A phase correction method and system based on piecewise polynomial fitting

Abstract

The application relates to the technical field of electromagnetic signal direction finding, and discloses a phase correction method and system based on segmented polynomial fitting. The method is based on correction data, a relationship between phase difference correction data and frequency points is modeled through a segmented polynomial fitting method, and correction data of any frequency point is calculated. The application solves the problems of low direction finding accuracy, incapability of judging the effectiveness of correction, large storage space occupation and the like in the prior art.

Description

Technical Field

[0001] This invention relates to the field of electromagnetic signal direction finding technology, specifically a phase correction method and system based on piecewise polynomial fitting. Background Technology

[0002] There are five common methods for electromagnetic signal direction finding: interferometer method, amplitude comparison method, spatial spectrum estimation method, Doppler method, and time difference of arrival method. Because the interferometer method offers advantages such as high sensitivity and high accuracy in direction finding, it is widely used in fields such as electronic warfare and radar.

[0003] The direction finding principle of a phase interferometer is as follows: Figure 1 As shown.

[0004] Where the antenna element spacing is d, and the signal incident angle is θ. From Figure 1 It can be seen that, due to the order in which the electromagnetic wave signals arrive at the antenna array elements, the signal received by the left array element is delayed compared to the signal received by the right array element. c is the speed of light, which causes the signal received by the left array element to lag behind that received by the right array element in phase. as follows:

[0005]

[0006] Where f is the frequency of the electromagnetic wave signal. As shown in equation (3.1), for a system with a fixed l, the phase difference between the two array elements has a linear relationship with the signal frequency. However, in actual direction finding systems, due to individual differences in antennas, mutual coupling effects between antennas, temperature changes, and other reasons, there will be phase inconsistencies, i.e., phase deviations, between array element channels, which will affect the direction finding accuracy. Therefore, effective correction methods are needed to improve the direction finding performance.

[0007] Traditional phase correction methods such as Figure 2 As shown, it is mainly divided into phase difference correction data acquisition, phase difference correction data distribution, and phase difference correction data application.

[0008] For broadband direction finding systems, the phase difference correction data of a subset of frequencies is typically acquired first (the relationship between phase difference and frequency is as follows). Figure 3 (As shown), then the phase difference correction data at the corresponding frequency point is sent to the subsequent processing platform, based on the phase difference. Theoretically, there is a linear relationship between the phase difference correction data and the signal frequency f. The processing platform uses phase difference correction data from some frequency points to linearly interpolate and calculate the correction data for other frequency points, thereby compensating for channel phase deviation and reducing phase inconsistency. The specific interpolation method is as follows:

[0009]

[0010] Where (x0,y0) is the starting (frequency point, phase difference) correction data, (x1,y1) is the ending (frequency point, phase difference) correction data, and x∈(x0,x1).

[0011] The above method has the following problems:

[0012] 1) Lack of judgment on the validity of the collected data. If the correction frequency point is invalid but is still used, it will directly affect the direction finding accuracy of the interpolation frequency point; in addition, the validity of the correction cannot be judged.

[0013] 2) such as Figure 3 As shown, the phase difference correction data exhibits a segmented relationship with frequency (the reason is explained in 5.1.1). Furthermore, due to multi-stage reflections and antenna coupling in the link, the phase difference data in each segment is not strictly linearly related to frequency, but rather exhibits fluctuations and glitches. Linear interpolation methods cannot effectively capture these characteristics.

[0014] 3) The calibration time and the number of initial calibration frequency points are strongly positively correlated. Furthermore, as the number of data collected increases, the amount of calibration data sent to the subsequent processing platform will increase, thus occupying more storage space. Summary of the Invention

[0015] To overcome the shortcomings of existing technologies, this invention provides a phase correction method and system based on piecewise polynomial fitting, which solves the problems of low direction finding accuracy, inability to determine the effectiveness of correction, and large storage space consumption in existing technologies.

[0016] The technical solution adopted by the present invention to solve the above problems is:

[0017] A phase correction method based on piecewise polynomial fitting is proposed. Based on the correction data, the relationship between the phase difference correction data and the frequency point is modeled by piecewise polynomial fitting, and the correction data at any frequency point is calculated.

[0018] As a preferred technical solution, the steps include:

[0019] S1, Data Acquisition: Acquire phase difference correction data at each frequency point;

[0020] S2, Data Processing: The acquired phase difference correction data is processed based on piecewise polynomial fitting;

[0021] S3, Parameter Sending: Sends the processed phase difference correction data to the subsequent processing platform;

[0022] S4, Parameter Application: The subsequent processing platform applies the processed phase difference correction data to reconstruct phase difference correction data at arbitrary frequencies.

