.
[0017] Reference is made to the accompanying drawings and specific embodiments described in the specification Figure 1 , a flowchart of a target data real-time selection and smoothing method based on error estimation provided by an embodiment of the present application is shown.
[0018] The embodiment of the present application provides a target data real-time selection and smoothing method based on error estimation, which can be realized by a target data real-time selection and smoothing device based on error estimation. The target data real-time selection and smoothing device based on error estimation can be a terminal or a server. The processing flow of the target data real-time selection and smoothing method based on error estimation can include the following steps:
[0019] S1: Obtain flight trajectory data including time, position, speed and height.
[0020] S2: Perform trajectory outlier selection processing on the flight trajectory data by a three-level hierarchical selection method to obtain target trajectory data.
[0021] The three-level hierarchical selection is a multi-level discrimination method for identifying outliers in flight trajectory data, which performs data filtering from three levels of single point measurement rationality, front and rear frame dynamic consistency and multi-frame trend fitting in turn.
[0022] It should be noted that this hierarchical structure can realize joint discrimination and processing of isolated and string type abnormal data, and significantly improve the quality of trajectory data and the stability of subsequent processing.
[0023] In one possible implementation, S2 specifically includes: S201: Perform single point measurement rationality selection on the flight trajectory data to determine whether the time, position, speed and height measurement of each flight trajectory data meets a first preset condition. If yes, the flight trajectory data is retained and the next step is entered. Otherwise, the flight trajectory data is rejected.
[0024] The first preset condition is specifically: Wherein, t denotes the sampling time, x denotes the target flight position vector, denotes the module length, denotes the upper limit of the reasonable value of the target flight position, Represents the target velocity vector. This indicates the upper limit of a reasonable value for the target's flight speed. h Indicates flight altitude. Indicates the minimum flight altitude. Indicates the maximum flight altitude.
[0025] Optionally, the upper limit of the reasonable value of the target flight position is generally set to , = 16700, the target flight altitude range is generally set to 0. rice.
[0026] In this embodiment of the invention, by setting a first preset physical constraint condition for measuring elements such as time, position, speed and altitude, obviously unreasonable single-point abnormal data can be quickly eliminated, such as exceeding the flight boundary, abnormal speed change or altitude drift, thereby effectively reducing the subsequent computational burden in the initial stage of data processing and improving the overall basic reliability and screening efficiency of trajectory data.
[0027] S202: After the single-point measurement element rationality check, the remaining flight trajectory data undergoes a rationality check between the preceding and following frames. The difference between the current frame's flight trajectory data and the previous frame's flight trajectory data in the received data queue is calculated, and it is determined whether the difference meets the second preset condition. If yes, the flight trajectory data is retained, and the process proceeds to the next step. Otherwise, the flight trajectory data is discarded.
[0028] The second pre-defined condition is as follows: in, Indicates the first k Frame sampling time, Indicates the first k -1 frame sampling time Indicates the first k The spatial location vector of a frame. Indicates the first k- A spatial position vector for one frame.
[0029] It should be noted that the difference between two adjacent frames (i.e., the difference between the current position and the previous position, and the corresponding time interval) reflects the instantaneous displacement and motion trend of the target. Using this difference information, the target's velocity can be estimated, the continuity of motion can be determined, and potential abnormal changes can be detected.
[0030] In the embodiment of the present application, by reasonably checking the time difference and position difference of the adjacent two frames of trajectory data, dynamic abnormal conditions such as time disorder, pseudo stationary point or abnormal jump can be effectively identified, so as to avoid speed calculation distortion and illegal data insertion, and help to protect the authenticity and stability of the trajectory data in terms of time continuity and motion consistency.
[0031] S203: After the flight trajectory data retained after the rationality check of the previous and next two frames of data, a multi-frame robust fitting check is performed to obtain target trajectory data.
[0032] In a possible implementation, S203 specifically includes: S2031: initialize the fitting queue to be empty, and define the maximum length of the fitting queue N .
[0033] S2032: read the current time trajectory data in the flight trajectory data retained after the rationality check of the previous and next two frames of data.
[0034] S2033: determine whether the current fitting queue is empty. If yes, directly press the current flight trajectory data into the queue and jump to S2039. Otherwise, proceed to the next step.
[0035] S2034: calculate the difference value between the adjacent frames of flight trajectory data in the window, and determine whether the difference value meets the second preset condition. If yes, proceed to the next step. Otherwise, determine that the current frame of flight trajectory data is an abnormal point and jump to S2038.
[0036] S2035: determine whether the length of the current fitting queue is less than N . If yes, add the current frame of flight trajectory data to the current fitting queue and jump to S2039. Otherwise, proceed to the next step.
