An efficient system and method for processing objective data from a helicopter flight simulator
By developing an efficient system and method for processing objective data from helicopter flight simulators, the problems of low data processing efficiency and low accuracy in helicopter simulators have been solved, achieving efficient and accurate data processing and meeting the needs of simulator development.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA HELICOPTER RES & DEV INST
- Filing Date
- 2026-05-08
- Publication Date
- 2026-06-30
AI Technical Summary
The processing of flight test data for helicopter flight simulators suffers from low efficiency, low accuracy, and difficulty in handling large-scale data. In particular, due to the high coupling of its aerodynamic characteristics, existing technologies are unable to meet the high-efficiency processing requirements of objective data in simulator development.
An efficient processing system and method for objective data from helicopter flight simulators is proposed, including modules for data preprocessing, segmentation, batch processing, outlier handling, curve fitting, and filtering. Through polynomial curve fitting, filtering, and data resampling techniques, efficient processing of objective data from helicopter flight simulators is achieved.
It significantly improves the efficiency and accuracy of data processing, reduces outlier residues, lowers fitting errors, ensures data consistency and high sampling rate requirements, and shortens processing time by 80%.
Smart Images
Figure CN122309930A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of helicopter flight simulator technology, specifically relating to an efficient system and method for processing objective data from helicopter flight simulators. Background Technology
[0002] In the aerospace field, flight test data is a crucial basis for evaluating aircraft performance. Traditional flight test data processing methods mainly include manual processing and simple computer-aided processing. Helicopters, due to the complex motion of their rotors, wake characteristics, aerodynamic interference, and blade deformation, exhibit highly coupled aerodynamic characteristics, making them more complex than fixed-wing aircraft. Traditional methods suffer from low efficiency, low accuracy, and difficulty in handling large-scale data when processing complex flight test data. The Shenyang Aircraft Design Institute of the Aviation Industry Corporation of China (AVIC) has provided a method and apparatus for centralized processing of aircraft flight test data (patent publication number CN117591515A). This invention includes the following steps: a. Obtaining aircraft flight test description data and aircraft flight test data; b. Structuring the flight test description data and associating it with the aircraft flight test data through filenames; c. Writing the aircraft flight test data associated with the flight test description into a Hadoop distributed file system; d. Converting the aircraft flight test data written to the Hadoop distributed file system into CSV format and storing it in a column database; e. Obtaining query templates with one or more data attributes filled in by various professionals, retrieving them from the column database to form secondary database flight parameter data for download.
[0003] Although the aforementioned invention patents cover aircraft flight test data processing and emphasize efficient storage methods for large-scale data, they do not delve into the processing of actual flight test data parameters collected by the aircraft. In particular, the aerodynamic characteristics of helicopters are highly coupled, and the processing of objective data from helicopter simulators is significantly more complex than the flight test data of fixed-wing aircraft covered by the aforementioned invention patents.
[0004] To meet the requirements of helicopter simulator development, a series of meticulous processing operations must be performed on helicopter flight test data to generate the objective data needed for simulator development. This includes, but is not limited to, outlier removal, curve fitting, and automated batch processing. The efficiency and effectiveness of these processing steps are crucial to simulator development. Summary of the Invention
[0005] Purpose of the Invention: In the development of flight simulators, objective data serves as the basis for the calibration and verification of the simulator's simulation model. The quality of objective data processing directly affects the realism of the simulator simulation and the training effect on flight personnel. This invention aims to provide an efficient system and method for processing objective data from helicopter flight simulators, mainly including endpoint outlier processing, curve fitting processing, batch data processing, and data resampling.
[0006] According to the first aspect of the present invention, a high-efficiency processing system for objective data of a helicopter flight simulator is proposed. The objective data of the simulator refers to the data used for performance verification and objective testing of the helicopter simulator. The objective data consists of test parameter data such as helicopter configuration parameters, control parameters, flight status parameters, atmospheric data parameters, engine parameters, and fuel system parameters.
[0007] The system includes a data preprocessing module, a data segmentation module, a batch processing module, an outlier processing module, a curve fitting processing module, and a filtering processing module.
[0008] The data preprocessing module is used to preprocess the acquired complete flight data, including data fusion and time series correction.
[0009] The data segmentation module determines the start and end times of actions based on helicopter configuration, changes in control inputs, and flight status, and performs action segmentation on the original simulator objective data to obtain data for each action segment.
[0010] The batch processing module provides a batch processing entry point for multiple sets of action segment data, and imports multiple sets of action segment data and multiple data parameters into one or more other processing modules for processing at the same time.
[0011] The outlier processing module is used to identify and remove outliers from the action segment data after batch data processing.