[0023] As a preferred technical solution, step S2 includes the following steps:

[0024] S21, Phase difference estimation: Perform phase difference estimation to obtain the transformed phase difference correction data;

[0025] S22, Piecewise polynomial fitting: Piecewise polynomial fitting is performed based on the phase difference correction data transformed in step S21.

[0026] S23, determine whether the current correction meets the requirements.

[0027] As a preferred technical solution, in step S21, it is assumed that the number of points in each segment of correction data is at least T. 翻转点数 That is, the maximum phase difference between two adjacent frequency points is Only then can we determine whether the data has been flipped;

[0028] Assuming we have current data Data with the next shot Represents frequency f i Phase difference data at time, and Abbreviated as TJ 翻转大到小 , Abbreviated as TJ 翻转小到大 In addition, there is a threshold T for the number of unflipped items. 未翻转 Segment count threshold T 段数 The detailed process for estimating the phase difference is as follows:

[0029] Assuming the phase difference of the data in this correction is 1 / 3 And initialized to 2π; unflipped number N 未翻转 Set to 0; segment number N 段数 Set to 0; toggle the start flag F 翻转开始 Set to false, flip time marker F 翻转时刻 It is false; the latest flip time is f. now The previous flip time was f. Last All are initialized to 0;

[0030] When traversing the data, execute steps (1)-(2) in sequence:

[0031] (1) Perform steps (1.1), (1.2), or (1.3):

[0032] (1.1) If TJ 翻转大到小 True:

[0033] Flip the start sign F 翻转开始 Set to true. Transform into Unflipped number N未翻转 Set to 0;

[0034] (1.1.1) If the flip time flag F 翻转时刻 If it is false, then use f. now Assignment f Last Then update f now f (i+1) Flip the time marker F 翻转时刻 Set to true;

[0035] (1.2) If TJ 翻转小到大 True:

[0036] Flip the start sign F 翻转开始 Set to false, unflipped number N 未翻转 Set to 0;

[0037] (1.3) Other situations:

[0038] If the start flag F is flipped 翻转开始 True: Transform into Unflipped number N 未翻转 Increment the count by 1;

[0039] (2) Perform steps (2.1), (2.2), or (2.3):

[0040] (2.1) If N 未翻转 =T 未翻转 :

[0041] (2.1.1) Number of segments N 段数 Increment the counter by 1, and flip the start flag F. 翻转开始 Set to false, unflipped number N 未翻转 Set to 0, flip time flag F 翻转时刻 Set to false, and set the current time and the previous T. 未翻转 The phase difference data between each time point is recovered, i.e., the phase difference is subtracted by 2π.

[0042] (2.1.2) If N 段数 ≤T 段数 And N 段数 ≥2: Based on the latest flip time f now and the previous flip time f Last Calculate the latest phase difference value Therefore, the phase difference value of the corrected data can be calculated.

[0043]

[0044] (2.2) If the data has been traversed and the start flag F has been toggled 翻转开始 True:

[0045] (2.2.1) Number of segments N 段数 Increment the count by 1 if N 段数 ≤T 段数 And N 段数 ≥2: Use f now Assignment f Last Then update f now f (i+1) Calculate the latest phase difference value Therefore, the phase difference value of the corrected data can be calculated.

[0046]

[0047] (2.3) Other situations:

[0048] Continue iterating through the data.

[0049] As a preferred technical solution, step S22 includes the following steps:

[0050] S221, Determine whether segmentation is required for this correction:

[0051] For the estimated phase difference value Assume that the phase difference range of (0, 2π) includes at least T points. 点数 That's why segmentation is needed, T 点数 If we define the set threshold for the number of points, then:

[0052] 1) If Then there is no need to execute the segmentation criterion, and this correction is considered as a segment of correction data;

[0053] 2) If Then the segmentation criteria need to be executed, and steps S223 to S224 are required to segment the correction data;

[0054] S222, Data Filtering: Remove invalid correction data;

[0055] S223, Segmentation: Based on estimated phase difference value And to filter valid correction data and determine the range of a certain correction data segment:

[0056] Assuming we have current data Valid data from the previous shot And T 翻转偏差 Indicates the tolerable flip deviation threshold:

[0057] 1) If If the result is true, then the termination frequency of this data segment can be determined to be f. nowAnd the starting frequency of the next data segment is f now ;

[0058] 2) In other cases, the segment is undetermined, and the data traversal continues;

[0059] S224, Determine the feasibility of intra-segment correction data: Determine whether the correction data collection for this segment is feasible;

[0060] S225, Polynomial Fitting: Using Polynomials To fit the correction data, the specific method is as follows:

[0061] y represents phase difference correction data, x represents frequency data, and the fitting coefficient is a. n n∈[0,p], p≥1, p represents the highest power of x, and the number of valid correction data in this segment is N. 段内有效 and correction data Solving using the least squares method, we obtain:

[0062] a=(X T X) -1 X T Y(5.2)

[0063] Where a represents the coefficient matrix, Y represents the phase difference matrix, and X represents the frequency matrix.

