Method and system for evaluating the precision of a ratio measurement of a plurality of one-dimensional frequency energy density spectra

Abstract

The application discloses a kind of multi-group one-dimensional frequency energy density spectrum ratio measurement precision evaluation method and system, belong to marine observation technical field, comprising: the training data sample of establishing including M group synchronous observation sea wave one-dimensional frequency energy density spectrum sequence;21 similarity evaluation value sequences are constructed;21 similarity evaluation value sequences are sorted according to the order of evaluation value from big to small, and at least two dimensions in sequence segmented average, overall average, fluctuation amplitude, fluctuation rate, distribution concentration and dispersion degree are analyzed, to determine the comprehensive weight coefficient of 21 similarity evaluation values;The one-dimensional frequency energy density spectrum sequence of two synchronous observations to be evaluated is obtained, and the corresponding 21 similarity evaluation values thereof are calculated;The final similarity evaluation value NRD is calculated, to evaluate the ratio measurement precision of two to be evaluated sequences.The application constructs a multidimensional evaluation system, can comprehensively reflect the similarity of two sea wave spectrum sequences.

Description

Technical Field

[0001] This invention belongs to the field of marine observation technology, and in particular relates to a method and system for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra. Background Technology

[0002] One-dimensional frequency energy density spectrum of ocean waves can accurately describe the distribution of energy of the constituent waves at different frequencies. It is a key basic data for ocean wave research, marine engineering design, ocean wave forecasting and marine disaster prevention and mitigation.

[0003] Currently, in the field of ocean observation, there is a lack of a systematic and comprehensive method to evaluate the similarity and consistency of one-dimensional frequency energy density spectrum sequences of two simultaneously observed ocean waves. In practical applications, such as verifying the consistency between simulated and measured spectra of wave spectrum models, comparing the similarity of wave fields in different sea areas or at different times, evaluating the effectiveness of numerical wave model outputs, and performing sea state classification based on wave spectra, similarity evaluation is essential for quantitative judgment. However, existing technologies cannot effectively assess the similarity and consistency of one-dimensional frequency energy density spectrum sequences observed simultaneously in the field by the wave spectrum observation instrument under test and a standard wave spectrum observation instrument.

[0004] Therefore, it is of great significance to develop a comprehensive, accurate, and adaptive method and system for evaluating the accuracy of comparison of multiple sets of one-dimensional frequency energy density spectra in order to overcome the limitations of existing technologies. Summary of the Invention

[0005] The present invention aims to address the problem in the prior art of lacking a method for evaluating the similarity and consistency of two synchronously observed one-dimensional frequency energy density spectrum sequences of ocean waves.

[0006] To achieve the above-mentioned objectives, the first objective of this invention is to provide a method for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra, comprising: Establish training data samples containing M sets of synchronously observed one-dimensional frequency energy density spectrum sequences of ocean waves, where M≥80. Each set of training data samples contains two synchronously observed one-dimensional frequency energy density spectrum sequences, and each sequence contains N one-dimensional frequency energy density spectra, where N≥180. Based on the training data samples, 21 similarity evaluation value sequences are constructed. The 21 similarity evaluation value sequences include: 18 sequences consisting of 9 one-dimensional correlation coefficients and 9 one-dimensional cosine coefficients of the 9 spectral feature value sequences; 2 sequences consisting of the average one-dimensional correlation coefficient and average one-dimensional cosine coefficient of the one-dimensional frequency energy density spectrum sequence; and 1 sequence consisting of the average minimum-to-maximum ratio of the total wave energy sequence. The 21 similarity evaluation value sequences were sorted in descending order of evaluation value, and analyzed from at least two dimensions, including sequence segment average, overall average, fluctuation amplitude, fluctuation rate, and distribution concentration and dispersion, to determine the comprehensive weight coefficient K of the 21 similarity evaluation values. i (i=1,2,...,21); Obtain the one-dimensional frequency-energy density spectrum sequences of two synchronous observations to be evaluated, and calculate their corresponding 21 similarity evaluation values ​​R. i ; Through formula The final similarity rating value (NRD) is calculated to evaluate the comparison accuracy of the two sequences to be evaluated.

[0007] Preferably, the nine spectral characteristic values ​​include: effective wave height, average period, total wave energy, peak frequency, peak energy value, spectral width, frequency spectrum centroid, 0.5 times the peak frequency total energy, and 0.5 times the peak energy frequency bandwidth.

[0008] Preferably, the comprehensive weighting coefficient K for determining the 21 similarity evaluation values ​​is... i include: Calculate the average value of each similarity evaluation value sequence in the first 1 / 4, first 1 / 2, first 3 / 4 and the whole range respectively, and determine its corresponding first weight coefficient, second weight coefficient, third weight coefficient and fourth weight coefficient; Calculate the standard deviation and the ratio of the standard deviation to the mean for each similarity evaluation value sequence, and determine the corresponding fifth and sixth weight coefficients. Calculate the central tendency and dispersion of the evaluation value distribution for each similarity evaluation value sequence, and determine its corresponding seventh and eighth weight coefficients; Through formula K i = k1 First weighting coefficient + k2 Second weighting coefficient + k3 Third weighting coefficient + k4 Fourth weighting coefficient + k5 Fifth weighting coefficient + k6 Sixth weighting coefficient + k7 Seventh weighting coefficient + k8 The eighth weighting coefficient is used to calculate the comprehensive weighting coefficient K. i k1 to k8 are the preset weights for each dimension.

[0009] Preferably, the recommended values ​​for the preset weights k1 to k8 for each dimension are 0.05, 0.10, 0.10, 0.50, 0.05, 0.05, 0.10, and 0.05, respectively.

[0010] Preferably, the similarity evaluation value NRD ranges from 0 to 1. The closer the NRD value is to 1, the more similar the two one-dimensional frequency energy density spectrum sequences are.