[0037] S2036: perform local fitting of the flight trajectory data in the current sliding window by using the robust least square method, and calculate the fitting error.
[0038] Specifically, in the case that there are noises, outliers or even continuous abnormalities (such as spot-type outliers) in the trajectory data, the traditional least square method is highly sensitive to abnormal values, and is easily deviated by a small number of deviated points, resulting in distortion of the model. Therefore, the present application introduces the M-estimation idea and the robust least square method, introduces a weighting mechanism to the observation error, and adaptively suppresses the interference of abnormal points on the model. The observation equation is constructed in the sliding window, and the weight is updated through the iteration process, so that the observation points with large residuals are given small weights, thereby reducing their influence, and realizing the robust fitting of the real trajectory trend. Compared with the fixed weight fitting method, this method can dynamically adapt to the distribution characteristics of the observation error, and is more in line with the data uncertainty requirements in complex flight environments.
[0039] In a possible implementation, S2036 specifically includes: S20361: Collecting track measurement points collected within the current sliding window, and constructing an observation equation: m wherein, z represents an observation vector, H represents a coefficient matrix, represents a to-be-estimated parameter, v represents an observation error vector.
[0040] S20362: Assigning initial weights to each observation value in the observation equation to form a diagonal weight matrix: wherein, represents an initial observation weight matrix corresponding to the observation equation, represents a function operation for constructing a diagonal matrix, represents an initial weight value of an i-th observation value, i i = 1, 2… i , m represents a total number of track observation data within the current sliding window. m S20363: Defining a robust objective function based on the diagonal weight matrix by using the M-estimation idea:
[0041] wherein, represents a minimum value, min represents a robust loss function, represents a residual corresponding to an i-th observation value. i S20364: Taking a partial derivative of the robust objective function to obtain a robust condition equation:
[0042] wherein, represents an influence function in the robust estimation, represents a model coefficient of an i-th observation value and a j-th parameter of a coefficient matrix, represents a partial derivative, represents a to-be-estimated track model parameter, i represents a dimension of the to-be-estimated parameter. j j n
[0043] S20365: Calculate the equivalent weight factor, and determine the equivalent weight matrix according to the equivalent weight factor: wherein, represents the single-point weight value after the robust correction of the i th observation value, i represents the equivalent weight factor of the i th observation value. i
[0044] S20366: Substitute the equivalent weight matrix into the robust condition equation, construct the robust least squares equation and solve it to obtain the trajectory parameter estimate: wherein, H represents the coefficient matrix, T represents the transpose of the matrix, represents the equivalent weight matrix, represents the inverse of the matrix, represents the original observation vector, represents the weighted observation vector, represents the matrix.
[0045] S20367: Calculate the residual according to the trajectory parameter estimate, and update the full matrix according to the residual: wherein, represents the equivalent weight matrix updated after the i th iteration, k represents the equivalent weight factor updated after the i th iteration, represents the first-order derivative of the robust loss function with respect to the residual at the i th iteration, k represents the residual of the i th observation point at the i th iteration, represents the i th true observation value, k represents the trajectory parameter estimate vector obtained in the i th iteration. i k i k
[0046] S20368: Substitute the updated full matrix into the robust least squares equation again to obtain the estimate.
[0047] In this embodiment of the invention, by introducing a robust weighting mechanism, the tolerance of the fitting process to outliers and the accuracy of overall trajectory modeling are effectively improved. In scenarios with speckled outliers or non-Gaussian distributed errors, the influence of outliers can be gradually eliminated through multiple rounds of weight adjustment and iterative convergence, achieving robust estimation of trajectory parameters.
[0048] S20369: Calculate the fitting error based on the estimated and actual values.
[0049] Fitting error is an important indicator that measures the degree of deviation between the trajectory fitting model and the actual observed data. It is usually expressed by residuals (i.e., the difference between the model's predicted value and the actual observed value). In trajectory data processing, fitting error is not only used to evaluate the fitting accuracy of the current model, but also as a basis for determining whether the data is abnormal.
[0050] S2037: Determine whether the fitting error meets the fitting accuracy condition. If yes, mark the current frame flight trajectory data as a normal point and update the current residual to the error baseline. Otherwise, mark the current frame flight trajectory data as an outlier, replace the original value with the fitted value, and increment the outlier counter by 1.
[0051] The specific fitting accuracy conditions are as follows: in, Indicates the first k Frame fitting residuals of frames Indicates the error threshold. Indicates the first k -1 frame fitting residual.
[0052] S2038: Update and maintain the current fitting queue based on the fitting results and outlier counts.