[0064]

[0065]

[0066] Based on the number of valid correction data N 段内有效 Using N p Fitting with a polynomial of degree N or a polynomial of degree 1, where N p >1, as detailed below:

[0067] 1) If [N] 段内有效 >(N p +1)], then use N n Fitting with a polynomial of degree 1;

[0068] 2) If (N) 段内有效 ≥2)∩[N 段内有效 ≤(N p If +1)], then a first-order polynomial is used for fitting.

[0069] As a preferred technical solution, the specific method for removing invalid correction data in step S222 is as follows:

[0070] During calibration, the true frequency of the signal source is known to be f. r The total number of array elements is N 阵元The actual frequency obtained during calibration is f. a The amplitude of array element i is PA i The judgment criteria are as follows:

[0071] 1) If The data is valid.

[0072] 2) In other cases, the data is invalid;

[0073] Among them, T f T represents the threshold value for the difference between the actual frequency and the true frequency. PA_1 T represents the threshold value for the amplitude of a single array element. PA_2 The threshold representing the difference in amplitude between two array elements.

[0074] As a preferred technical solution, in step S224, the method for determining whether the data collected for this correction segment is feasible is as follows:

[0075] The number of invalid correction data in this segment is N. 段内无效 The number of valid correction data is N 段内有效 The proportion of invalid correction data is T 该段有效 The threshold used to determine whether the corrected data segment is valid;

[0076] 1) If If true, then the data collected in this correction segment meets the polynomial fitting requirements and is valid and feasible, proceed to step S225;

[0077] 2) If Therefore, the data collected in this correction segment does not meet the requirements and is invalid, and it is not feasible to perform polynomial fitting.

[0078] 3) If (N) 段内有效 If the number of data points is less than or equal to 1, then the data in this segment is too small to meet the requirements, but it is valid and not feasible to perform polynomial fitting.

[0079] As a preferred technical solution, in S23, the method for determining whether the current correction meets the requirements is as follows:

[0080] The number of invalid correction data points N in this correction data set 筛选无效 The number of effective correction data N 筛选有效 The effective correction data ratio is T 本次有效 The threshold for determining whether the corrected data is valid; in addition, the number of invalid segments is N. 无效段数 The number of effective segments is N 有效段数 The proportion of effective segments is T 段数有效 A threshold representing the proportion of valid segments;

[0081] 1) If If true, proceed to step S3;

[0082] 2) In other cases, if the data collected for this calibration does not meet the requirements, the reported calibration data is invalid, and the calibration process is terminated.

[0083] As a preferred technical solution, in step S4, the subsequent processing platform uses the following formula to reconstruct phase difference correction data at any frequency:

[0084]

[0085] in, Let f represent the phase difference and f represent the frequency matrix. This represents the starting frequency of the i-th segment of correction data. Let f represent the termination frequency of the i-th segment of correction data, where f = (1 / f) 1 f 2 …f p ) 1×(p+1) This compensates for phase inconsistencies between channels.

[0086] A phase correction system based on piecewise polynomial fitting, used to implement the aforementioned phase correction method based on piecewise polynomial fitting, includes the following modules connected in sequence:

[0087] Data acquisition module: used to acquire phase difference correction data at each frequency point;

[0088] Data processing module: used to process the acquired phase difference correction data based on piecewise polynomial fitting;

[0089] Parameter distribution module: Used to distribute the processed phase difference correction data to the subsequent processing platform;

[0090] Parameter application module: Used to reconstruct phase difference correction data of arbitrary frequencies using the phase difference correction data processed by the subsequent processing platform.

[0091] Compared with the prior art, the present invention has the following advantages:

[0092] (1) It has the ability to judge data validity and correction confidence: By filtering by amplitude and frequency, invalid data can be filtered out, improving the accuracy of correction data, thereby enhancing the stability and robustness of correction; In addition, by statistically analyzing the number of valid and invalid points, the confidence of correction can be obtained.