[0011] The second objective of this invention is to provide a comparative accuracy evaluation system for multiple sets of one-dimensional frequency energy density spectra, comprising: The sample building module builds training data samples containing M sets of synchronously observed one-dimensional frequency energy density spectrum sequences of ocean waves, where M≥80. Each set of training data samples contains two synchronously observed one-dimensional frequency energy density spectrum sequences, and each sequence contains N one-dimensional frequency energy density spectra, where N≥180. The feature analysis module constructs 21 similarity evaluation value sequences based on the training data samples. The 21 similarity evaluation value sequences include: 18 sequences consisting of 9 one-dimensional correlation coefficients and 9 one-dimensional cosine coefficients of 9 spectral feature value sequences; 2 sequences consisting of the average one-dimensional correlation coefficient and average one-dimensional cosine coefficient of the one-dimensional frequency energy density spectrum sequence; and 1 sequence consisting of the average minimum-maximum ratio of the total wave energy sequence. The weighting module sorts the 21 similarity evaluation value sequences in descending order of evaluation value, and analyzes them from at least two dimensions, including sequence segment average, overall average, fluctuation amplitude, fluctuation rate, and distribution concentration and dispersion, to determine the comprehensive weight coefficient K of the 21 similarity evaluation values. i (i=1,2,...,21); The data acquisition module acquires the one-dimensional frequency-energy density spectrum sequences of the two synchronous observations to be evaluated, and calculates their corresponding 21 similarity evaluation values ​​R. i ; The accuracy evaluation module uses formulas. The final similarity rating value (NRD) is calculated to evaluate the comparison accuracy of the two sequences to be evaluated.

[0012] Preferably, the nine spectral feature values ​​in the feature analysis module include: effective wave height, average period, total wave energy, peak frequency, peak energy value, spectral width, frequency spectrum centroid, 0.5 times the peak frequency total energy, and 0.5 times the peak energy frequency bandwidth.

[0013] Preferably, the weight calculation module is specifically used for: Calculate the average value of each similarity evaluation value sequence in the first 1 / 4, first 1 / 2, first 3 / 4 and the whole range, and determine its corresponding first weight coefficient, second weight coefficient, third weight coefficient and fourth weight coefficient; Calculate the standard deviation and the ratio of the standard deviation to the mean for each similarity evaluation value sequence, and determine its corresponding fifth and sixth weight coefficients; Calculate the central tendency and dispersion of the evaluation value distribution for each similarity evaluation value sequence, and determine its corresponding seventh and eighth weight coefficients; Through formula K i = k1 First weighting coefficient + k2 Second weighting coefficient + k3 Third weighting coefficient + k4 Fourth weighting coefficient + k5 Fifth weighting coefficient + k6 Sixth weighting coefficient + k7 Seventh weighting coefficient + k8 The eighth weighting coefficient is used to calculate the comprehensive weighting coefficient K. i k1 to k8 are the preset weights for each dimension.

[0014] A third objective of this invention is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for evaluating the accuracy of comparison of multiple sets of one-dimensional frequency energy density spectra.

[0015] A fourth objective of this invention is to provide a computer program product, comprising a computer program that, when executed by a processor, implements the aforementioned method for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra.

[0016] The advantages and positive effects of this application are: This invention constructs a multi-dimensional evaluation system from multiple levels, including spectral feature values, overall spectral shape, and energy distribution, and introduces various evaluation indicators such as correlation coefficient, cosine of the included angle, and minimum-maximum ratio, which can comprehensively reflect the similarity of two wave spectral sequences.

[0017] This invention determines the weights of each evaluation index by establishing training data samples and using a multi-dimensional weighted analysis method, avoiding the subjectivity of manually setting weights and making the evaluation results more objective and reliable. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart of the first embodiment of the present invention is shown; Figure 2 A model block diagram of the first embodiment of the present invention is shown; Figure 3 A flowchart of the unknown comprehensive weighting coefficients in the first embodiment of the present invention is shown. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 The first embodiment provides a method for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra, mainly including: Establish training data samples containing M sets of synchronously observed one-dimensional frequency energy density spectrum sequences of ocean waves, where M≥80. Each set of training data samples contains two synchronously observed one-dimensional frequency energy density spectrum sequences, and each sequence contains N one-dimensional frequency energy density spectra, where N≥180. Based on the training data samples, 21 similarity evaluation value sequences are constructed. The 21 similarity evaluation value sequences include: 18 sequences consisting of 9 one-dimensional correlation coefficients and 9 one-dimensional cosine coefficients of the 9 spectral feature value sequences; 2 sequences consisting of the average one-dimensional correlation coefficient and average one-dimensional cosine coefficient of the one-dimensional frequency energy density spectrum sequence; and 1 sequence consisting of the average minimum-to-maximum ratio of the total wave energy sequence. The 21 similarity evaluation value sequences were sorted in descending order of evaluation value, and analyzed from at least two dimensions, including sequence segment average, overall average, fluctuation amplitude, fluctuation rate, and distribution concentration and dispersion, to determine the comprehensive weight coefficient K of the 21 similarity evaluation values. i (i=1,2,...,21); Obtain the one-dimensional frequency-energy density spectrum sequences of two synchronous observations to be evaluated, and calculate their corresponding 21 similarity evaluation values ​​R. i ; Through formula The final similarity rating value (NRD) is calculated to evaluate the comparison accuracy of the two sequences to be evaluated.