[0053] Specifically, if the current fitting queue has not reached its maximum length N Then the flight trajectory data Push the data to the end of the fitting queue. If the current fitting queue is full, delete the data at the head of the queue and then add the flight trajectory data. Add it to the end of the queue to maintain a constant sliding window length. If the count of consecutive abnormal data exceeds the threshold (i.e., the anomaly counter)... If the error occurs, then perform the following operations: Clear the fitting queue. Reset the exception counter. The fitting process is reinitialized to prevent trajectory drift caused by accumulated anomalies.
[0054] S2039: After completing the selection and fitting of the trajectory data in the current frame, output the target trajectory data.
[0055] In the embodiments of the present application, by introducing a three-level hierarchical detection mechanism, the flight trajectory data is subjected to multi-level outlier identification and elimination, which can significantly improve the accuracy and robustness of anomaly detection. The first layer of single-point measurement element rationality detection can quickly eliminate obvious measurement anomalies that violate physical constraints, the second layer of before-and-after frame difference test enhances the judgment ability of time continuity and speed rationality, and the third layer of multi-frame robust fitting detection combines the M-estimation idea and the weighted least squares method to effectively identify and correct stringed spot-type outliers, avoiding their interference with the overall trajectory trend. The multi-layer structure is shallow to deep and gradually refined, which can simultaneously consider the detection and processing of isolated and related anomalies, and improve the data quality and system stability of subsequent trajectory smoothing, prediction and storage.
[0056] S3: Calculate the sampling interval coefficient of variation of the target trajectory data, and determine whether the sampling interval coefficient of variation is less than a preset sampling interval coefficient of variation. If yes, determine that the target trajectory data is equidistant sampling data, and proceed to S4. Otherwise, determine that the target trajectory data is non-equidistant sampling data, and proceed to S5.
[0057] The sampling interval coefficient of variation is an important index for measuring the stability of the sampling time interval of trajectory data, and is defined as the ratio of the standard deviation of the sampling time interval sequence to the mean value thereof. It reflects the uniformity and regularity of the trajectory data on the time axis, and the smaller the value, the closer the sampling to equidistant, and the more stable the data structure.
[0058] In one possible implementation, the calculation method of the sampling interval coefficient of variation specifically includes: Based on the sampling time sequence of the target trajectory data, a sampling interval sequence is determined: Wherein, denotes the time difference between adjacent two samplings, denotes the timestamp of the i-th sampling point in the target trajectory data, b denotes the timestamp of the i-1-th sampling point in the target trajectory data, denotes the total number of sampling intervals. b d The mean value of the sampling interval sequence is calculated:
[0059] Wherein, denotes the mean value of the sampling interval sequence. Based on the mean value, the standard deviation of the sampling interval sequence is calculated:
[0060] Wherein, a standard deviation of the sampling interval sequence, a difference between each sampling interval and the mean value.
[0061] Based on the mean value and the standard deviation of the sampling interval sequence, a sampling interval coefficient of variation is calculated: wherein, the sampling interval coefficient of variation.
[0062] In the embodiments of the present application, by calculating the sampling interval coefficient of variation, the time sampling structure of the trajectory data is effectively judged, so as to realize the automatic classification of equal interval and non-equal interval data. As an important statistical index for measuring sampling stability, the coefficient can accurately reflect the consistency of the timestamp interval, avoiding the misuse of mismatched smoothing algorithms due to sampling frequency fluctuations. At the same time, with the help of the discrimination mechanism, the system can adaptively select the corresponding smoothing processing strategy according to the data characteristics, improve the trajectory fitting accuracy, and enhance the intelligence and robustness of the overall data processing flow.
[0063] S4: The target trajectory data sampled at equal intervals is smoothed by an equal interval center smoothing processing method, and S6 is entered.
[0064] The equal interval center smoothing processing is a local weighted fitting method based on a sliding window, which is suitable for trajectory data with basically consistent time sampling intervals. The method selects a plurality of trajectory observation points symmetrically with the current time as the center, constructs a quadratic polynomial model with time offset as the variable, estimates the model parameters by the least square method, and thus obtains the smoothed trajectory information such as position, first derivative (velocity) and second derivative (acceleration). Since the samples are uniformly distributed on the time axis, the weight distribution is stable during the fitting process, which can effectively eliminate local jitter, reduce noise influence, improve the continuity and smoothness of the trajectory curve, and at the same time preserve the original motion trend.