[0093] (2) Improved correction accuracy: By adopting the piecewise polynomial fitting method, the relationship between phase difference correction data and frequency can be better modeled, which is closer to the actual situation of the system and has universality;

[0094] (3) Reduced calibration time and storage space: In the entire calibration method, it is time-consuming to collect phase difference calibration data at each frequency point. In this invention, by using piecewise polynomials to fit the calibration data, the number of frequency points collected can be reduced, thereby reducing calibration time. In the traditional calibration method, the calibration data (frequency, phase difference) needs to be sent to the subsequent processing platform. However, in the existing method, it is only necessary to send the polynomial coefficients and their corresponding frequency range to the processing platform and apply them, thus reducing the storage space occupied. Attached Figure Description

[0095] Figure 1 A schematic diagram of an ideal model of a two-element antenna phase interferometer;

[0096] Figure 2 This is a schematic diagram of a traditional phase correction method;

[0097] Figure 3 Phase difference correction data / frequency relationship diagram (before transformation);

[0098] Figure 4 This is a schematic diagram of the phase correction method based on piecewise polynomial fitting of the present invention;

[0099] Figure 5 This is a data processing flowchart of the present invention;

[0100] Figure 6 Phase difference correction data / frequency relationship diagram (after transformation). Detailed Implementation

[0101] The present invention will be further described in detail below with reference to the embodiments and accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0102] Example 1

[0103] like Figures 1 to 6 As shown, the purpose of this invention is to propose a phase correction method that, without affecting the use of the broadband direction finding system of the phase interferometer, models the relationship between phase difference correction data and frequency points based on a limited amount of correction data through piecewise polynomial fitting, thereby better characterizing characteristics such as segmentation, fluctuation, and glitches; removes abnormal data through data filtering and provides the confidence level of the correction; calculates correction data at any frequency point through piecewise polynomial fitting, reducing correction time; and reduces storage usage by distributing multiple polynomial parameters instead of distributing correction data at multiple frequency points.

[0104] This invention proposes a phase correction method based on piecewise polynomial fitting, the architecture of which is as follows: Figure 4As shown. The phase difference correction data acquisition is consistent with traditional methods, both using a fixed frequency interval f. 间隔 This section focuses on data processing, parameter distribution, and parameter application for acquiring phase difference correction data.

[0105] 5.1 Data Processing:

[0106] like Figure 3 As shown, since the phase output value range of the signal phase detector is (0, 2π), when the signal frequency changes continuously, it may exceed the phase detection range, causing the phase difference data to flip. This results in a segmented relationship between the phase difference correction data and the frequency. This invention is based on the number of phase difference data flips, N. 翻转 The correction data is divided into (N) 翻转 +1) paragraph, based on this Figure 3 There are three segments of correction data. Because the phase difference data near the flipping point fluctuates too much, directly using the initial correction data for polynomial fitting yields poor results and requires a high order (high complexity). Therefore, a data processing flow based on piecewise polynomial fitting was designed, such as... Figure 5 As shown, the specific implementation details are as follows.

[0107] 5.1.1 Phase difference estimation:

[0108] according to We can obtain:

[0109]

[0110] Where f2 > f1 (unit: 1MHz), These are the phase difference data at frequency f2. This is the phase difference data at frequency f1; theoretically, when l is constant, the same frequency interval has the same phase difference value. To estimate f... 间隔 The phase difference value needs to be determined by identifying the flip point, which is accompanied by a transformation of the phase difference data. The relationship between the transformed phase difference correction data and frequency is as follows: Figure 6 As shown (the relationship before transformation is as follows) Figure 3 (As shown).

[0111] This invention assumes that the number of points in each segment of correction data is at least T. 翻转点数 That is, the maximum phase difference between two adjacent frequency points is Only then can we determine whether the data has been flipped. Assuming we have the current data... Data with the next shot and will Abbreviated as TJ 翻转大到小 , Abbreviated as TJ 翻转小到大In addition, there is a threshold T for the number of unflipped items. 未翻转 Segment count threshold T 段数 The detailed process is as follows:

[0112] Assuming the phase difference of the data in this correction is 1 / 3 And initialized to 2π; unflipped number N 未翻转 Set to 0; segment number N 段数 Set to 0; toggle the start flag F 翻转开始 Set to false, flip time marker F 翻转时刻 It is false; the latest flip time is f. now The previous flip time was f. Last All are initialized to 0;

[0113] Traversing the data: (1)

[0115] (1.1) If TJ 翻转大到小 True:

[0116] Flip the start sign F 翻转开始 Set to true. Transform into Unflipped number N 未翻转 Set to 0;

[0117] (1.1.1) If the flip time flag F 翻转时刻 If it is false, then use f. now Assignment f Last Then update f now f (i+1) Flip the time marker F 翻转时刻 Set to true;

[0118] (1.2) If TJ 翻转小到大 True:

[0119] Flip the start sign F 翻转开始 Set to false, unflipped number N 未翻转 Set to 0;

[0120] (1.3) Other cases (i.e., when the flipping condition is not met):

[0121] If the start flag F is flipped 翻转开始 True: Transform into Unflipped number N 未翻转 Increment the count by 1; (2)

[0123] (2.1) If N 未翻转 =T 未翻转 :

[0124] (2.1.1) Number of segments N段数 Increment the counter by 1, and flip the start flag F. 翻转开始 Set to false, unflipped number N 未翻转 Set to 0, flip time flag F 翻转时刻 Set to false, and set the current time and the previous T. 未翻转 The phase difference data between each time point is recovered, i.e., the phase difference is subtracted by 2π.

