A method for dynamic envelope extraction of spacecraft parameters based on curve matching search

Abstract

This invention proposes a method for extracting the dynamic envelope of spacecraft parameters based on curve matching search, comprising the following steps: reading telemetry parameter files to obtain parameter sequence curves as templates for matching search; reading historical telemetry data to obtain historical parameter sequence curves as objects for traversal and retrieval; traversing the historical parameter sequence curves, with a window length equal to the number of values ​​in the template sequence and the initial point of the window sequence as the initial matching point, and calculating the matching vector and matching value between each window sequence and the parameter sequence curve; sorting the matching values ​​in ascending order and correspondingly sorting the matching starting points; optimizing the matching starting points for interference reduction based on distance boundaries; extracting the optimized matching starting points based on matching thresholds and performing curve positioning to form a curve set; and calculating the upper and lower dynamic envelopes of the spacecraft telemetry parameters based on the maximum and minimum values ​​of the curve points in the curve set.

Description

Technical Field

[0001] This invention belongs to the field of spacecraft testing and health assessment technology, and particularly relates to a method for dynamic envelope extraction of spacecraft parameters based on curve matching search. Background Technology

[0002] The dynamic envelope of spacecraft telemetry parameters is crucial data for verifying whether the spacecraft's design and performance meet requirements. It can be used for anomaly detection and health assessment. However, the diverse types and large scale of spacecraft telemetry parameters make manually selecting historical curves and extracting the dynamic envelope costly and inconsistent.

[0003] Current automated methods typically simulate the dynamic envelope of a specific telemetry parameter by setting upper and lower threshold limits for curve fluctuations using expert knowledge. However, these methods rely on experience, are inefficient, and are difficult to widely apply to large-scale spacecraft telemetry parameters.

[0004] Currently, methods based on neural networks and adaptive threshold estimation design and train network structures for specific data features, calculating adaptive upper and lower thresholds through network reconstruction bias to generate dynamic envelopes for telemetry parameters. While these methods can estimate the envelope, they are only applicable to specific data ranges and cannot expand the input category range. Furthermore, they somewhat ignore the authenticity of the dynamic envelope data, failing to extract the true dynamic envelope of spacecraft parameters. Curve matching search technology is an important method in data mining, utilizing data similarity to retrieve fitting curves. However, its application in spacecraft parameter dynamic envelope extraction is extremely limited. This is because several neighboring points before and after the starting point of the historical telemetry parameter curve matching can affect the localization of the fitting curve, resulting in numerous interference points in the curve matching search results and reducing the accuracy of spacecraft parameter dynamic envelope extraction. Summary of the Invention

[0005] To address the limitations of current machine learning methods such as adaptive threshold estimation in calculating the dynamic envelope of spacecraft parameters over a wide range and in extracting the dynamic envelope of real spacecraft, as well as the problem that curve matching search methods suffer from numerous interference points, this invention proposes a method for extracting the dynamic envelope of spacecraft parameters based on curve matching search.

[0006] The present invention is achieved through the following technical solution.

[0007] A method for dynamic envelope extraction of spacecraft parameters based on curve matching search includes the following steps:

[0008] Step 1: Read the telemetry parameter file and obtain the parameter sequence curve as a template for matching search;

[0009] Step 2: Read historical telemetry data and obtain historical parameter sequence curves as the objects for traversal and retrieval;

[0010] Step 3: Traverse the historical parameter sequence curves described in Step 2. The window length is the number of values ​​in the template sequence, and the initial point of the window sequence is the initial matching point. Calculate the matching vector and matching value between each window sequence and the parameter sequence curve described in Step 1.

[0011] Step 4: Sort the matching values ​​in ascending order and sort the matching starting points accordingly;

[0012] Step 5: Optimize the matching starting point from Step 4 based on the distance boundary to reduce interference;

[0013] Step 6: Based on the matching threshold, extract the matching starting point optimized in Step 5, locate the curve, and form a curve set;

[0014] Step 7: Calculate the upper and lower dynamic envelopes of the spacecraft telemetry parameters based on the maximum and minimum values ​​of the points on the curve set described in Step 6.

[0015] The beneficial effects of this invention are:

[0016] 1. This invention fills the gap in the application of curve matching search methods in the dynamic envelope extraction of spacecraft parameters. It can effectively eliminate curve matching interference points, improve the authenticity and accuracy of dynamic envelope extraction of spacecraft parameters, and provide prior input for other supervised learning methods, such as fault diagnosis methods based on metric learning, thereby improving the efficiency of anomaly detection of spacecraft parameters.