[0022] To better understand the technical solution of the present invention, a non-limiting description is provided below: Please see Figure 2 and Figure 3 ; Each wave's one-dimensional frequency energy density spectrum sequence contains The one-dimensional correlation coefficient of the effective wave height sequences of two synchronously observed one-dimensional frequency energy density spectrum sequences of ocean waves is: The one-dimensional cosine coefficient of the included angle of the effective wave height sequence is: The one-dimensional correlation coefficient of the average periodic series is The one-dimensional cosine coefficient of the included angle of the average periodic sequence is The one-dimensional correlation coefficient of the total wave energy sequence is The one-dimensional cosine coefficient of the included angle in the total wave energy sequence is: The one-dimensional correlation coefficient of the spectral peak frequency sequence is The one-dimensional cosine coefficient of the included angle of the spectral peak frequency sequence is The one-dimensional correlation coefficient of the spectral peak energy value sequence is The one-dimensional cosine coefficient of the included angle of the spectral peak energy value sequence is The one-dimensional correlation coefficient of the spectral width sequence is The one-dimensional cosine coefficient of the included angle of the spectral width sequence is The one-dimensional correlation coefficient of the frequency spectrum centroid sequence is The one-dimensional cosine coefficient of the centroid sequence of the frequency spectrum is The one-dimensional correlation coefficient of the total energy sequence at 0.5 times the peak frequency is: The one-dimensional cosine coefficient of the included angle of the total energy sequence at 0.5 times the spectral peak frequency is: The one-dimensional correlation coefficient of the 0.5-fold peak energy frequency bandwidth sequence is The one-dimensional cosine coefficient of the included angle of the 0.5 times peak energy frequency bandwidth sequence is: , The average of the one-dimensional correlation coefficients is , The average value of the one-dimensional cosine coefficients of the included angle is , The average of the minimum-to-maximum ratio of the total energy of each wave is: Similarity evaluation values ​​of multiple sets of one-dimensional frequency energy density spectra (Its value is between 0 and 1) and is obtained from the following: The closer the value is to 1, the more similar the two one-dimensional frequency energy density spectrum sequences are. =0 indicates that they are completely dissimilar =1 indicates perfect similarity:

[0023] in, , , , , , , , , , , , , , , , , , , , , There are 21 unknown comprehensive weight coefficients to be solved.

[0024] Solve , , , , , , , , , , , , , , , , , , , , The process for the 21 unknown comprehensive weighting coefficients is as follows: E1. Establish training data samples for the one-dimensional frequency energy density spectrum sequence of waves, and establish... ( The training data samples consist of ≥80 sets of simultaneously observed one-dimensional frequency energy density spectrum sequences of waves. Each set of training data samples contains two simultaneously observed one-dimensional frequency energy density spectrum sequences, and each sequence contains... ( ≥180) one-dimensional frequency energy density spectra; E2. Analyze the similarity of the spectral eigenvalue sequences, i.e., solve for... The one-dimensional correlation coefficient and one-dimensional cosine coefficient of the spectral eigenvalue sequence of a group of one-dimensional frequency-energy density spectral sequences are used to construct a similarity evaluation value sequence of 18 spectral eigenvalue sequences. Each sequence contains... The similarity evaluation values ​​specifically include: the one-dimensional correlation coefficient evaluation value sequence of the effective wave height sequence is as follows. ( =1,2,3… The one-dimensional cosine coefficient evaluation value sequence of the effective wave height sequence is as follows: ( =1,2,3… The one-dimensional correlation coefficient evaluation value sequence of the average periodic series is as follows: ( =1,2,3… The evaluation value sequence of the one-dimensional cosine coefficient of the mean periodic sequence is as follows: ( =1,2,3… The one-dimensional correlation coefficient evaluation value sequence of the total wave energy sequence is as follows: ( =1,2,3… The one-dimensional cosine coefficient evaluation value sequence of the total wave energy sequence is as follows: ( =1,2,3… The one-dimensional correlation coefficient evaluation value sequence of the spectral peak frequency sequence is as follows: ( =1,2,3… The one-dimensional cosine coefficient evaluation value sequence of the spectral peak frequency sequence is as follows: ( =1,2,3… The one-dimensional correlation coefficient evaluation value sequence of the spectral peak energy value sequence is as follows: ( =1,2,3… The one-dimensional cosine coefficient evaluation value sequence of the spectral peak energy value sequence is as follows: ( =1,2,3… The one-dimensional correlation coefficient evaluation value sequence of the spectral width sequence is as follows: ( =1,2,3… The one-dimensional cosine coefficient evaluation value sequence of the spectral width sequence is as follows: ( =1,2,3… The one-dimensional correlation coefficient evaluation value sequence of the frequency spectrum centroid sequence is as follows: ( =1,2,3… The one-dimensional cosine coefficient evaluation value sequence of the centroid sequence of the frequency spectrum is as follows: ( =1,2,3… The one-dimensional correlation coefficient evaluation value sequence of the total energy sequence with 0.5 times the peak frequency is as follows: ( =1,2,3… The evaluation value sequence of the one-dimensional cosine coefficient of the included angle of the total energy sequence of the 0.5 times peak frequency is as follows: ( =1,2,3… The one-dimensional correlation coefficient evaluation value sequence of the 0.5 times peak energy frequency bandwidth sequence is as follows: ( =1,2,3… The evaluation value sequence of the one-dimensional cosine coefficient of the included angle of the 0.5 times peak energy frequency bandwidth sequence is as follows: ( =1,2,3… ); E3. Analyze the similarity of one-dimensional frequency energy density spectrum sequence data samples, i.e., solve for... The average one-dimensional correlation coefficient and average one-dimensional cosine coefficient of the one-dimensional frequency energy density spectrum sequence are used to construct a similarity evaluation value sequence for two one-dimensional frequency energy density spectrum sequences. Each sequence contains... The similarity evaluation values ​​specifically include: the average one-dimensional correlation coefficient evaluation value sequence of the one-dimensional frequency-energy density spectrum sequence. ( =1,2,3… Each similarity score in the sequence represents a value in a set of one-dimensional frequency-energy density spectrum sequences. The average value of the one-dimensional correlation coefficients; the evaluation value sequence of the average one-dimensional cosine coefficient of the one-dimensional frequency energy density spectrum sequence is as follows: ( =1,2,3… Each similarity score in the sequence represents a value in a set of one-dimensional frequency-energy density spectrum sequences. The average value of the cosine coefficients of the one-dimensional included angle; E4. Analyze the similarity of the total wave energy sequence with the training data samples of the one-dimensional frequency energy density spectrum sequence, i.e., solve the following respectively. The average minimum-maximum ratio of the wave total energy sequence from the one-dimensional frequency energy density spectrum sequence is used to construct a similarity evaluation value sequence for the wave total energy sequence, i.e., the average minimum-maximum ratio evaluation value sequence of the wave total energy sequence is... ( =1,2,3… Each similarity score in the sequence represents a value in a set of wave total energy sequences. The average of the minimum and maximum ratios; E5. The similarity evaluation value sequences of the one-dimensional frequency energy density spectrum are reordered. A total of 21 similarity evaluation value sequences of the one-dimensional frequency energy density spectrum were constructed above, including similarity evaluation value sequences of 18 spectral feature value sequences, 2 similarity evaluation value sequences of the one-dimensional frequency energy density spectrum sequences, and 1 similarity evaluation value sequence of the wave total energy sequence. Each similarity evaluation value sequence contains... The 21 similarity score sequences are then reordered in descending order of score. For the ... ( =1,2,3…21) similarity evaluation value sequences ( =1,2,3… ), sorted into a sequence according to the evaluation values ​​from largest to smallest. ,in ; E6. Analyze the similarity of the first quarter of the one-dimensional frequency-energy density spectrum similarity evaluation value sequence. To further understand the gradient distribution of the similarity evaluation values ​​of each part in the one-dimensional frequency-energy density spectrum similarity evaluation value sequence, and the influence of each part's similarity evaluation value on the overall similarity evaluation value, and thus assign different levels of importance to them; divide each one-dimensional frequency-energy density spectrum similarity evaluation value sequence, which has been sorted from largest to smallest, into four equal parts. Select the first part and analyze its similarity and influence as a factor affecting the overall similarity evaluation; analyze the similarity of the first quarter of the 21 one-dimensional frequency-energy density spectrum similarity evaluation value sequences, that is, calculate the average similarity evaluation value and its corresponding weight coefficient of the first quarter of the 21 sorted one-dimensional frequency-energy density spectrum similarity evaluation value sequences. ( The average similarity score of the first quarter of the sorted one-dimensional frequency energy density spectrum similarity score sequences (=1,2,3…21) is:

[0025] In the formula above Indicates to Rounding up to the nearest integer, the corresponding weighting coefficient is:

[0026] E7. Analyze the similarity of the first half of the one-dimensional frequency-energy density spectrum similarity evaluation value sequence. To further understand the gradient distribution of the similarity evaluation values ​​of each part in the one-dimensional frequency-energy density spectrum similarity evaluation value sequence, and the influence of each part's similarity evaluation value on the overall similarity evaluation value, and thus assign different levels of importance to them; divide each one-dimensional frequency-energy density spectrum similarity evaluation value sequence, which has been sorted from largest to smallest, into two equal parts. Select the first part and analyze its similarity and influence as a factor affecting the overall similarity evaluation; analyze the similarity of the first half of the 21 one-dimensional frequency-energy density spectrum similarity evaluation value sequences, that is, calculate the average similarity evaluation value and its corresponding weight coefficient of the first half of the 21 sorted one-dimensional frequency-energy density spectrum similarity evaluation value sequences. ( The average similarity score of the first half of the sorted one-dimensional frequency energy density spectrum similarity score sequences (=1,2,3…21) is:

[0027] In the formula above Indicates to Rounding up to the nearest integer, the corresponding weighting coefficient is:

[0028] E8. Analyze the similarity of the first 3 / 4 of the one-dimensional frequency energy density spectrum similarity evaluation value sequence. To further understand the gradient distribution of the similarity evaluation values ​​of each part in the one-dimensional frequency energy density spectrum similarity evaluation value sequence, and the influence of each part's similarity evaluation value on the overall similarity evaluation value, and thus assign different levels of importance to them; divide each one-dimensional frequency energy density spectrum similarity evaluation value sequence, which has been sorted from largest to smallest, into 4 equal parts, select the first 3 parts, and analyze their similarity and influence as a factor affecting the overall similarity evaluation; analyze the similarity of the first 3 / 4 of the 21 one-dimensional frequency energy density spectrum similarity evaluation value sequences, that is, solve for the average similarity evaluation value and its corresponding weight coefficient of the first 3 / 4 of the 21 sorted one-dimensional frequency energy density spectrum similarity evaluation value sequences; ( The average similarity score of the first 3 / 4 of the sorted one-dimensional frequency energy density spectrum similarity score sequences (=1,2,3…21) is:

[0029] In the formula above Indicates to Rounding up to the nearest integer, the corresponding weighting coefficient is:

[0030] E9. Analyze the overall similarity of the one-dimensional frequency energy density spectrum similarity evaluation value sequences. Analyze the overall similarity of the 21 one-dimensional frequency energy density spectrum similarity evaluation value sequences, that is, calculate the average similarity evaluation value and its corresponding weight coefficient for each of the 21 sequences, as a factor influencing the overall similarity evaluation; ( The average similarity evaluation value of the sequence of one-dimensional frequency energy density spectrum similarity evaluation values ​​(=1,2,3…21) is:

[0031] The corresponding weighting coefficients are:

[0032] E10. Analyze the fluctuation range of evaluation values ​​in the one-dimensional frequency energy density spectrum similarity evaluation value sequence. Specifically, analyze the fluctuation range of evaluation values ​​in the 21 one-dimensional frequency energy density spectrum similarity evaluation value sequences, i.e., calculate the standard deviation and corresponding weight coefficient of each of the 21 one-dimensional frequency energy density spectrum similarity evaluation value sequences, as a factor affecting the overall similarity evaluation. ( The standard deviation of the sequence of one-dimensional frequency energy density spectrum similarity evaluation values ​​(=1,2,3…21) is:

[0033] The corresponding weighting coefficients are:

[0034] E11. Analyze the rate of change of evaluation values ​​in the one-dimensional frequency energy density spectrum similarity evaluation value sequence. Specifically, analyze the rate of change of evaluation values ​​in the 21 one-dimensional frequency energy density spectrum similarity evaluation value sequences. That is, calculate the ratio of the standard deviation to the average value and its corresponding weighting coefficient for each of the 21 one-dimensional frequency energy density spectrum similarity evaluation value sequences, as a factor influencing the overall similarity evaluation. ( The ratio of the standard deviation to the mean of the sequence of one-dimensional frequency energy density spectrum similarity evaluation values ​​(e.g., =1,2,3…21) is:

[0035] The corresponding weighting coefficients are:

[0036] E12. Analyze the concentration and dispersion of evaluation values ​​in the one-dimensional frequency energy density spectrum similarity evaluation value sequence. Analyze the concentration and dispersion of evaluation values ​​in the 21 one-dimensional frequency energy density spectrum similarity evaluation value sequences as a factor affecting the overall similarity evaluation; ( =1,2,3…21) sequences of one-dimensional frequency energy density spectrum similarity evaluation values ( =1,2,3… The minimum evaluation value in ) The sum of the evaluation values ​​of the sequences is:

[0037] The central tendency of the evaluation value distribution in this sequence is:

[0038] The corresponding weighting coefficients are:

[0039] The degree of dispersion of the evaluation value distribution in this sequence is: The corresponding weighting coefficients are:

[0040] E13. A multi-dimensional comprehensive evaluation of the similarity of one-dimensional frequency-energy density spectrum sequences is performed, specifically by calculating the comprehensive weighting coefficients of the 21 one-dimensional frequency-energy density spectrum similarity evaluation values. ( The comprehensive weighting coefficients for the similarity evaluation values ​​of the one-dimensional frequency energy density spectrum (=1,2,3…21) are:

[0041] in: ,recommend =0.05, =0.10, =0.10, =0.50, =0.05, =0.05, =0.10, =0.05.

[0042] The following example uses multiple sets of one-dimensional frequency energy density spectra observed simultaneously by two sets of wave buoys in a certain sea area as an implementation case to explain in detail the invention. , , , , , , , , , , , , , , , , , , , , Specific implementation methods for 21 unknown comprehensive weighting coefficients.

[0043] E1. Establish training data samples for wave one-dimensional frequency energy density spectrum sequences: 100 sets of one-dimensional frequency energy density spectrum sequences observed simultaneously by two sets of wave buoys in a certain sea area are used as training data samples. Each set of training data samples contains two synchronously observed one-dimensional frequency energy density spectrum sequences, and each sequence contains 200 one-dimensional frequency energy density spectra.

[0044] E2. Analyze the similarity of the spectral eigenvalue sequences: This paper analyzes the similarity of spectral feature value sequences of training data samples of one-dimensional frequency-energy density spectral sequences. Specifically, it calculates the one-dimensional correlation coefficient and one-dimensional cosine coefficient of the spectral feature value sequences of 100 sets of one-dimensional frequency-energy density spectral sequences, constructing 18 similarity evaluation value sequences, each containing 100 similarity evaluation values. As an implementation example, the first value of the one-dimensional correlation coefficient evaluation value sequence of the significant wave height sequence is provided. =0.997521, the first value in the one-dimensional cosine coefficient evaluation value sequence of the effective wave height sequence. =0.999103, the first value of the one-dimensional correlation coefficient evaluation value of the average periodic series. =0.992811, the first value of the one-dimensional cosine coefficient evaluation value sequence of the average periodic sequence. =0.999697, the first value of the one-dimensional correlation coefficient evaluation value sequence of the total wave energy sequence. =0.994716, the first value in the one-dimensional cosine coefficient evaluation value sequence of the total wave energy sequence. =0.996592, the first value of the one-dimensional correlation coefficient evaluation value sequence of the spectral peak frequency sequence. =0.896278, the first value in the one-dimensional cosine coefficient evaluation value sequence of the spectral peak frequency sequence. =0.987518, the first value of the one-dimensional correlation coefficient evaluation value sequence of the spectral peak energy value sequence. =0.980529, the first value in the one-dimensional cosine coefficient evaluation value sequence of the spectral peak energy value sequence. =0.986197, the first value in the one-dimensional correlation coefficient evaluation value sequence of the spectral width sequence. =0.975557, the first value in the one-dimensional cosine coefficient evaluation value sequence of the spectral width sequence. =0.999588, the first value in the one-dimensional correlation coefficient evaluation sequence of the frequency spectrum centroid sequence. =0.990380, the first value in the one-dimensional cosine coefficient evaluation value sequence of the frequency spectrum centroid sequence. =0.999559, the first value of the one-dimensional correlation coefficient evaluation value of the total energy sequence at 0.5 times the peak frequency. =0.976088, the first value of the one-dimensional cosine coefficient evaluation value sequence of the total energy sequence of the spectral peak frequency, 0.5 times the peak frequency. =0.983736, the first value of the one-dimensional correlation coefficient evaluation value of the 0.5 times peak energy frequency bandwidth sequence. =0.640716, the first value in the one-dimensional cosine coefficient evaluation value sequence of the 0.5-fold peak energy frequency bandwidth sequence. =0.949551. The solution process for other evaluation values ​​is similar.

[0045] E3. Analyze the similarity of one-dimensional frequency energy density spectrum sequences: This study analyzes the similarity of one-dimensional frequency-energy density spectral sequence data samples. Specifically, it calculates the average one-dimensional correlation coefficient and average one-dimensional cosine coefficient of the included angle for 100 sets of one-dimensional frequency-energy density spectral sequences, constructing two similarity evaluation value sequences, each containing 100 similarity evaluation values. As an implementation example, the first value of the average one-dimensional correlation coefficient evaluation value sequence for the one-dimensional frequency-energy density spectral sequences is provided. =0.959277, the first value of the evaluation value sequence of the average one-dimensional cosine coefficient of the one-dimensional frequency energy density spectrum sequence. =0.975147. The solution process for other evaluation values ​​is similar.