[0125] (2.1.2) If N 段数 ≤T 段数 And N 段数 ≥2: Based on the latest flip time f now and the previous flip time f Last Calculate the latest phase difference value Therefore, the phase difference value of the corrected data can be calculated.

[0126]

[0127] (2.2) If the data has been traversed and the start flag F has been toggled 翻转开始 True:

[0128] (2.2.1) Number of segments N 段数 Increment the count by 1 if N 段数 ≤T 段数 And N 段数 ≥2: Use f now Assignment f Last Then update f now f (i+1) Calculate the latest phase difference value Therefore, the phase difference value of the corrected data can be calculated.

[0129] (2.3) Other cases (i.e., the unflipped count value N) 未翻转 The set threshold T was not reached. 未翻转 hour):

[0130] Continue iterating through the data.

[0131] 5.1.2 Piecewise Polynomial Fitting:

[0132] The piecewise polynomial fitting process is based on the phase difference correction data transformed in step 5.1.1, and the specific implementation details are as follows.

[0133] 5.1.2.1 Determine whether segmentation is required for this correction:

[0134] After step 5.1.1, the estimated phase difference can be obtained. And assume that the phase difference range of (0, 2π) includes at least T points. 点数 If segmentation is necessary, then:

[0135] 1) If Then there is no need to execute the segmentation criteria (5.1.2.3, 5.1.2.4), and this correction can be considered as a segment of correction data;

[0136] 2) If Then the segmentation criteria (5.1.2.3, 5.1.2.4) need to be applied to segment the correction data.

[0137] 5.1.2.2 Data Filtering:

[0138] The collected correction data may be invalid for various reasons. Using it will directly affect the final calculated phase difference correction data, thus impacting the correction accuracy. Therefore, data screening is necessary. During correction, the system knows the true frequency of the signal source is f. r The total number of array elements is N 阵元 The actual frequency obtained during calibration is f. a The amplitude of array element i is PA i The judgment criteria are as follows:

[0139] 1) If The data is valid.

[0140] 2) In other cases, the data is invalid;

[0141] Where T f The threshold value representing the difference between the actual frequency and the true frequency; T PA_1 T represents the threshold value for the amplitude of a single array element. PA_2 The threshold representing the difference in amplitude between two array elements.

[0142] 5.1.2.3 Segmentation:

[0143] Based on the estimated phase difference This process, including filtering valid correction data, can determine the range of a given correction data segment. Assuming we have current data... Valid data from the previous shot And T 翻转偏差 Indicates the tolerable flip deviation threshold:

[0144] 1) If If the result is true, then the termination frequency of this data segment can be determined to be f. now And the starting frequency of the next data segment is f now ;

[0145] 2) In other cases, the segment is undetermined, and the data continues to be traversed.

[0146] 5.1.2.4 Determining the feasibility of intra-segment correction data:

[0147] After the aforementioned steps, it can be determined that the number of invalid correction data in this segment is N. 段内无效 The number of valid correction data is N 段内有效 The proportion of invalid correction data is T 该段有效 The threshold used to determine whether the corrected data segment is valid;

[0148] 1) If If true, then the data collected in this correction phase meets the polynomial fitting requirements and is valid and feasible, so proceed to the next step;

[0149] 2) If Therefore, the data collected in this correction segment does not meet the requirements and is invalid, and it is not feasible to perform polynomial fitting.

[0150] 3) If (N) 段内有效 If the number of data points is less than 1, then the data in this segment is too small, does not meet the requirements but is valid, and is not feasible.

[0151] 5.1.2.5 Polynomial Fitting:

[0152] Polynomial fitting uses polynomials To fit and correct the data (frequency points, phase differences). In this invention, y represents the phase difference correction data, x represents the frequency point data, and the fitting coefficient is a. n , n∈[0,p]. After the above steps, the number N of valid correction data in this segment can be obtained. 段内有效 and correction data Solving using the least squares method, we obtain:

[0153] a=(X T X) -1 X T Y(5.2) where the matrix

[0154] This invention is based on the number of valid correction data N 段内有效 Using N p (N p >1) Fitting with a polynomial of degree one or first degree, as follows:

[0155] 1) If [N] 段内有效 >(N p +1)], then use N p Fitting is performed using a polynomial of degree n.