[0012] The curve fitting processing module is used to correct the data after outlier removal;
[0013] The filtering module is used to recover the true information as much as possible from the motion segment data containing interference. That is, the signal has various frequency components, and the module filters out specific frequency band components (generally noise) and retains the desired components.
[0014] In one possible embodiment, the system further includes a smoothing module for generating a continuous and smooth data curve when the action segment data exhibits a stepped shape, using smoothing techniques.
[0015] In one possible embodiment, the system further includes a data resampling module, which is used to improve the low-sampling-rate objective data to high-sampling-rate data that meets the objective testing requirements of the helicopter simulator when the sampling rate of the original simulator objective data does not meet the objective testing requirements of the helicopter simulator.
[0016] According to a second aspect of the present invention, an efficient method for processing objective data from a helicopter flight simulator is proposed, employing the aforementioned efficient system for processing objective data from a helicopter flight simulator, comprising the following steps:
[0017] The entire flight data was acquired and preprocessed;
[0018] The preprocessed data of the entire sortie is segmented, and the start and end times of the actions are determined based on the helicopter configuration, changes in control inputs and flight status. The original simulator objective data is segmented to generate action segment data.
[0019] It provides a batch processing entry point, allowing multiple sets of action segmented data to be simultaneously imported into one or more other processing modules for processing.
[0020] Identify and remove outliers from the action segment data after batch data processing;
[0021] The discarded points were corrected by using a polynomial curve fitting method.
[0022] In one possible embodiment, the specific process of acquiring and preprocessing the entire flight data includes:
[0023] Data from different sources is merged to form a complete file of data for the entire flight;
[0024] Some parameters in the acquired flight data are extracted from the bus, while others require external sensor acquisition. Some parameters will have a certain delay, so time delay correction is needed for some parameters in the flight data during preprocessing.
[0025] In one possible embodiment, the preprocessed data of the entire sortie is segmented, the start and end times of the action are determined based on the helicopter configuration, changes in control inputs and flight status, and the original simulator objective data is segmented to generate action segment data.
[0026] It provides a batch processing entry point for multiple sets of motion segment data, allowing multiple sets of motion segment data and multiple data parameters to be imported into one or more other processing modules for processing at the same time.
[0027] In one possible embodiment, the specific process of identifying and removing outliers from batch-processed data includes:
[0028] The ordinary seven-point second-order algorithm is used to identify and remove isolated outliers and clusters of outliers in the middle of the data based on the innovation value E and the number of consecutive occurrences of acceptable outliers R.
[0029] For outliers at the front end of the data, a seven-point second-order algorithm calculation is performed in reverse starting from the 18th point of the data. The result of the reverse calculation is used to replace the result of the forward identification, thus completing the identification and removal of outliers.
[0030] In one possible embodiment, the locations of the identified and removed outliers are corrected using Lagrange interpolation.
[0031] Select the preset or manually set fitting length and polynomial power;
[0032] Based on the principle of least squares, the polynomial coefficients are solved to obtain the fitting function;
[0033] The fitting function is used to calculate the fitting value for each data point, and the fitting values are used to form a smoothed dataset.
[0034] In one possible embodiment, a filtering module is invoked to recover the true information from the action segment data containing noisy information, using a first-order filtering and low-pass filtering method.
[0035] When using a first-order filtering algorithm, the data sequence is forward filtered, and then the same data sequence is reverse filtered. The corresponding points of the two filtering results are added together and the average value is taken as the final filtered output.
[0036] Alternatively, when using the Butterworth low-pass filter algorithm, set the low-pass cutoff frequency and the high-impedance cutoff frequency to filter the data.
[0037] In one possible embodiment, when some parameters of the original simulator objective data exhibit a stepped shape, spline processing is used to generate a continuous and smooth data curve.
[0038] In one possible embodiment, when the sampling rate of the original simulator objective data does not meet the objective testing requirements of the helicopter simulator, the low sampling rate objective data is upsampled to a high sampling rate that meets the objective testing requirements of the helicopter simulator. The specific process of upsampling the low sampling rate objective data to a high sampling rate that meets the objective testing requirements of the helicopter simulator includes:
[0039] For switch parameters and jump variable parameters, front-point interpolation is used to increase data density;
[0040] For other types of parameters, linear interpolation is used to increase data density.
[0041] According to a third aspect of the present invention, a computer-readable storage medium is provided, wherein a computer program is stored therein; when the computer program is executed by a processor, it implements the above-described method for efficiently processing objective data of a helicopter flight simulator.