[0046] E4. Analyze the similarity of the total wave energy sequence: This paper analyzes the similarity of the wave total energy sequence among the training data samples of the one-dimensional frequency energy density spectrum sequence. Specifically, it calculates the average minimum-to-maximum ratio of the wave total energy sequence for each of 100 sets of one-dimensional frequency energy density spectrum sequences, constructing a sequence of evaluation values ​​for the average minimum-to-maximum ratio of the wave total energy sequence. As an implementation example, the i-th value of the evaluation value sequence for the average minimum-to-maximum ratio of the wave total energy sequence is given. =0.896720. The solution process for other evaluation values ​​is similar.

[0047] E5. Reorder the similarity evaluation value sequence of the one-dimensional frequency energy density spectrum: A total of 21 similarity evaluation value sequences for one-dimensional frequency energy density spectra were constructed, including similarity evaluation value sequences for 18 spectral feature value sequences, 2 similarity evaluation value sequences for 2 one-dimensional frequency energy density spectrum sequences, and 1 similarity evaluation value sequence for 1 wave total energy sequence. Each similarity evaluation value sequence contains 100 similarity evaluation values. These 21 similarity evaluation value sequences were then reordered in descending order of evaluation value. The first similarity evaluation value sequence is given as an implementation example. (in (e.g., 1, 2, 3…100), rearranged into a sequence according to the evaluation values ​​in descending order. (in =1,2,3…100), the first value of the sequence. =0.998497, the 100th value =0.997292. The solution process for other evaluation value sequences is similar.

[0048] E6. Analyze the similarity of the first quarter of the one-dimensional frequency-energy density spectrum similarity evaluation value sequence: Analyze the similarity of the first quarter of the 21 one-dimensional frequency-energy density spectrum similarity evaluation value sequences, that is, calculate the average similarity evaluation value and its corresponding weight coefficient for the first quarter of each of the 21 sorted one-dimensional frequency-energy density spectrum similarity evaluation value sequences. As an implementation case, the average similarity evaluation value of the first quarter of the first sorted one-dimensional frequency-energy density spectrum similarity evaluation value sequence is given as follows: =0.998447.

[0049] In the formula above Indicates to Rounding up to the nearest integer, the corresponding weighting coefficient is: =0.048970.

[0050] The solution process for other evaluation value sequences is similar.

[0051] E7. Analyze the similarity of the first half of the one-dimensional frequency energy density spectrum similarity evaluation value sequence: This paper analyzes the similarity of the first half of 21 one-dimensional frequency-energy density spectrum similarity evaluation value sequences, specifically by calculating the average similarity evaluation value and its corresponding weight coefficient for each of the first half of the 21 sorted sequences. As an implementation example, the average similarity evaluation value of the first half of the first sorted one-dimensional frequency-energy density spectrum similarity evaluation value sequence is given as follows: =0.998402.

[0052] In the formula above Indicates to Rounding up to the nearest integer, the corresponding weighting coefficient is: =0.049115.

[0053] The solution process for other evaluation value sequences is similar.

[0054] E8. Analyze the similarity of the first 3 / 4 of the one-dimensional frequency energy density spectrum similarity evaluation value sequence: This study analyzes the similarity of the first three-quarters of 21 one-dimensional frequency-energy density spectrum similarity evaluation value sequences. Specifically, it calculates the average similarity evaluation value and its corresponding weight coefficient for each of the first three-quarters of the 21 sorted sequences. As an implementation example, the average similarity evaluation value for the first three-quarters of the first sorted one-dimensional frequency-energy density spectrum similarity evaluation value sequence is given as follows: =0.998302.

[0055] In the formula above Indicates to Rounding up to the nearest integer, the corresponding weighting coefficient is: =0.049195.

[0056] The solution process for other evaluation value sequences is similar.

[0057] E9. Analyze the overall similarity of the one-dimensional frequency energy density spectrum similarity evaluation value sequence: The overall similarity of 21 one-dimensional frequency-energy density spectrum similarity evaluation value sequences is analyzed. Specifically, the average similarity evaluation value and its corresponding weight coefficient for each of the 21 sequences are calculated, serving as a factor influencing the overall similarity evaluation. As an implementation example, the average similarity evaluation value of the first one-dimensional frequency-energy density spectrum similarity evaluation value sequence is given as follows: =0.998113.

[0058] The corresponding weighting coefficients are: =0.049264.

[0059] The solution process for other evaluation value sequences is similar.

[0060] E10. Analyze the fluctuation range of evaluation values ​​in the one-dimensional frequency energy density spectrum similarity evaluation value sequence: This study analyzes the fluctuation range of the evaluation values ​​in 21 one-dimensional frequency-energy density spectrum similarity evaluation value sequences. Specifically, it calculates the standard deviation and corresponding weight coefficient of each of the 21 sequences, considering them as factors influencing the overall similarity evaluation. As an implementation example, the standard deviation of the first one-dimensional frequency-energy density spectrum similarity evaluation value sequence is given as follows: =0.000372.

[0061] The corresponding weighting coefficients are: =0.004160.

[0062] The solution process for other evaluation value sequences is similar.

[0063] E11. Analyze the fluctuation rate of evaluation values ​​in the one-dimensional frequency energy density spectrum similarity evaluation value sequence: Analyze the fluctuation rate of evaluation values ​​in the 21 one-dimensional frequency energy density spectrum similarity evaluation value sequences, that is, calculate the ratio of the standard deviation to the average value and its corresponding weight coefficient for each of the 21 one-dimensional frequency energy density spectrum similarity evaluation value sequences, as a factor affecting the overall similarity evaluation. As an implementation case, the ratio of the standard deviation to the average value of the first one-dimensional frequency energy density spectrum similarity evaluation value sequence is given as: =0.000373.

[0064] The corresponding weighting coefficients are: =0.003478.

[0065] The solution process for other evaluation value sequences is similar.