[0156] 2) If (N)段内有效 ≥2)∩[N 段内有效 ≤(N p If +1)], then a first-order polynomial is used for fitting, i.e., linear fitting.

[0157] 5.1.3 Determine whether the current correction meets the requirements:

[0158] After step 5.1.2, we know that N is the number of invalid correction data in this correction data. 筛选无效 The number of effective correction data N 筛选有效 The effective correction data ratio is T 本次有效 The threshold for determining whether the corrected data is valid; in addition, the number of invalid segments is N. 无效段数 The number of effective segments is N 有效段数 The proportion of effective segments is T 段数有效 A threshold representing the proportion of valid segments;

[0159] 1) If If true, proceed to the next step;

[0160] 2) In other cases, if the data collected for this calibration does not meet the requirements, the reported calibration data is invalid, and the process ends (i.e., this calibration is invalid and calibration is exited).

[0161] 5.2 Parameter Issuance and Application:

[0162] After step 5.1, we know the number of effective segments N. 有效段数 And the corresponding parameters for each segment (starting frequency, ending frequency, parameters of the p-th (p≥1) degree polynomial), i.e. Due to the judgment in step 5.1.2.4, there may be cases of frequency discontinuity. Without loss of generality, the present invention considers adjusting the frequency range to make it continuous, as shown in the following example:

[0163] Assuming the correction frequency range is [2000MHz, 3000MHz], two correction data segments are obtained after processing, with parameters [2040MHz, 2480MHz, a0] and [2600MHz, 2980MHz, a1] respectively; after adjustment, the parameters are [2000MHz, 2600MHz, a0] and [2600MHz, 3000MHz, a1] respectively.

[0164] In summary, the adjusted parameters are as follows: The adjusted parameters are then sent to the subsequent processing platform, which can then apply the following formula:

[0165]

[0166] Where f = (1 f 1 f 2 … f p ) 1×(p+1) This allows for the reconstruction of phase difference correction data at arbitrary frequencies, thereby compensating for phase inconsistencies between channels.

[0167] Example 2

[0168] like Figures 1 to 6 As shown, as a further optimization of Embodiment 1, this embodiment also includes the following technical features based on Embodiment 1:

[0169] A typical 4-baseline phase interferometer direction finding system is used as an example of the present invention, but the implementation of the present invention is not limited to this example. Assume that the four channels of the 4-baseline phase interferometer direction finding system are A, B, C, and D, and the correction data between channels A and B are shown in Table 1.

[0170] Table 1. Calibration data between channels A and B

[0171]

[0172]

[0173] The meanings of the symbols in the table are as follows:

[0174] Phi AB Phase difference between antenna elements A and B

[0175] Pa A The amplitude of antenna element A

[0176] Pa B The amplitude of antenna element B

[0177] Pa C The amplitude of antenna array element C

[0178] Pa D The amplitude of antenna element D

[0179] Phi′ AB Phase difference between A and B after transformation

[0180] Examples of the present invention are based on frequency interval f 间隔 =Acquire and correct data at 60MHz, and set the threshold T for the number of points in each segment to determine whether the data is flipped. 翻转点数 The threshold for unflipped counts is 10. 未翻转 The threshold for segment counting is 5. 段数 The threshold T for determining whether segmentation is required for this correction is set to 5. 点数The value is 15. After step 5.1.1, we have:

[0181] 1) The segmentation requirement in step 5.1.2.1 is met, as shown below.

[0182]

[0183] 2) Data transformation: In this example, the correction data does not require transformation;

[0184] In this invention, a threshold T is set for the difference between the actual frequency and the true frequency. f For 3MHz, the single correction amplitude threshold T PA_1 The correction amplitude difference threshold T is 50. PA_2 The tolerable flip deviation threshold T is 30. 翻转偏差 Set the threshold T to 1; this sets the threshold for determining whether the corrected data segment is valid. 该段有效 It is 20%. After piecewise polynomial fitting in step 5.1.2, we have:

[0185] 1) The actual frequency of the calibrated frequency point of 10920MHz is 10930MHz, and the difference of 10MHz exceeds the threshold T. f Data that does not meet the frequency filtering requirements is invalid.

[0186] 2) At the calibration frequency of 12840MHz, the amplitude difference between antenna element A and element D exceeds the threshold T. PA_2 The data does not meet the amplitude filtering requirements and is therefore invalid.

[0187] 3) There are three correction data segments: [start frequency, end frequency (excluding this frequency point), N] 段内无效 N 段内有效 The feasibility assessment results for each correction data segment are as follows: [10680MHz, 11040MHz, 1, 5], [11040MHz, 12960MHz, 1, 31], and [12960MHz, 12960MHz, 0, 1]. If true, the requirement is met and the condition is valid; (31>1) is true, which satisfies the requirement and is valid; (1≤1) is true, which does not satisfy the requirement but is still valid.