[0042] In summary, the beneficial effects of the present invention are as follows:
[0043] This invention achieves significant technical improvements in four aspects: batch processing of objective data from helicopter simulators, handling of endpoint outliers, filtering, smoothing, and data resampling. Specific effects are as follows:
[0044] 1) Endpoint outlier handling
[0045] a) Traditional outlier handling methods typically fail to identify outliers at endpoints and can only be corrected manually, which can easily lead to residual outliers and affect the effectiveness of objective data processing. This invention employs a reverse inspection method, which can accurately identify and correct outliers at endpoints.
[0046] b) Advantages:
[0047] Endpoint outliers can be directly processed using the methods of this system, eliminating the need for manual checks and reducing the time spent on data inspection and processing, thereby improving the effectiveness of objective data processing in the simulator.
[0048] 2) Curve fitting processing
[0049] a) Traditional curve fitting methods are prone to overfitting or underfitting when dealing with nonlinear objective data. This invention uses a polynomial parameter fitting method, which can select the optimal polynomial fitting equation based on the characteristics of the curve by using different orders, thereby reducing overfitting and underfitting.
[0050] b) Advantages:
[0051] Compared with traditional algorithms, the fitting error is reduced by an average of 40%, and the algorithm can adapt to different types of objective data, including noisy data and sparse data.
[0052] c) Experimental data:
[0053] In fitting objective data from helicopter simulators, the mean squared error (MSE) of traditional methods is 0.045, while the MSE of the method of this invention is reduced to 0.018, resulting in a significant improvement in fitting performance.
[0054] 3) Batch data processing
[0055] a) Traditional objective data processing typically employs a single-file, line-by-line processing method, which is inefficient and struggles to guarantee consistency. This invention enables batch processing of objective data and supports automated processing of multiple files.
[0056] b) Advantages:
[0057] Batch processing speed is 60% faster than traditional methods, and data consistency is ensured through unified processing rules.
[0058] c) Experimental data
[0059] In the processing of objective data from helicopter simulators, traditional methods require 48 hours to process 100G of data, while the method of this invention only requires 4 hours, significantly improving efficiency.
[0060] 4) Data resampling
[0061] a) For objective data with low sampling frequency, this system employs a data resampling technique. This technique uses interpolation and filtering to increase the data sampling rate, meeting the high sampling rate requirements of objective data from helicopter simulators. Simultaneously, the algorithm is optimized to reduce noise and distortion. Traditional resampling methods typically target single datasets and are insufficient for batch processing of multiple datasets. This system implements batch resampling functionality.
[0062] b) Advantages:
[0063] The data processed by this system can meet the high sampling rate requirements of objective data for helicopter simulators, and the batch resampling efficiency is 80% higher than that of traditional methods.
[0064] c) Experimental data:
[0065] In the resampling of objective data from helicopter simulators, it is necessary to interpolate 5Hz data to 100Hz, with a data volume as high as 10G. Using this system for resampling processing, the processing time is reduced by 80%. Attached Figure Description
[0066] Figure 1 This is a schematic diagram of a high-efficiency system for processing objective data from a helicopter flight simulator according to a preferred embodiment of the present invention;
[0067] Figure 2 A schematic diagram illustrating the seven-point second-order algorithm for identifying isolated outliers;
[0068] Figure 3 A schematic diagram illustrating the identification of patchy outliers using the 7-point second-order algorithm;
[0069] Figure 4 A schematic diagram illustrating the identification of outlier parameters in a patch using a 7-point second-order algorithm;
[0070] Figure 5 This is an illustration illustrating the failure to identify endpoint outliers.
[0071] Figure 6 A schematic diagram illustrating outlier correction examples;
[0072] Figure 7 A schematic diagram illustrating the polynomial curve fitting method for processing objective data from a helicopter simulator.
[0073] Figure 8 A diagram illustrating low sampling rate data. Detailed Implementation
[0074] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0075] It should be noted that if the embodiments of the present invention involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.
[0076] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.
[0077] Example 1
[0078] like Figure 1 As shown, a high-efficiency processing system for objective data of helicopter flight simulators is provided. The objective data of the simulator refers to the data used for performance verification and objective testing of helicopter simulators. The objective data consists of test parameter data such as helicopter configuration parameters, control parameters, flight status parameters, atmospheric data parameters, engine parameters, and fuel system parameters.
[0079] The system includes a data preprocessing module, a data segmentation module, a batch processing module, an outlier processing module, a filtering module, a data resampling module, and a smoothing module.
[0080] The data preprocessing module is used to preprocess the acquired complete flight data, including data fusion and time series correction;
[0081] The data segmentation module determines the start and end times of actions based on helicopter configuration, changes in control inputs, and flight status, and performs action segmentation on the original simulator objective data to obtain data for each action segment.