[0065] In one possible implementation, S4 specifically includes: S401: Selecting a plurality of trajectory data points centered on the current time from the target trajectory data sampled at equal intervals: wherein, the timestamp of the i-th trajectory point, n the measurement value of the i-th trajectory point, the half-width of the sliding window. n N S402: Constructing a first quadratic polynomial trajectory model:
[0066] wherein, actual trajectory observation value of the i-th sampling point, k actual trajectory observation value of the i-th sampling point, , , model parameters to be fitted, time offset of the i-th point relative to the center point, k time offset of the i-th point relative to the center point, noise term.
[0067] S403: Establish the residual sum of squares of the first quadratic polynomial trajectory model as the least squares target, and construct the least squares target function: wherein, least squares target function, half width of the sliding window.
[0068] It should be noted that the residual sum of squares is established as the least squares target because the residual sum of squares can measure the overall deviation between the model predicted value and the actual observation value, and the model parameters that make this deviation “minimized” are found through mathematical optimization means.
[0069] S404: Partially derive the multiple fitting parameters in the least squares target function and set them to zero to obtain the equation group: wherein, partial derivative, s least squares target function.
[0070] S405: Solve the equation group to obtain the fitting parameter coefficient expression: wherein, square sum of the relative positions of the sampling points in the sliding window, fourth power sum of the sampling points in the sliding window, weight function for constructing the fitting coefficient weight function for constructing the fitting coefficient weight function for constructing the fitting coefficient weight function for constructing the fitting coefficient weight function for constructing the fitting coefficient weight function for constructing the fitting coefficient
[0071] S406: Substitute the fitting parameter coefficient expression into the first quadratic polynomial trajectory model to obtain the smoothed position value, the first derivative estimate, and the second derivative estimate, and complete the smoothing process: in, Indicates the current center point k Smoothed position estimate at the location, I This represents the relative index position variable within the sliding window. These represent the basic smoothing weighting coefficients used for location estimation. This represents the smoothing weight coefficient corresponding to the first-order time term. This represents the smoothing weight coefficient corresponding to the second-order time term. Indicates the current sliding window's... I Measurement values of each trajectory point This represents the weighting coefficient corresponding to the first derivative. The second derivative corresponds to the 1st derivative. l The weighting coefficients for each sampling point Indicates the first l The original observations of each sampling point Indicates the current center point k The estimated acceleration at that location, Indicates the first l The second derivative weighting coefficients for each sampling point.
[0072] Optionally, when N When the value is 10, it is a 21-point center smoothing. The formula for 21-point center smoothing is: in, Indicates the smoothed position value. This represents the first derivative (velocity) estimate. This represents the second derivative (acceleration) estimate.
[0073] In this embodiment of the invention, trajectory data is locally fitted using a quadratic polynomial through equal-interval center smoothing. A least-squares optimization model is constructed using symmetrical observation points within a sliding window to accurately estimate key parameters such as current position, velocity, and acceleration. Under the premise of stable sampling intervals, this method can fully utilize time symmetry to achieve balanced weight distribution and trend preservation, significantly reducing the impact of local noise and measurement errors, and improving the continuity and smoothness of the trajectory curve.
[0074] S5: By using non-equal interval least squares fitting, the target trajectory data sampled at non-equal intervals is smoothed, and then proceeds to S6.
[0075] Among them, non-uniform interval least squares fitting is a smoothing method designed for trajectory data with uneven sampling time intervals. This method uses the actual sampling time as the independent variable to construct a quadratic polynomial trajectory model, and uses the least squares method to fit and estimate parameters such as position, velocity, and acceleration, thus avoiding the fitting bias problem of the uniform interval model under time misalignment or dense sampling conditions.
[0076] In one possible implementation, S5 specifically includes S501: Select multiple trajectory data points centered at the current time from the non-equally spaced target trajectory data: in, Indicates the first n The timestamp of each trajectory point Indicates the first n Measurement values of each trajectory point N This represents the half width of the sliding window.
[0077] S502: Constructing the second quadratic polynomial trajectory model: in, Represents the measured values of the trajectory points. , , All represent the model parameters to be fitted. This represents the random Gaussian noise term.
[0078] S503: Transform the second quadratic polynomial trajectory model into a matrix expression: in, Represents the observation vector. Represents the design matrix. Represents the model parameter vector. W This represents the observation noise vector.
[0079] S504: Based on matrix expressions, the least squares method is used to estimate the parameters of the smooth curve. in, Represents the parameters of the smooth curve. The coefficients of the constant term in the quadratic polynomial fitting are represented. Denotes the coefficient of the linear term. Denotes the coefficient of the quadratic term. T This represents the transpose of a matrix.
[0080] S505: Substitute the smooth curve parameters and the current center point time into the fitting model to calculate the smooth value of the current trajectory point: wherein, represents the smooth estimation value of the current center point.
[0081] S506: Output the obtained smooth value as the trajectory estimation result at the current time, and complete the smoothing process.