[0188] 1) Two sets of calibration data that meet the requirements are shown in Table 2:

[0189] Table 2 shows the two correction data segments that meet the requirements.

[0190]

[0191]

[0192] Set to use N p =Fitting with a 5th-degree polynomial or a first-degree polynomial, we have: the first segment of data (N 段内有效 ≥2)∩[N 段内有效 ≤(N p +1)=(5>2)∩(5≤(5+1)) is true, so a first-order polynomial should be used for fitting; [N 段内有效 >(N p If [+1)] = [31>(5+1)] is true, then N should be used. p =5th degree polynomial fitting; according to equation (5.2), use f a and Phi A ′ B The fitting parameters for the two polynomial segments can be calculated as follows: and

[0193] Set a threshold T to determine whether the current correction is effective. 本次有效 The threshold T for the proportion of effective segments is 80%. 段数有效 It is 80%, and there are: It can be concluded that the calibration meets the requirements;

[0194] After adjusting the two sets of parameters obtained from the solution, they are sent to the subsequent processing platform. The platform can then use equation (5.3) to obtain the piecewise polynomial in this example, which is:

[0195]

[0196] According to the above formula, the processing platform can recover the phase difference correction data at any frequency point, thereby compensating for the phase deviation. In this example, the phase difference data of the frequency points between 11000MHz and 11300MHz are reconstructed in 20MHz steps, as shown in Table 3.

[0197] Table 3. Phase difference data for reconstructing frequencies between 11000MHz and 11300MHz using a 20MHz step.

[0198]

[0199]

[0200] As shown in Table 3, the phase difference at any frequency point can be recovered after processing the correction data using this method. Increasing the frequency step significantly improves the correction rate. Furthermore, storing multiple parameters (polynomial fitting parameters and frequency range) instead of storing the correction data reduces memory usage. This example demonstrates that the present invention meets the correction data requirements of digital receivers.

[0201] As described above, the present invention can be implemented well.

[0202] All features disclosed in all embodiments of this specification, or steps in all methods or processes implied in the disclosure, may be combined and / or extended or replaced in any way, except for mutually exclusive features and / or steps.

[0203] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Based on the technical essence of the present invention, any simple modifications, equivalent substitutions, and improvements made to the above embodiments within the spirit and principles of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A phase correction method based on piecewise polynomial fitting, characterized in that, Based on the correction data, the relationship between phase difference correction data and frequency points is modeled using a piecewise polynomial fitting method, and the correction data at any frequency point is calculated, including the following steps: S1, Data Acquisition: Acquire phase difference correction data at each frequency point; S2, Data Processing: The acquired phase difference correction data is processed based on piecewise polynomial fitting; S3, Parameter Sending: Sends the processed phase difference correction data to the subsequent processing platform; S4, Parameter Application: The subsequent processing platform applies the processed phase difference correction data to reconstruct phase difference correction data at arbitrary frequencies; Step S2 includes the following steps: S21, Phase difference estimation: Perform phase difference estimation to obtain the transformed phase difference correction data; S22, Piecewise polynomial fitting: Piecewise polynomial fitting is performed based on the phase difference correction data transformed in step S21. S23, determine whether the current correction meets the requirements; Step S22 includes the following steps: S221, Determine whether segmentation is required for this correction: For the estimated phase difference value Assuming a segment The phase difference range includes at least 10 points. That's why it needs to be segmented. If we define the set threshold for the number of points, then: 1) If If the segmentation criterion is not required, the current correction is considered as a segment of correction data. 2) If If the segmentation criteria are not met, steps S223 to S224 are required to segment the correction data. S222, Data Filtering: Remove invalid correction data; S223, Segmentation: Based on estimated phase difference value And to filter valid correction data and determine the range of a certain correction data segment: Assuming we have current data Valid data from the previous shot ,and Indicates the tolerable flip deviation threshold: 1) If If true, then the termination frequency of this data segment can be determined as... And the starting frequency of the next data segment is ; 2) In other cases, the segment is undetermined, and the data traversal continues; S224, Determine the feasibility of intra-segment correction data: Determine whether the correction data collection for this segment is feasible; S225, Polynomial Fitting: Using Polynomials To fit and correct the data, the specific method is as follows: , The fitting coefficient is , , express The highest power, the number of valid correction data in this segment. Solving using the least squares method, we obtain: (5.2) in, Represents the coefficient matrix. Represents the phase difference matrix. Represents the frequency matrix, , , ; Based on the number of valid correction data ,use Fitting is performed using either a polynomial of degree 1 or a first-degree polynomial, where... The details are as follows: 1) If [N 段内有效 >(N p +1)], then use Fitting with a polynomial of degree 1; 2) If (N) 段内有效 ≥2)∩[N 段内有效 ≤(N p If +1)], then a first-order polynomial is used for fitting.