[0082] The batch processing module provides a batch processing entry point for multiple sets of action segment data, and simultaneously imports multiple sets of action segment data into one or more of the outlier processing module, filtering processing module, smoothing processing module, and data resampling module for processing;
[0083] The outlier processing module is used to identify and remove outliers from the action segment data after batch data processing.
[0084] The curve fitting processing module is used to correct the data after outlier removal;
[0085] The filtering module is used to recover the true information as much as possible from the action segment data containing interference. That is, the signal has various frequency components, and the module filters out specific frequency band components (generally noise) and retains the desired components.
[0086] The system also includes a smoothing module, which is used to generate a continuous and smooth data curve when the action segment data has a stepped shape.
[0087] The system also includes a data resampling module. When the sampling rate of the original simulator objective data does not meet the objective test requirements of the helicopter simulator, the data resampling module is used to increase the low sampling rate objective data to a high sampling rate data that meets the objective test requirements of the helicopter simulator.
[0088] Example 2
[0089] An efficient method for processing objective data from a helicopter flight simulator, employing the aforementioned efficient system for processing objective data from a helicopter flight simulator, includes the following steps:
[0090] The specific process of preprocessing the acquired data for the entire flight includes:
[0091] Data from different sources is merged to form a complete file of data for the entire flight;
[0092] Some parameters in the acquired flight data are extracted from the bus, while others require external sensor acquisition. Some parameters will have a certain delay, so time delay correction is needed for some parameters in the flight data during preprocessing.
[0093] The preprocessed data of the entire sortie is segmented, and the start and end times of the actions are determined based on the helicopter configuration, changes in control inputs and flight status. The original simulator objective data is segmented to generate action segment data.
[0094] It provides a batch processing entry point for multiple sets of action segment data, allowing multiple sets of action segment data to be simultaneously imported into one or more of the outlier processing module, filtering module, smoothing module, and resampling module for processing;
[0095] Traditional manual data processing methods require extensive manual corrections, which is extremely costly in terms of manpower and resources, especially for large datasets and numerous test points. Improving processing efficiency while ensuring data validity is crucial for the objective data processing work of simulators. The batch data processing module of this invention significantly improves work efficiency, reducing workload by 80% compared to manual processing.
[0096] Because helicopter simulator data is massive, covering various maneuvers such as hovering, climbing, descending, and level flight, and each maneuver segment contains multiple parameters to be processed, each parameter may require outlier handling, filtering, smoothing, and resampling. Considering that there are multiple datasets for the same maneuver in the objective data, and that the parameter characteristics have a certain similarity, batch processing can simultaneously process multiple data files and multiple parameters in a single batch.
[0097] By employing the batch processing module of this system, the parameters in the first action segment data file are processed first. After all parameters in that file are processed, the parameters in the next file are automatically processed, until all action segment data files have been processed. Batch processing can replace a large amount of repetitive work by engineers, greatly improving data processing efficiency.
[0098] The data after batch processing is used to identify and remove outliers. First, the ordinary seven-point second-order algorithm is used to identify and remove isolated outliers and clusters of outliers. Then, the optimized seven-point second-order algorithm is used to identify and remove endpoint outliers.
[0099] 1) Ordinary seven-point second-order algorithm
[0100] Outlier correction is performed using a standard 7-point second-order algorithm. Outlier correction refers to the entire process of identifying, removing, and correcting outliers in data. It can correct outliers in objective data.
[0101] The seven-point second-order algorithm uses a simple, efficient, and very practical low-order polynomial sliding fitting method to identify and remove outliers. Due to the influence of outliers, in order to avoid treating outliers as normal values and vice versa, the algorithm uses a forward difference formula to identify outliers. The calculation formula for the first 6 points is shown in (1).
[0102] (1)
[0103] Starting from point 7, the calculation formula is as follows:
[0104] (2)
[0105] In the formula: For flight test data, This is the interpolated data.
[0106] First, a check is performed. If the first 6 points are not outliers, the above formula is used to calculate point by point in chronological order. and new information .
[0107] For outliers, whose values are much larger or smaller than normal values, formula (3) can be used as a criterion for identification:
[0108] (3)
[0109] like Figure 3 As shown, under normal circumstances, the values of patches of fields in objective data will be relatively close. In this case, formula (4) can be used to determine the value of patches of fields:
[0110] (4)
[0111] If the frequency of consecutive outliers is less than or equal to the acceptable R value (R is the number of consecutive outliers that are acceptable, which can be set, usually R=3), then these points are considered outliers.