[0017] 2. This invention effectively removes interference points in the preceding and following positions of the fitting curve data by optimizing the matching points based on distance boundaries, achieving high-precision selection of the real fitting curve, improving the quality of dynamic envelope extraction of spacecraft parameters, and increasing the testing efficiency of spacecraft anomaly detection based on the envelope discrimination method.

[0018] 3. Compared with neural networks and adaptive threshold estimation methods, this invention can extract the dynamic envelope of real spacecraft parameters and is applicable to the dynamic envelope extraction of multiple types of telemetry parameters of spacecraft without training.

[0019] 4. This invention does not require assumptions about the distribution of process data; it is a purely data-driven algorithm with wider applicability.

[0020] 5. This invention fills the gap in the application of curve matching search methods in the dynamic envelope extraction of spacecraft parameters. Attached Figure Description

[0021] Figure 1This is a historical measured sequence curve of temperature parameters on the +Y side of a spacecraft's midplate, according to an embodiment of the present invention.

[0022] Figure 2 This is a template sequence curve of temperature parameters on the Y-side of the midplate in a spacecraft according to an embodiment of the present invention;

[0023] Figure 3 This is a curve diagram obtained from historical measured sequence curves using the curve matching search method in an embodiment of the present invention;

[0024] Figure 4 This is a dynamic upper and lower envelope diagram of the temperature on the Y side of the middle plate in a spacecraft according to an embodiment of the present invention;

[0025] Figure 5 This is a two-dimensional dot plot of the matching vector before optimization in an embodiment of the present invention;

[0026] Figure 6 This is a two-dimensional dot plot of the optimized matching vector in an embodiment of the present invention. Detailed Implementation

[0027] Exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the embodiments shown and described in the drawings are merely exemplary and are intended to illustrate the principles and spirit of the present invention, and are not intended to limit the scope of the present invention.

[0028] like Figure 1 As shown, the present invention provides a method for dynamic envelope extraction of spacecraft parameters based on curve matching search, which specifically includes the following steps:

[0029] Step 1: Read the telemetry parameter file and obtain the parameter sequence curve as a template for matching search;

[0030] In this embodiment, the telemetry parameter file is read to retrieve the telemetry parameters to be detected within a specified time interval. The structured telemetry parameter data generated after format conversion is as follows:

[0031] item content Parameter name String Parameter code (parameter number) String time Digital vectors Engineering value Digital vectors Source code String vectors

[0032] In this embodiment, obtaining the parameter sequence curve as a template for matching search specifically involves: using the engineering value sequence x within a specified time interval. 1×n temp As a template in curve matching search, n and 1 represent the template sequence length and parameter type, respectively. In specific implementation, considering the actual scenario of spacecraft dynamic envelope detection, the template sequence length n>1.

[0033] Step 2: Read historical telemetry data and obtain historical parameter sequence curves as the objects for traversal and retrieval;

[0034] In this embodiment, reading the telemetry historical data refers to reading telemetry parameters within a specified historical time interval. The structured telemetry parameter data generated after format conversion is as follows:

[0035] item content Parameter name String Parameter code (parameter number) String time Digital vectors Engineering value Digital vectors Source code String vectors

[0036] The sequence of engineering values ​​X for the specified historical time interval 1×m As the object of traversal retrieval, where m and 1 represent the sequence length and parameter types respectively, m>n.

[0037] Step 3: Traverse the historical parameter sequence curves described in Step 2. The window length is the number of values ​​in the template sequence, and the initial point of the window sequence is the initial matching point. Calculate the matching vector and matching value between each window sequence and the parameter sequence curve described in Step 1. The specific steps are as follows:

[0038] 3.1 Using a window of template sequence length for spacecraft telemetry data X 1×m The process involves iterating through the sequences within the window and calculating matching vectors between them and the template sequence. Specifically:

[0039] Traversing curve X 1×m With a window size of 1×n and a sliding window step size of k=1, we obtain the window sequence set Y={y 1×n 1,y 1 ×n 2,…,y 1×n m-n+1 The set of window sequences contains m-n+1 sequences. The initial point p of the window sequence is called the initial matching point, and the corresponding set of starting coordinate points is...

[0040] P = {p1, p2, ..., p} m-n+1} = {1,2,…,m-n+1}.