2. The phase correction method based on piecewise polynomial fitting according to claim 1, characterized in that, In step S21, it is assumed that the number of points in each segment of correction data is at least [number missing]. That is, the maximum phase difference between two adjacent frequency points is Only then can we determine whether the data has been flipped; Assuming we have current data and , Represents frequency Phase difference data at time, and ; In addition, there is a threshold for counting the number of unflipped items. The detailed process of estimating the phase difference is as follows: Assuming the phase difference of the data in this correction is 1 / 3 and initialized to Flip the start flag Set to false, flip the time marker It is false; the latest flip time is The previous flip time was All are initialized to 0; When traversing the data, execute steps (1)-(2) in sequence: (1) Perform steps (1.1), (1.2), or (1.3): (1.1) If True: Flip the start sign Set to true. Unflipped number Set to 0; (1.1.1) If the time flag is flipped If it is false, then use Assignment Then update for Flip the time marker Set to true; (1.2) If True: Flip the start sign Set to false, number not flipped Set to 0; (1.3) Other situations: If you flip the start flag True: Unflipped number Increment the count by 1; (2) Perform steps (2.1), (2.2), or (2.3): (2.1) If : (2.1.1) Number of segments Increment the counter by 1, and flip the start flag. Set to false, number not flipped Set to 0, toggle the time stamp. Set to false, and set the current time and the previous time... The phase difference data between each time point is used to recover the data, that is, the phase difference is subtracted from the phase difference. ; (2.1.2) If According to the latest flip time and the previous flip moment Calculate the latest phase difference value ; (2.2) If the data has been traversed and the start flag has been flipped True: (2.2.1) Number of segments Increment the count by 1 if :use Assignment Then update for Calculate the latest phase difference value ; (2.3) Other situations: Continue iterating through the data.

3. The phase correction method based on piecewise polynomial fitting according to claim 1, characterized in that, In step S222, the specific method for removing invalid correction data is as follows: During calibration, the true frequency of the signal source is known to be... The total number of array elements is The actual frequency obtained during calibration is Array element The amplitude is The judgment criteria are as follows: 1) If If so, the data is valid; 2) In other cases, the data is invalid; in, The threshold representing the difference between the actual frequency and the true frequency. The threshold representing the amplitude of a single array element. The threshold representing the difference in amplitude between two array elements.

4. The phase correction method based on piecewise polynomial fitting according to claim 1, characterized in that, In step S224, the method for determining whether the data collected for this correction segment is feasible is as follows: The number of invalid correction data in this segment is The number of effective correction data is The proportion of invalid correction data is , The threshold used to determine whether the corrected data segment is valid; 1) If If true, then the data collected in this correction segment meets the polynomial fitting requirements and is valid and feasible, proceed to step S225; 2) If If the data collected in this section does not meet the requirements and is invalid, then it is not feasible to perform polynomial fitting. 3) If If the number of data points in that segment is too small, it does not meet the requirements but is valid, and therefore is not feasible for polynomial fitting.

5. The phase correction method based on piecewise polynomial fitting according to claim 4, characterized in that, In S23, the method for determining whether the current correction meets the requirements is as follows: Number of invalid correction data in this correction data Effective number of corrected data The effective correction data ratio is , The threshold for determining whether the corrected data is valid; in addition, the number of invalid segments is obtained. The number of valid segments is The proportion of effective segments is , A threshold representing the proportion of valid segments; 1) If If true, proceed to step S3; 2) In other cases, if the data collected for this calibration does not meet the requirements, the reported calibration data is invalid, and the calibration process is terminated.

6. A phase correction method based on piecewise polynomial fitting according to any one of claims 1 to 5, characterized in that, In step S4, the subsequent processing platform uses the following formula to reconstruct phase difference correction data at any frequency: (5.3) in, Indicates phase difference, Represents the frequency matrix, This represents the starting frequency of the i-th segment of correction data. This indicates the termination frequency of the i-th segment of correction data. This compensates for phase inconsistencies between channels.

7. A phase correction system based on piecewise polynomial fitting, characterized in that, A phase correction method based on piecewise polynomial fitting as described in any one of claims 1 to 6 comprises the following modules connected in sequence: Data acquisition module: used to acquire phase difference correction data at each frequency point; Data processing module: used to process the acquired phase difference correction data based on piecewise polynomial fitting; Parameter distribution module: Used to distribute the processed phase difference correction data to the subsequent processing platform; Parameter application module: Used to reconstruct phase difference correction data of arbitrary frequencies using the phase difference correction data processed by the subsequent processing platform.

Images