[0112] Theory shows that the 7-point second-order algorithm can identify both isolated outliers and clusters of outliers.
[0113] First, we examine the seven-point second-order algorithm's ability to identify isolated outliers. Taking 3.88385, 3.88385, 3.88621, 3.88503, 3.88149, 3.88503, 0, 3.88621, 3.91212, 3.9239, 3.91801, 3.88149, and 3.89092 as examples, the process of the seven-point second-order algorithm identifying outliers is as follows:
[0114] a) First, confirm that the first 6 points of this parameter are not outliers. Substitute the first 7 points (note that this is not the first 7 points of this data set, but the first 7 points of this parameter) 3.71775, 3.72011, 3.72129, 3.64354, 3.72246, 3.64, 3.70833 into equation (1) to obtain... The values are 3.72967, 3.70959, 3.69487, 3.6855, 3.68149, and 3.68275 respectively. Using (2), we obtain... =3.68953, calculated using equation (3) Then, using equation (3) for judgment, after calculation, the first 6 points Therefore, the first 6 points are not outliers;
[0115] b) Using (1), calculate the values of 3.88385, 3.88385, 3.88621, 3.88503, 3.88149, 3.88503, 0, ..., which are 3.89619, 3.8926, 3.87892, 3.88293, 3.88267, 3.88348, 0.92453, ..., according to formula (3), the E values are 0.0315389, 0.0311003, 0.0317648, 0.028084, 0.0208309, 0.0163166, 0.830504, ...
[0116] c) Based on the calculation results, determine whether there are outliers. Substitute them into (4) and calculate to obtain the result. Therefore, a value of 0 may be an outlier, which requires special attention. In this case, 0 cannot be directly assumed to be an outlier; further judgment is needed to determine whether the R points following the value of 0 satisfy the condition. or This outlier condition means that only when the number of consecutive outlier points is less than or equal to the R (the number of times an outlier is allowed to occur) set by the engineer can these points be considered outliers.
[0117] d) Continue calculating the data after the value 0. According to (4), set the R value to 3, and check the two data after the value 0, 3.88621 and 3.91212. Obviously, these two values do not meet the requirements. Therefore, the value 0 is considered an outlier and needs to be removed. If the three data points following the value 0 are also outliers, then the four data points, including the value 0, are reasonable variations and not outliers.
[0118] e) The results of the seven-point second-order algorithm for identifying isolated outliers are as follows: Figure 2 As shown:
[0119] Modify the data to 3.88385, 3.88385, 3.88621, 3.88503, 3.88149, 3.88503, 0, 0, 0, 3.9239, 3.91801, 3.88149, 3.89092, and perform outlier identification following the steps outlined above. The results are as follows: Figure 3 As shown:
[0120] If the number of outliers in a cluster exceeds three, but the R-value is still set to three, it will also lead to incorrect identification. The algorithm will consider this cluster of outliers to be normal mutations rather than outliers. However, setting the R-value too large will significantly reduce the algorithm's efficiency. Therefore, the program allows engineers to set the R-value based on the data, such as... Figure 4As shown, there are more than 10 outliers in this segment of objective data. With the R value set to 10, the outliers were successfully identified.
[0121] The process of identifying outliers using the 7-point second-order algorithm can be summarized as follows:
[0122] 1) Confirm whether the first six points are outliers;
[0123] 2) If outliers exist among the first six points, remove them and correct them using interpolation. If they are not outliers, calculate according to the formula. The value;
[0124] 3) Calculate each data point E;
[0125] 4) Based on Determine the relationship between the magnitude of E and E. Is it an outlier?
[0126] 5) Continue the calculation, repeating steps 3) to 4) until the data segment has been checked.
[0127] Suppose a certain Although it is a droma value, it cannot be directly considered a droma value; further judgment is needed. , Do the following R points also satisfy this condition? This condition applies only when the number of consecutive outlier points is less than the R value (the allowed number of outlier occurrences) set by the engineer; this is a crucial point to note. Based on experience, the R value is usually set to 3.
[0128] In real-world objective data, outliers may appear at the endpoints, requiring verification of whether the seven-point second-order algorithm can identify them.
[0129] Suppose that 3.88385, 3.88385, 3.88621, 3.88503, 3.88149, 3.88503, 0, 3.88621, 3.91212, 3.9239, 3.9180, 3.88149, 3.89092 are the starting data points of the objective data. There is one outlier within this range. Additionally, outliers exist in the middle range 3.88385, 3.88385, 3.88621, 3.88503, 3.88149, 3.88503, 0, 0, 0, 3.9239, 3.91801, 3.88149, 3.89092. Using the seven-point second-order algorithm to identify outliers in this segment of objective data yields a different result: the value of 0 at the starting data point is not considered an outlier. Figure 5 As shown:
[0130] The black line represents the original data curve, while the red line represents the curve after outliers have been identified, removed, and corrected. As you can see, the algorithm can identify outliers that appear at non-endpoints, but it cannot effectively identify outlier points at endpoints.