[0082] In the embodiment of the application, by using the non-equidistant least squares fitting method, adaptive smoothing processing is performed on trajectory data with uneven time distribution, which can effectively avoid the fitting deviation of the traditional equidistant method when facing dense sampling, time jitter or interval mutation. At the same time, by constructing a polynomial model with real time stamp as a variable and solving the fitting parameters in matrix form, the smooth estimation of position data is realized without sacrificing time accuracy, which ensures the continuity and dynamic consistency of the trajectory. It significantly improves the adaptability and robustness of the trajectory processing system in complex sampling environment, and provides high-quality trajectory basis for subsequent flight state evaluation and trend prediction.
[0083] S6: Save the target trajectory data after smoothing to the trajectory processing result cache area through a multi-layer cache mechanism.
[0084] The multi-layer cache mechanism is a hierarchical storage strategy for improving the efficiency and stability of trajectory data processing, which is usually composed of a single-point cache, a batch aggregation cache and a persistent storage area. The mechanism manages data of different granularities and lifecycles through hierarchical management, realizing efficient flow between real-time reception, short-term buffering and long-term saving.
[0085] In one possible implementation, S6 specifically includes: S601: Initialize the multi-layer cache structure, which includes a single-point cache area, a batch aggregation cache area and a persistent storage area.
[0086] The single-point cache area is used to temporarily store single trajectory data just received, has the characteristics of fast read-write speed and low access delay, and mainly undertakes the functions of real-time access and fast response of trajectory data in the system. The cache area supports atomic update operation and can instantly overwrite old data when new data arrives, which is suitable for fast detection.
[0087] The batch aggregation cache area is used to collect a certain number of track data entries and perform batch packaging processing, and its core function is to relieve the system burden caused by frequent writing and improve the data writing efficiency. When the data accumulation reaches the preset batch threshold, it can be uniformly persisted or further analyzed and calculated, has the advantages of controllable cache capacity, unified operation, adaptive batch processing logic, etc., effectively improves the overall throughput capacity and stability of the processing rhythm of the system.
[0088] The persistent storage area is responsible for reliably writing the processed and cache-managed data to long-term storage media such as databases, file systems, or cloud storage, etc., for track history archiving, post-analysis or fault backtracking. This area has the characteristics of data not easy to lose, strong queryability, stable structure, etc., and through asynchronous writing, index recording, fault rollback mechanism, etc., to ensure the integrity, security and traceability of key track information.
[0089] S602: receiving the target track data after smoothing processing.
[0090] S603: real-time monitoring of the load state of the single-point cache area and the batch aggregation cache area, when the load of the single-point cache area exceeds the first preset threshold, emptying the content of the single-point cache area. When the load of the batch aggregation cache area exceeds the second preset threshold, the packaging and persistent storage process of the current data batch is performed in advance, and the standby buffer is enabled.
[0091] Optionally, the load state of the cache area is: Among them, represents the current load ratio of the first-level cache area, represents the number of data items stored in the current first-level cache area, represents the maximum capacity upper limit of the first-level cache area.
[0092] Optionally, the load state of the aggregation cache area is: Among them, represents the current load ratio of the second-level batch aggregation cache area, and represents the number of data entries temporarily stored in the current second-level cache area, represents the maximum capacity upper limit of the second-level cache area.
[0093] In the embodiments of the present application, by monitoring the load state of the cache area and the batch aggregation cache area in real time, the dynamic perception and intelligent regulation of the trajectory data cache pressure are realized. When the cache load approaches the set threshold, the system can automatically empty, package or dump the cache content, preventing data accumulation from causing overflow or delay, while ensuring the continuity of data flow and the efficiency of processing. At the same time, monitoring the load state not only helps to improve the stability and response speed of the system in the high-frequency data input scene, but also enhances the flexibility and robustness of the trajectory data processing flow, avoiding data loss or service blocking problems caused by exhausted cache resources.
[0094] S604: Store the received smoothed target trajectory data in the result cache area including a multi-layer cache structure.
[0095] In a possible implementation, S604 specifically includes: S6041: Based on atomic operation, update the received each piece of trajectory data to the single-point cache area, and clear the content stored in the last time period.
[0096] Atomic operation refers to the smallest operation unit that cannot be interrupted or divided during execution, that is, the operation is either completely executed or not executed at all, and it will not be interrupted or interrupted by any other task in the middle. In a multi-threaded or concurrent execution environment, atomic operation is often used to ensure the consistency and integrity of shared data, and to prevent race conditions or data overwrite conflicts.