[0041] 3.2 Calculate the relationship between each window sequence and the template curve x using the curve matching method. 1×n temp The matching vector. Specifically:

[0042] For each window sequence y, first calculate its relationship with x. 1×n temp The scaling factor α of (hereinafter referred to as x) is defined as follows:

[0043]

[0044] At this point, the first matching coefficient d1(x,y) is calculated, and it is defined as follows:

[0045]

[0046] The second matching coefficient d2(x,y) is defined as follows:

[0047]

[0048] As described above, after processing each window sequence y using the curve matching method, the matching vector d(x,y) is defined as follows:

[0049]

[0050] 3.3 The matching vector is symmetricized, i.e., d(x,y) = d(y,x). Based on this, the matching value v(x,y) of y is defined as follows:

[0051] v(x,y)=β(d1(x,y)+d1(y,x))+(1-β)(d2(x,y)+d2(y,x))

[0052] Where β represents the weight coefficient, which measures the relative importance of two vector elements in the matching vector d(x,y), 0<β<1, and unless otherwise specified, β is usually set to 0.5.

[0053] It can be seen that the set of window sequences Y = {y 1×n 1,y 1×n 2,…,y 1×n m-n+1 After the above processing, a set of matching vectors D = {d} can be obtained. 2×1 1,d 2×1 2,…,d 2×1 m-n+1} and the matching value set V = {v1, v2, ..., v m-n+1}

[0054] Step 4: Sort the matching values ​​in ascending order and sort the matching starting points accordingly;

[0055] The idea behind this step is that the set of matching starting coordinate points P obtained in step three represents the location of each window sequence, and each window sequence corresponds to a matching value in the set of matching values ​​V. Therefore, the element positions of set P and set V correspond one-to-one, and their position mapping relationship is as follows:

[0056] Therefore, in this embodiment, the elements of the matching value set V are sorted in ascending order to obtain a new sorted matching value set V′, which is then mapped... We obtain the sorted set of new matching starting coordinates P′.

[0057] Step 5: Optimize the matching starting point from Step 4 based on the distance boundary to reduce interference;

[0058] The idea behind this step is that during curve matching search in step three, the window sequence corresponding to the starting point of the matching curve and the points before and after it has a low matching value, making it easy to be identified as the matching curve. This results in poor positioning and matching accuracy of the matching curve. The points before and after the starting point are called interference points.

[0059] To address this issue, this embodiment utilizes a distance boundary-based interference reduction method to optimize curve matching search performance. Specifically, it includes a distance boundary-based method for judging the interference of the matching starting point and a distance boundary-based method for optimizing the interference reduction of the matching starting point. The algorithms for the two methods are as follows:

[0060]

[0061]

[0062]

[0063] In Algorithm 1, the distance boundary value `thres` is affected by the distance boundary coefficient `θ` and the template sequence length `n`. The `int(θ×n)` function represents rounding. The boundary coefficient ranges from 0 to θ to 1. When θ > 1, the false negative rate for matching the starting point is relatively high. Algorithm 1 outputs a judgment flag. If the output is `True`, it means that the current starting coordinate point can be merged into the set `S` in Algorithm 2; otherwise, it cannot.

[0064] Step Six: Based on the matching threshold, extract the matching starting points optimized in Step Five, and perform curve localization to form a curve set; the specific steps are as follows:

[0065] 6.1 For two elements in the matching vector, perform threshold judgment according to the matching threshold, and let the matching thresholds of the first and second elements be d1_thres and d2_thres respectively;

[0066] 6.2 Traverse the set of matching starting coordinates P″ after interference reduction to obtain the matching vector at the corresponding position, and compare it with the threshold. The position point where both elements are less than the matching threshold is considered to be the starting position point of the adaptation curve.

[0067] 6.3 Take the curves of the corresponding positions in the window sequence set Y obtained in step 3 as the localized adaptation curves to obtain an adaptation curve set X.

[0068] The algorithm for the above steps is as follows:

[0069]

[0070] Step 7: Calculate the upper and lower dynamic envelopes of the spacecraft telemetry parameters based on the maximum and minimum values ​​of the points on the curves in the curve set described in Step 6; the specific rules are as follows:

[0071] If the set of curves Then the sequence of engineering values ​​X for the specified historical time interval 1×m Does not include adaptation curves;

[0072] If the number of elements in the curve set X is less than 2, it is considered that the current data volume cannot effectively extract the dynamic upper and lower envelopes; otherwise, for all sequence curves in set X, traverse each point of all curves and calculate the maximum and minimum values ​​of all curves at this point. After traversal, two sequences are obtained, namely the maximum value sequence and the minimum value sequence. The length of each sequence is equal to the length n of the template sequence curve, which serves as the dynamic upper and lower envelopes of the extracted spacecraft parameters.

[0073] The following is an example of extracting the dynamic envelope of the temperature on the Y side of the middle plate of a certain spacecraft.