[0131] Using the standard 7-point second-order algorithm to remove endpoint outliers is not ideal, as it fails to identify outliers at the endpoints. Therefore, the algorithm needs to be optimized to identify and remove endpoint outliers. A detailed description follows.
[0132] 2) Optimization of the seven-point second-order algorithm
[0133] In-depth theoretical research revealed that the reason why the seven-point second-order algorithm cannot identify endpoint outliers lies in the different calculation formulas: for the first 6 points, the calculation formula is as shown in (1), while for other data points it is calculated according to (2). However, the formula for calculating the E value is also used, so when an outlier appears at the endpoint, because Changes in [something] will cause changes in E, which in turn will affect [something]. This condition for judgment. That is, if The calculation uses the first 6 points, which may lead to the inability to correctly identify outliers. According to formula (2), the first 6 points are used in the calculation of the first 12 points, so the first 12 points all belong to the endpoint range.
[0134] For this reason, for identifying endpoint outliers, a method can be adopted: reverse-engineer the objective data and perform a 7-point second-order algorithm for outlier identification. The result of this reverse algorithm is then used to replace the original identification results for the first 12 endpoints. This method can only be applied to offline processing of objective data. However, in practical engineering, to improve the efficiency of the algorithm, it is not necessary to reverse-engineer all the objective data once; it is only necessary to reverse-engineer from the 18th point. Furthermore, since a forward algorithm has already been performed, points 13 to 18 contain no outliers. Therefore, the outlier identification results for endpoints 1 to 12 calculated in reverse are also reliable.
[0135] Now there is only one problem left: if there are outliers among the first 6 points, we can use methods such as polynomial fitting to process the outlier data.
[0136] As can be seen, the upgraded seven-point second-order algorithm can effectively identify all outlier points and remove them. Finally, the program calls the polynomial fitting method to correct the removed points.
[0137] To ensure data integrity, after identifying and removing outliers, it is necessary to correct the outliers. If n consecutive data points (i.e., y...) are identified... k y k+1 , ..., y k+n ) is an outlier, and we can find m correct data points (i.e., y) ahead. k-1 yk-2 , ..., y k-m And find m correct data points (i.e., y) afterward. k+n+1 y k+n+2 , ..., y k+n+m The correction value can be obtained using the Lagrange interpolation formula shown below:
[0138] (5)
[0139] The above formula is used to correct the outlier positions, and the corrected values are used to replace the original outliers.
[0140] Outliers appeared in the right fuel temperature parameter in the simulator's objective data. These outliers should be removed and corrected during objective data preprocessing. The program uses an optimized seven-point second-order algorithm to identify and remove outliers, and then uses a polynomial fitting method in the filtering module to correct the removed points.
[0141] like Figure 6 As shown, the black curve represents the original objective data, with outliers existing both at the endpoints and in the middle of the data. The red curve represents the objective data after removal and correction. It can be seen that the optimized seven-point second-order algorithm can effectively identify all outlier points in the simulator's objective data, and then use a polynomial fitting method to correct the removed points.
[0142] Curve fitting is performed on the data after outlier processing. Polynomial fitting technology is used to smooth the data after outlier processing to generate continuous data curves.
[0143] In practical applications, conventional curve fitting algorithms can over-process objective data peaks, leading to data distortion. Furthermore, they often produce unsatisfactory interpolation results when data sampling rates are low or data points are missing. Therefore, it is necessary to optimize conventional curve fitting algorithms by using polynomial curve fitting algorithms to overcome these shortcomings.
[0144] The principle of the polynomial curve fitting algorithm is to call the fitting function and the fitting verification function to calculate the curve data points. First, an appropriate order and number of data points are selected. Then, the data segment is fitted, and the fitting formula is obtained. The x-value is then substituted into the fitting formula, and the result of the fitting formula is used to replace the original data. The data segment is advanced until all data for that parameter has been fitted.
[0145] The well-fitted analytical function is:
[0146] (6)
[0147] in, is the coefficient, and n is the degree of the polynomial.
[0148] The fitting function returns the coefficients of an nth-order polynomial p(x), which is the best fit to the data in y. The coefficients in p are decreasing, and the length of p is n+1.