[0097] S6042: Append each piece of received trajectory data to the batch aggregation cache area, and judge whether the number of data in the cache meets the preset batch size. If yes, start the persistent write process, and package the cache content as a complete batch. Otherwise, perform a single cleaning operation on the temporary cache file that exceeds the set preservation time limit.
[0098] It should be noted that the persistent write process is a key link for stably and efficiently storing the trajectory data batch processed and managed by smoothing into a long-term storage medium. This process is usually triggered in an asynchronous manner to avoid blocking the main processing thread, and the packaging operation is started when the data reaches the preset batch threshold, and is uniformly submitted to the persistent carrier such as database, file system or cloud storage.
[0099] S6043: The packaged trajectory data batch is submitted to the persistent storage area in an asynchronous write mode. If the cache is successful, update the trajectory data index and record the log. Otherwise, trigger the fault handling process, write back and switch to the standby storage path for re-caching.
[0100] The asynchronous writing mode is a mechanism of decoupling data writing operation from the main processing flow, and can submit the trajectory data batches to the persistent storage system in the background without blocking the front-end data receiving and processing.
[0101] Optionally, after S604, the method further includes: S605: cleaning the temporary cache files that exceed the set storage time limit, and performing compression operation on the historical trajectory data archive files.
[0102] In the embodiment of the application, by orderly storing the smoothed trajectory data according to the multi-layer cache structure, combining the atomic update, batch encapsulation and asynchronous writing mechanism, efficient and safe management of data from access to persistence is realized. At the same time, combined with the regular cleaning and archiving compression strategy, the cache expansion and storage waste can be effectively prevented, and the integrity, stability and sustainability of the trajectory data in high-frequency circulation are ensured.
[0103] Reference is made to the accompanying drawings Figure 2 The accompanying drawings show a structure schematic diagram of a target data real-time selection and smoothing system based on error estimation provided by the application.
[0104] The application further provides a target data real-time selection and smoothing system 20 based on error estimation, which is applied to the target data real-time selection and smoothing method based on error estimation described above, and includes: A processor 201.
[0105] A memory 202, the memory 202 stores computer readable instructions, and when the computer readable instructions are executed by the processor 201, the target data real-time selection and smoothing method based on error estimation as in the method embodiment is realized.
[0106] The target data real-time selection and smoothing system 20 based on error estimation provided by the application can execute the target data real-time selection and smoothing method based on error estimation described above, and realize the same or similar technical effects. To avoid repetition, the application will not be described again.
[0107] The technical scheme provided by the embodiment of the application has at least the following beneficial effects: In the embodiment of the present application, by establishing a temperature and humidity balance model based on physical principles, the system can simulate temperature and humidity changes and predict future temperature and humidity values according to historical environmental data. In combination with a self-tuning PID controller, the system can adjust the temperature and humidity of the maintenance room in real time, ensuring that the system can dynamically adapt to environmental changes, enhancing adaptability and improving control effect. At the same time, the self-tuning PID controller quickly adjusts the temperature and humidity according to the prediction results, avoiding temperature and humidity control lag and energy waste.
[0108] It should be understood that the processor in the embodiment of the present application can be a central processing unit (CPU), and the processor can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.
[0109] It should also be understood that the memory in the embodiments of the present application can be volatile or nonvolatile memory, or can include both volatile and nonvolatile memory. The nonvolatile memory can be read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), electrically EPROM (EEPROM), or flash memory. The volatile memory can be random access memory (RAM) used as external cache. By way of example, and not limitation, many forms of random access memory (RAM) are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).
[0110] The above-described embodiments can be implemented in whole or in part by software, hardware (such as a circuit), firmware, or any combination thereof. When implemented in software, the above-described embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center through a wired (such as infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.
[0111] It should be understood that the term "and / or" herein merely describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents that the associated objects before and after it are in an "or" relationship, but it can also represent an "and / or" relationship, which can be understood according to the context before and after it.
[0112] In the present application, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or the like means any combination of the items, including any combination of single or multiple items. For example, at least one of a, b, or c can represent a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.
[0113] It should be understood that in various embodiments of the present application, the size of the sequence number of the above-described processes does not mean the order of execution, and the execution order of the processes should be determined according to their functions and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0114] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.
[0115] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the devices, apparatuses and units described above can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.
[0116] In several embodiments provided by the present application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.
[0117] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.
[0118] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit.
[0119] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or parts of the present application that essentially contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.
[0120] The computer readable storage medium provided by the embodiment of the present application stores a computer program, and the program is executed by a processor to realize the error estimation based target data real-time selection and smoothing method as described in the method embodiment.
[0121] The computer readable storage medium provided by the present application can realize the steps and effects of the error estimation based target data real-time selection and smoothing method of the above method embodiment. To avoid repetition, the present application will not be described again.