[0066] E12. Analysis of the concentration and dispersion of evaluation values ​​in the one-dimensional frequency energy density spectrum similarity evaluation value sequence: Analyze the concentration and dispersion of evaluation values ​​in the 21 one-dimensional frequency energy density spectrum similarity evaluation value sequences as a factor affecting the overall similarity evaluation. The first similarity evaluation value sequence is given as an implementation case. ( The degree of centralization of the evaluation value distribution in the range (=1,2,3…100) is shown below, where and For sequence ( The minimum evaluation value and the sum of all evaluation values ​​in the range (1, 2, 3…100).

[0067] =0.971949, the corresponding weighting coefficient is: =0.049585.

[0068] The degree of dispersion of the evaluation values ​​in this sequence is: =0.028051, the corresponding weighting coefficient is: =0.020060.

[0069] The solution process for other evaluation value sequences is similar.

[0070] E13. Multi-dimensional comprehensive evaluation of the similarity of one-dimensional frequency-energy density spectrum sequences: The comprehensive weighting coefficients for the 21 one-dimensional frequency-energy density spectrum similarity evaluation values ​​are calculated separately. An implementation case is provided. The combined weighting coefficients of the 21 similarity evaluation values ​​are as follows, and recommendations are made. =0.05, =0.10, =0.10, =0.5, =0.05, =0.05, =0.10, =0.05.

[0071] =0.043255; =0.043122; The solution process for other comprehensive weighting coefficients is similar, and the results are as follows.

[0072] =0.043893; =0.045054; =0.043893; =0.043022; =0.062552; =0.049641; =0.046315; =0.044671; =0.051099; =0.047385; =0.045494; =0.046517; =0.045497; =0.044222; =0.070672; =0.045681; =0.047860; =0.046123; =0.044033.

[0073] Based on the already solved , , , , , , , , , , , , , , , , , , , , The model for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra is established using 21 comprehensive weighting coefficients, as shown below. The similarity evaluation values ​​for multiple sets of one-dimensional frequency energy density spectra are also presented. (Its value is between 0 and 1) is obtained from the following.

[0074]

[0075] A second embodiment provides a comparative accuracy evaluation system for multiple sets of one-dimensional frequency energy density spectra, comprising: The sample building module builds training data samples containing M sets of synchronously observed one-dimensional frequency energy density spectrum sequences of ocean waves, where M≥80. Each set of training data samples contains two synchronously observed one-dimensional frequency energy density spectrum sequences, and each sequence contains N one-dimensional frequency energy density spectra, where N≥180. The feature analysis module constructs 21 similarity evaluation value sequences based on the training data samples. The 21 similarity evaluation value sequences include: 18 sequences consisting of 9 one-dimensional correlation coefficients and 9 one-dimensional cosine coefficients of 9 spectral feature value sequences; 2 sequences consisting of the average one-dimensional correlation coefficient and average one-dimensional cosine coefficient of the one-dimensional frequency energy density spectrum sequence; and 1 sequence consisting of the average minimum-maximum ratio of the total wave energy sequence. The weighting module sorts the 21 similarity evaluation value sequences in descending order of evaluation value, and analyzes them from at least two dimensions, including sequence segment average, overall average, fluctuation amplitude, fluctuation rate, and distribution concentration and dispersion, to determine the comprehensive weight coefficient K of the 21 similarity evaluation values. i (i=1,2,...,21); The data acquisition module acquires the one-dimensional frequency-energy density spectrum sequences of the two synchronous observations to be evaluated, and calculates their corresponding 21 similarity evaluation values ​​R. i ; The accuracy evaluation module uses formulas. The final similarity rating value (NRD) is calculated to evaluate the comparison accuracy of the two sequences to be evaluated.

[0076] The nine spectral feature values ​​in the feature analysis module include: significant wave height, average period, total wave energy, peak frequency, peak energy value, spectral width, frequency spectrum centroid, 0.5 times peak frequency total energy, and 0.5 times peak energy frequency bandwidth.

[0077] The weight calculation module is specifically used for: Calculate the average value of each similarity evaluation value sequence in the first 1 / 4, first 1 / 2, first 3 / 4 and the whole range, and determine its corresponding first weight coefficient, second weight coefficient, third weight coefficient and fourth weight coefficient; Calculate the standard deviation and the ratio of the standard deviation to the mean for each similarity evaluation value sequence, and determine its corresponding fifth and sixth weight coefficients; Calculate the central tendency and dispersion of the evaluation value distribution for each similarity evaluation value sequence, and determine its corresponding seventh and eighth weight coefficients; Through formula K i = k1 First weighting coefficient + k2 Second weighting coefficient + k3 Third weighting coefficient + k4 Fourth weighting coefficient + k5 Fifth weighting coefficient + k6 Sixth weighting coefficient + k7 Seventh weighting coefficient + k8 The eighth weighting coefficient is used to calculate the overall weighting coefficient K. i k1 to k8 are the preset weights for each dimension.

[0078] In the third embodiment, a computer-readable storage medium stores a computer program that, when executed by a processor, implements the above-described method for evaluating the accuracy of comparison of multiple sets of one-dimensional frequency energy density spectra.

[0079] Fourth embodiment: A computer program product, including a computer program that, when executed by a processor, implements the above-described method for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra.