[0149] The fitting validation function is a function that calculates the corresponding y-value using polynomial coefficients and x-values. In this algorithm, the polynomial coefficients are the polynomial coefficients calculated from the previous fitting function, i.e., ... The function calculates the p(x) value, or y value, for each point based on the x value. The set of these y values is the processed dataset.
[0150] use To fit these data points, the essence is to find the coefficients a, b, and c that minimize the sum of squared vertical distances between all data points and the fitted curve, i.e., minimize the sum of squared errors E.
[0151] (7)
[0152] To find the optimal coefficients a, b, and c, the following system of equations needs to be solved:
[0153] (8)
[0154] (9)
[0155] (10)
[0156] Expanding equations (8), (9), and (10), we obtain the following linear equations:
[0157] (11)
[0158] (12)
[0159] (13)
[0160] Substituting the known fixed-point numbers into equations (11), (12), and (13), we can obtain a system of equations containing coefficients a, b, and c. Solving this system of equations will give us the values of coefficients a, b, and c, and thus the values of the fitted polynomial.
[0161] Figure 7 shows the results of processing the objective data using the polynomial fitting method.
[0162] Theoretically, the shorter the fitting length, the closer the processed result is to the original curve; the higher the power, the closer the result is to the original curve. When smoothing data, if the chosen fitting length is too short or the power is too high, it's equivalent to no smoothing at all. However, if the fitting length is too long or the power is too low, although smoothing can be achieved, it may distort the data, rendering the data processing meaningless. Therefore, in practical processing, it is crucial to choose an appropriate fitting length and power.
[0163] Based on past experience, the default data length is 13 data points, and the highest power of the fit is 3. To accommodate more situations, engineers can manually set the fit length and polynomial power to modify the processed curve, achieving different smoothing effects and meeting the processing needs of different data.
[0164] Based on the research and processing of the data, the advantages of the polynomial curve fitting method are as follows:
[0165] Some parameter data points are missing, which can be filled in;
[0166] If the parameter sampling rate is insufficient, interpolation can be performed.
[0167] After curve fitting, the filtering module is called to remove noise from the data, using first-order filtering and low-pass filtering methods.
[0168] a) First-order filtering
[0169] Ordinary first-order filtering methods can introduce phase differences, especially noticeable after repeated processing. To eliminate these phase differences, the system upgrades the first-order filtering method. Since the values at each point in the acquired data are constant, both forward and reverse processing are applicable. The system processes the data once in the forward direction and once in the reverse direction, then sums the results and calculates the average. This process cancels out the phase differences, thus eliminating them.
[0170] b) Low-pass filter
[0171] Conventional low-pass filters abruptly truncate data in the frequency domain, preventing smooth data transitions. Butterworth digital filters avoid this. A Butterworth digital filter requires a low-pass cutoff frequency and a high-impedance cutoff frequency. The high-impedance cutoff frequency must be higher than the low-pass cutoff frequency and lower than the quiescent frequency. During data filtering, frequencies lower than the low-pass cutoff frequency pass through, frequencies higher than the high-impedance cutoff frequency are filtered, and frequencies in between gradually decrease from passing through to filtering until no frequencies can pass through.
[0172] A Butterworth digital filter was used to implement low-pass filtering and process flight test data, as shown in Figure 7. The black curve represents the original data for a certain parameter, and the red curve represents the data after low-pass filtering. It can be seen that the processed data is smoother.
[0173] In one possible embodiment, when some parameters of the original simulator objective data exhibit a stepped shape, spline processing is used to generate a continuous and smooth data curve.
[0174] When the sampling rate of the original simulator objective data does not meet the objective test requirements of the helicopter simulator, the data resampling module is used to improve the low sampling rate objective data to a high sampling rate data that meets the objective test requirements of the helicopter simulator.
[0175] The sampling rate of some parameters in the objective data obtained from the real aircraft is low, which can lead to distortion or even loss of information in the data, failing to meet the objective data requirements of the simulator. As shown in Figure 8, due to the low sampling rate, the curve of this data exhibits a stepped shape. Directly using such data as objective data for the simulator will cause discontinuities in the simulator's flight state, resulting in visual stuttering, which does not match the actual flight state of the helicopter. Therefore, it is necessary to resample this type of data. Data resampling includes linear interpolation and forward interpolation methods.
[0176] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0177] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.