[0122] The technical solutions provided by the embodiment of the present application have at least the following beneficial effects: In the embodiment of the present application, by establishing a temperature and humidity balance model based on physical principles, the system can simulate temperature and humidity changes and predict future temperature and humidity values according to historical environmental data. Combined with a self-tuning PID controller, the system can adjust the temperature and humidity of the maintenance room in real time, ensuring that the system can dynamically adapt to environmental changes, enhancing adaptability and improving control effect. At the same time, the self-tuning PID controller quickly adjusts the temperature and humidity according to the prediction results, avoiding temperature and humidity control lag and energy waste.
[0123] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.
[0124] The following points need to be explained: (1) The drawings of the embodiment of the present application only involve the structures related to the embodiment of the present application, and other structures can refer to the usual design.
[0125] (2) In the drawings used to describe the embodiments of the present application, the thickness of layers or regions are exaggerated or reduced, for clarity, and are not necessarily drawn to scale. It will be understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, it can be "directly" on or under the other element or an intervening element can also be present.
[0126] (3) The embodiments of the present application and the features in the embodiments can be combined with each other without conflict if possible, to obtain new embodiments.
[0127] The above merely describes specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and should be subject to the protection scope of the claims.
Claims
1. A method for real-time selection and smoothing of target data based on error estimation, characterized in that, The method comprises the following steps: S1: acquiring flight trajectory data comprising time, position, speed and height; S2: performing trajectory outlier detection processing on the flight trajectory data through a three-level hierarchical detection method to obtain target trajectory data; S3: calculating a sampling interval variation coefficient of the target trajectory data, and determining whether the sampling interval variation coefficient is less than a preset sampling interval variation coefficient; if yes, determining that the target trajectory data is equidistant sampling data, and proceeding to S4; otherwise, determining that the target trajectory data is non-equidistant sampling data, and proceeding to S5; S4: performing smoothing processing on the equidistant sampling target trajectory data through an equidistant center smoothing processing method, and proceeding to S6; S5: performing smoothing processing on the non-equidistant sampling target trajectory data through a non-equidistant least square fitting method, and proceeding to S6; S6: saving the smoothed target trajectory data to a trajectory processing result cache area through a multi-layer cache mechanism.
2. The method of claim 1, wherein the error estimate based target data real-time selection and smoothing method is characterized by, The S2 specifically comprises the following steps: S201: performing single-point measurement element rationality detection on the flight trajectory data, determining whether the time, position, speed and height measurement elements of each flight trajectory data meet a first preset condition; if yes, retaining the flight trajectory data, and proceeding to the next step; otherwise, eliminating the flight trajectory data; S202: performing front-and-back two-frame data rationality detection on the flight trajectory data retained after the single-point measurement element rationality detection, calculating a difference value of the current frame flight trajectory data and the last frame flight trajectory data in the received data queue, and determining whether the difference value meets a second preset condition; if yes, retaining the flight trajectory data, and proceeding to the next step; otherwise, eliminating the flight trajectory data; S203: performing multi-frame robust fitting detection on the flight trajectory data retained after the front-and-back two-frame data rationality detection to obtain the target trajectory data.
3. The method of claim 2, wherein the error estimate is based on a difference between the target data and a predicted value of the target data. The S203 specifically comprises the following steps: S2031 : initialize the fitting queue to be empty, and define the maximum length of the fitting queue N ; S2032: reading the current frame trajectory data from the flight trajectory data retained after the front-and-back two-frame data rationality detection; S2033: determining whether the current fitting queue is empty; if yes, directly pushing the current flight trajectory data into the queue, and jumping to S2039; otherwise, proceeding to the next step; S2034: calculating a difference value between adjacent frame flight trajectory data in a window, and determining whether the difference value meets the second preset condition; if yes, proceeding to the next step; otherwise, determining that the current frame flight trajectory data is an abnormal point, and jumping to S2038; S2035: judging whether the current fitting queue length is less than N ; if yes, adding the current frame flight trajectory data to the current fitting queue, and jumping to S2039; otherwise, proceeding to the next step; S2036: performing robust least square local fitting on the flight trajectory data in the current sliding window, and calculating a fitting error; S2037: determining whether the fitting error meets a fitting accuracy condition; if yes, marking the current frame flight trajectory data as a normal point, updating the current residual error as an error reference; otherwise, marking the current frame flight trajectory data as an abnormal point, replacing the original value with a fitting value, and increasing an abnormal counter by 1; S2038: updating and maintaining the current fitting queue according to the fitting result and the abnormal counting condition; S2039: outputting the target trajectory data after completing the detection and fitting processing of the current frame trajectory data.