[0080] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented, in whole or in part, as a computer program product, the computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line, or wireless (e.g., infrared, wireless, microwave, etc.) means). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0081] The above description is only a preferred embodiment of the present invention. It should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra, characterized in that, include: Establish training data samples containing M sets of synchronously observed one-dimensional frequency energy density spectrum sequences of ocean waves, where M≥80. Each set of training data samples contains two synchronously observed one-dimensional frequency energy density spectrum sequences, and each sequence contains N one-dimensional frequency energy density spectra, where N≥180. Based on the training data samples, 21 similarity evaluation value sequences are constructed. The 21 similarity evaluation value sequences include: 18 sequences consisting of 9 one-dimensional correlation coefficients and 9 one-dimensional cosine coefficients of the 9 spectral feature value sequences; 2 sequences consisting of the average one-dimensional correlation coefficient and average one-dimensional cosine coefficient of the one-dimensional frequency energy density spectrum sequence; and 1 sequence consisting of the average minimum-maximum ratio of the total wave energy sequence. The 21 similarity evaluation value sequences were sorted in descending order of evaluation value, and analyzed from at least two dimensions, including sequence segment average, overall average, fluctuation amplitude, fluctuation rate, and distribution concentration and dispersion, to determine the comprehensive weight coefficient K of the 21 similarity evaluation values. i , i=1,2,...,21; Obtain the one-dimensional frequency-energy density spectrum sequences of two synchronous observations to be evaluated, and calculate their corresponding 21 similarity evaluation values ​​R. i ; Through formula The final similarity rating value (NRD) is calculated to evaluate the comparison accuracy of the two sequences to be evaluated.

2. The method for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra according to claim 1, characterized in that, The nine spectral characteristic values ​​include: effective wave height, average period, total wave energy, peak frequency, peak energy value, spectral width, frequency spectrum centroid, 0.5 times the peak frequency total energy, and 0.5 times the peak energy frequency bandwidth.

3. The method for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra according to claim 1, characterized in that, The comprehensive weighting coefficient K for determining the 21 similarity evaluation values. i include: Calculate the average value of each similarity evaluation value sequence in the first 1 / 4, first 1 / 2, first 3 / 4 and the whole range respectively, and determine its corresponding first weight coefficient, second weight coefficient, third weight coefficient and fourth weight coefficient; Calculate the standard deviation and the ratio of the standard deviation to the mean for each similarity evaluation value sequence, and determine the corresponding fifth and sixth weight coefficients. Calculate the central tendency and dispersion of the evaluation value distribution for each similarity evaluation value sequence, and determine its corresponding seventh and eighth weight coefficients; Through formula K i =k1 First weighting coefficient + k2 Second weighting coefficient + k3 Third weighting coefficient + k4 Fourth weighting coefficient + k5 Fifth weighting coefficient + k6 Sixth weighting coefficient + k7 Seventh weighting coefficient + k8 The eighth weighting coefficient is used to calculate the overall weighting coefficient K. i k1 to k8 are the preset weights for each dimension.

4. The method for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra according to claim 3, characterized in that, The recommended values ​​for the preset weights k1 to k8 for each dimension are 0.05, 0.10, 0.10, 0.50, 0.05, 0.05, 0.10, and 0.05, respectively.

5. The method for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra according to claim 1, characterized in that, The similarity evaluation value NRD ranges from 0 to 1. The closer the NRD value is to 1, the more similar the two one-dimensional frequency energy density spectrum sequences are.

6. A comparative accuracy evaluation system for multiple sets of one-dimensional frequency energy density spectra, characterized in that, include: The sample building module builds training data samples containing M sets of synchronously observed one-dimensional frequency energy density spectrum sequences of ocean waves, where M≥80. Each set of training data samples contains two synchronously observed one-dimensional frequency energy density spectrum sequences, and each sequence contains N one-dimensional frequency energy density spectra, where N≥180. The feature analysis module constructs 21 similarity evaluation value sequences based on the training data samples. The 21 similarity evaluation value sequences include: 18 sequences consisting of 9 one-dimensional correlation coefficients and 9 one-dimensional cosine coefficients of 9 spectral feature value sequences; 2 sequences consisting of the average one-dimensional correlation coefficient and average one-dimensional cosine coefficient of the one-dimensional frequency energy density spectrum sequence; and 1 sequence consisting of the average minimum-maximum ratio of the total wave energy sequence. The weighting module sorts the 21 similarity evaluation value sequences in descending order of evaluation value, and analyzes them from at least two dimensions, including sequence segment average, overall average, fluctuation amplitude, fluctuation rate, and distribution concentration and dispersion, to determine the comprehensive weight coefficient K of the 21 similarity evaluation values. i , i=1,2,...,21; The data acquisition module acquires the one-dimensional frequency-energy density spectrum sequences of the two synchronous observations to be evaluated, and calculates their corresponding 21 similarity evaluation values ​​R. i ; The accuracy evaluation module uses formulas. The final similarity rating value (NRD) is calculated to evaluate the comparison accuracy of the two sequences to be evaluated.

7. The system for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra according to claim 6, characterized in that, The nine spectral feature values ​​in the feature analysis module include: effective wave height, average period, total wave energy, peak frequency, peak energy value, spectral width, frequency spectrum centroid, 0.5 times peak frequency total energy, and 0.5 times peak energy frequency bandwidth.

8. The system for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra according to claim 6, characterized in that, The weight calculation module is specifically used for: Calculate the average value of each similarity evaluation value sequence in the first 1 / 4, first 1 / 2, first 3 / 4 and the whole range, and determine its corresponding first weight coefficient, second weight coefficient, third weight coefficient and fourth weight coefficient; Calculate the standard deviation and the ratio of the standard deviation to the mean for each similarity evaluation value sequence, and determine its corresponding fifth and sixth weight coefficients; Calculate the central tendency and dispersion of the evaluation value distribution for each similarity evaluation value sequence, and determine its corresponding seventh and eighth weight coefficients; Through formula K i = k1 First weighting coefficient + k2 Second weighting coefficient + k3 Third weighting coefficient +k4 Fourth weighting coefficient + k5 Fifth weighting coefficient + k6 Sixth weighting coefficient + k7 Seventh weighting coefficient + k8 The eighth weighting coefficient is used to calculate the overall weighting coefficient K. i k1 to k8 are the preset weights for each dimension.

9. A computer program product, characterized in that, The method includes a computer program that, when executed by a processor, implements the method for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra as described in any one of claims 1-5.

10. A computer-readable storage medium storing a computer program, characterized in that, When executed by the processor, the program implements the method for evaluating the comparative accuracy of multiple sets of one-dimensional frequency energy density spectra as described in any one of claims 1-5.

Images