Claims
1. A system for efficient processing of objective data of a helicopter flight simulator, characterized in that, The system includes a data preprocessing module, a data segmentation module, a batch processing module, an outlier processing module, a curve fitting processing module, and a filtering processing module. The data preprocessing module is used to preprocess the acquired complete flight data, including data fusion and time-series correction. The data segmentation module determines the start and end times of actions based on helicopter configuration, changes in control parameters, and flight status, and segments the original simulator objective data into action segments to obtain data for each action segment. The batch processing module provides a batch processing entry point for multiple sets of action segment data, simultaneously importing multiple sets of action segment data and multiple data parameters into one or more other processing modules for processing. The outlier processing module is used to identify and remove outliers from the action segment data after batch data processing. The curve fitting module is used to correct the data after outlier removal; the filtering module is used to restore the true information from the action segment data containing interference information.
2. A system for efficient processing of objective data of a helicopter flight simulator according to claim 1, characterized in that, The system also includes a smoothing module, which is used to generate a continuous and smooth data curve when the action segment data has a stepped shape.
3. The system for efficient processing of objective data of a helicopter flight simulator according to claim 1, characterized in that, The system also includes a data resampling module. When the sampling rate of the original simulator objective data does not meet the objective test requirements of the helicopter simulator, the data resampling module is used to improve the low sampling rate objective data to a high sampling rate data that meets the objective test requirements of the helicopter simulator.
4. A method for efficiently processing objective data from a helicopter flight simulator, employing the efficient processing system for objective data from a helicopter flight simulator as described in any one of claims 1-3, characterized in that, Includes the following steps: Acquire and preprocess all flight data; The entire preprocessed flight data is segmented, and the start and end times of the actions are determined based on the helicopter configuration, changes in control inputs, and flight status. The original simulator objective data is segmented to generate action segment data. Provides a batch processing entry point, allowing multiple sets of action segment data to be simultaneously imported into one or more other processing modules for processing; Identify and remove outliers from the action segment data after batch data processing; The discarded points were corrected by using a polynomial curve fitting method.
5. The efficient processing method for objective data from a helicopter flight simulator according to claim 1, characterized in that: The specific process of acquiring and preprocessing the data for the entire flight includes: Data from different sources is merged to form a complete file of data for the entire flight; Some parameter data in the entire flight data are extracted from the bus, while others need to be collected by external sensors. Some parameter data will have a certain delay. Therefore, time delay correction is performed on the delayed data in the preprocessing.
6. The efficient processing method for objective data from a helicopter flight simulator according to claim 1, characterized in that, The specific process of identifying and removing outliers from batch-processed data includes: The ordinary seven-point second-order algorithm is used to identify and remove isolated outliers and clusters of outliers in the middle of the data based on the innovation value E and the number of consecutive occurrences of acceptable outliers R. For outliers at the front end of the data, a seven-point second-order algorithm calculation is performed in reverse starting from the 18th point of the data. The result of the reverse calculation is used to replace the result of the forward identification, thus completing the identification and removal of outliers.
7. The efficient processing method for objective data from a helicopter flight simulator according to claim 1, characterized in that, For the locations of the identified and removed outliers, Lagrange interpolation is used to correct the data.
8. The efficient processing method for objective data from a helicopter flight simulator according to claim 1, characterized in that, The filtering module is invoked to restore the true information from the action segment data containing noise information. The methods used are first-order filtering and low-pass filtering.
9. The efficient processing method for objective data from a helicopter flight simulator according to claim 8, characterized in that, When using a first-order filtering algorithm, the data sequence is forward filtered, and then the same data sequence is reverse filtered. The corresponding points of the two filtering results are added together and the average value is taken as the final filtered output.
10. The efficient processing method for objective data from a helicopter flight simulator according to claim 8, characterized in that, When using the Butterworth low-pass filter algorithm, the low-pass cutoff frequency and the high-impedance cutoff frequency are set to filter the data.
11. The efficient processing method for objective data from a helicopter flight simulator according to claim 1, characterized in that, When some parameters of the original simulator's objective data exhibit a stepped shape, spline processing is used to generate continuous and smooth data curves.
12. The efficient processing method for objective data from a helicopter flight simulator according to claim 1, characterized in that, When the sampling rate of the original simulator objective data does not meet the objective test requirements of the helicopter simulator, the low sampling rate objective data is increased to a high sampling rate data that meets the objective test requirements of the helicopter simulator.
13. The efficient processing method for objective data from a helicopter flight simulator according to claim 12, characterized in that, For switch parameters and jump variable parameters, front-point interpolation is used to increase data density.
14. The efficient processing method for objective data from a helicopter flight simulator according to claim 12, characterized in that, For other types of parameters, linear interpolation is used to increase data density.
15. A computer-readable storage medium storing a computer program; characterized in that, When the computer program is executed by the processor, it implements an efficient method for processing objective data of a helicopter flight simulator according to any one of claims 4-14.