4. The method of claim 3, wherein the error estimate is based on a difference between the target data and a predicted value of the target data. The S2036 specifically comprises: S20361: Construct observation equation with the track measurement points collected in the current sliding window m ; S20362: assigning initial weights to each observation value in the observation equation to form a diagonal weight matrix; S20363: defining a robust objective function based on the diagonal weight matrix using the M-estimation idea; S20364: obtaining a partial derivative of the robust objective function to obtain a robust condition equation; S20365: calculating an equivalent weight factor and determining an equivalent weight matrix according to the equivalent weight factor; S20366: substituting the equivalent weight matrix into the robust condition equation to construct a robust least square equation and solving to obtain a trajectory parameter estimate; S20367: calculating a residual according to the trajectory parameter estimate, and updating a full matrix according to the residual; S20368: substituting the updated full matrix into the robust least square equation to obtain an estimate; S20369: calculating the fitting error according to the estimate and the actual value.
5. The method of claim 1, wherein the error estimate based target data real-time selection and smoothing method further comprises: The calculation method of the sampling interval coefficient of variation specifically comprises: determining a sampling interval sequence based on a sampling time sequence of the target trajectory data; calculating the average value of the sampling interval sequence; calculating the standard deviation of the sampling interval sequence based on the average value; calculating the sampling interval coefficient of variation based on the average value and the standard deviation of the sampling interval sequence.
6. The method of claim 1, wherein the error estimate based target data real-time selection and smoothing method further comprises: The S4 specifically comprises: S401: selecting a plurality of trajectory data points centered on the current time from the target trajectory data sampled at equal intervals; S402: constructing a first quadratic polynomial trajectory model; S403: establishing a residual sum of squares of the first quadratic polynomial trajectory model as a least square objective to construct a least square objective function; S404: obtaining an equation set by respectively taking partial derivatives of the plurality of fitting parameters in the least square objective function and setting them to zero; S405: solving the equation set to obtain a fitting parameter coefficient expression; S406: substituting the fitting parameter coefficient expression into the first quadratic polynomial trajectory model to obtain smoothed position values, first derivative estimates, and second derivative estimates, and completing the smoothing process.
7. The method of claim 1, wherein the error estimate based target data real-time selection and smoothing method further comprises: The S5 specifically comprises: S501: selecting a plurality of trajectory data points centered on the current time from the target trajectory data sampled at unequal intervals; S502: constructing a second quadratic polynomial trajectory model; S503: converting the second quadratic polynomial trajectory model into a matrix expression; S504: estimating a smoothed curve parameter using the least square method based on the matrix expression; S505: substituting the smoothed curve parameter and the current center point time into a fitting model to calculate a smoothed value of the current trajectory point; S506: outputting the obtained smoothed value as a trajectory estimation result at the current time to complete the smoothing process.
8. The method of claim 1, wherein the error estimate based target data real-time selection and smoothing method further comprises: The S6 specifically comprises: S601: initializing a multi-layer cache structure, the multi-layer cache structure comprising a single-point cache area, a batch aggregation cache area, and a persistent storage area; S602: receiving the target trajectory data after smoothing processing; S603: Real-time monitoring of the load state of the single-point cache area and the batch aggregation cache area, when the single-point cache area load exceeds the first preset threshold, the content of the single-point cache area is emptied; when the load of the batch aggregation cache area exceeds the second preset threshold, the encapsulation and persistent storage process of the current data batch is performed in advance, and the standby buffer is enabled; S604: Store the received smoothed target trajectory data in the result cache area including the multi-layer cache structure.
9. The real-time target data selection and smoothing method based on error estimation according to claim 8, characterized in that, The S604 specifically includes: S6041: Based on atomic operation, update the received each piece of trajectory data to the single-point cache area, and clear the content stored in the last time period; S6042: Append each piece of received trajectory data to the batch aggregation cache area, judge whether the number of data in the cache meets the preset batch size; if yes, start the persistent write process, and encapsulate the cache content as a complete batch; otherwise, perform a single cleaning operation on the temporary cache file exceeding the set saving time limit; S6043: Adopting asynchronous writing mode, submit the encapsulated trajectory data batch to the persistent storage area; if the cache is successful, update the trajectory data index, and record the log; otherwise, trigger the fault handling process, write back and switch to the standby storage path for re-caching.
10. A real-time selection and smoothing system of target data based on error estimation, characterized in that, including: a processor; a memory, the memory has computer readable instructions stored thereon, the computer readable instructions are executed by the processor to implement the real-time target data selection and smoothing method based on error estimation according to any one of claims 1 to 9.