[0074]

[0075] Figure 1 This is historical measured data of the temperature on the Y side of the middle plate of a certain spacecraft. Figure 2 This is template data for the temperature of the middle plate + Y side of the spacecraft to be tested. In the original working mode, test engineers used a dynamic threshold as a dynamic envelope to determine whether the curve met expectations.

[0076] Figure 3 To obtain the fitting curves from historical data obtained using the curve matching search method of this invention, the fitting curves are then extracted according to the dynamic upper and lower envelope extraction method for spacecraft parameters. Figure 4 The dynamic upper and lower envelopes of the temperature on the Y side of the spacecraft's midplate are shown in the image. The dark area between the upper and lower envelopes represents the range of normal value fluctuations and can be used for applications such as anomaly detection.

[0077] Figure 5 and Figure 6 These are two-dimensional point plots of the matching vector before and after reducing interference at the starting position of the matching. The light-colored points represent the points of the adaptation curve, and the dark-colored points are other points. It can be seen that the dark-colored interference points are close to the light-colored points, which will interfere with the matching threshold judgment.

[0078] The results of spacecraft dynamic envelope extraction show that the method of this invention is easier to automate than the original method, reduces reliance on expert experience, and provides a quantitative analysis tool. Furthermore, compared to conventional curve matching search methods, the improved matching search method proposed in this invention can effectively reduce the adverse effects of interference points on fitting curve detection, thereby improving the accuracy of spacecraft dynamic envelope extraction.

[0079] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A spacecraft parameter dynamic envelope extraction method based on curve matching search, characterized in that, The method comprises the following steps: Step one, reading a telemetry parameter file to obtain a parameter sequence curve as a template for matching search; Step two, reading telemetry history data to obtain a history parameter sequence curve as an object for traversal search; Step three, traversing the history parameter sequence curve in step two, with a window length being the number of values of the template sequence, and an initial point of the window sequence being a matching initial point, to calculate a matching vector and a matching value of each window sequence and the parameter sequence curve in step one; Step four, arranging the matching values in ascending order and sorting the matching initial points correspondingly; Step five, performing interference reduction optimization on the matching initial points in step four based on distance boundary; Step six, extracting the matching initial points after optimization in step five based on a matching threshold, and performing curve positioning to form a curve set; Step seven, calculating an upper and lower dynamic envelope of a spacecraft telemetry parameter according to the maximum and minimum values of each point in the curve set in step six; The reading of telemetry history data is reading telemetry parameters in a specified historical time interval; Step six comprises the following specific steps: 6.1 performing threshold judgment on two elements in the matching vector according to a matching threshold; 6.2 traversing the matching initial coordinate set after interference reduction to obtain a matching vector at a corresponding position, and comparing the matching vector with the threshold, and regarding a position point at which both elements are smaller than the matching threshold as a starting position point of the adaptive curve; 6.3 The set of window sequences obtained in step three is combined The set of adapted curves is obtained by combining the curves of the corresponding position points as adapted curves of the position points ; The calculation of the upper and lower dynamic envelope of the spacecraft telemetry parameter comprises the following specific rules: if the set of curves then the sequence of engineering values for the specified historical time interval does not contain an adapted curve; If the number of elements of the curve set is less than 2, it is considered that the current data volume cannot effectively extract the dynamic up and down envelope; otherwise, for all sequence curves in the set , all points of all curves are traversed, and the maximum value and the minimum value of the point at this moment in all curves are calculated respectively, after the traversal is completed, two sequences are obtained, which are the maximum value point sequence and the minimum value point sequence respectively, the length of each sequence is equal to the length of the template sequence curve , which is the dynamic up and down envelope line of the extracted spacecraft parameters.

2. The method of claim 1, wherein, The reading of the telemetry parameter file is reading a to-be-detected telemetry parameter in a specified time interval.

3. The method of claim 1 or 2, wherein, The obtaining of the parameter sequence curve as the template for matching search is that an engineering value sequence in the specified time interval is taken as a template in curve matching search.

4. The method of claim 1, wherein, The engineering value sequence in the specified historical time interval is taken as the object for traversal search.

5. The method of claim 1 or 2 or 4, wherein, Step three comprises the following specific steps: 3.1 traversing the spacecraft telemetry data by using a window with a template sequence length to calculate a matching vector of a sequence in the window and the template sequence; 3.2 calculating a matching vector of each window sequence and the template curve by using a curve matching method; 3.3 performing symmetrization processing on the matching vector to obtain a matching vector set and a matching value set.

Images