A ship model resistance test result evaluation method based on time series random truncation

Abstract

The application provides a ship model resistance test result evaluation method based on time series random truncation, and specifically comprises the following steps: S1: collecting a time series TS0 in a ship model resistance experiment to obtain a corresponding resistance-time curve; S2: obtaining a time series TS1 by using a resistance mean value R0 of the time series TS0; S3: obtaining a sample time series TSi by periodically truncating the time series TS1; S4: judging whether the sample time series in S3 is a valid sample time series; S5: performing convergence evaluation on the valid sample time series obtained in S4; and S6: outputting a class A uncertainty UF corresponding to RF. The application not only improves the richness of the sample time series, but also greatly improves the accuracy of the test result of the ship model resistance experiment under the collection of a limited length time series.

Description

Technical Field

[0001] This invention relates to the field of ship hydrodynamic testing, specifically to a method for evaluating the results of ship model resistance tests based on random truncation of time series. Background Technology

[0002] The carbon peak target and carbon neutrality vision are major national strategies. Carbon emissions from the shipping industry are a key concern for both the shipping and shipbuilding industries. International conventions and related planning documents regulating ship environmental pollution and reducing energy consumption and emissions have been introduced one after another, and standards and regulations are becoming increasingly stringent. Green and low-carbon development has become an inevitable trend for the shipping industry.

[0003] Verification of a ship's hydrodynamic performance is an essential step in the ship design phase, and testing techniques are one method for verifying the effectiveness of green and energy-saving technologies in ships. Model testing involves placing a ship model in a towing tank for testing, collecting parameters of the model under different operating conditions, and then using the "model-to-ship" prediction method to forecast the ship's navigation performance under ideal conditions. Therefore, the results of model testing directly determine the accuracy of the predicted performance of the actual ship. Accurately describing and outputting the test results is a crucial issue during model testing.

[0004] The model ship resistance test is a test to measure the resistance of a model ship in a towing tank. Its basic principle is to tow a model ship, manufactured to a specific scale, at a corresponding draft, using a trailer at a specific speed, and then measuring the hydrostatic resistance encountered by the model using a drag meter. A reasonable and scientific evaluation method, focusing on the accuracy and stability of the resistance results, can effectively ensure the precision of the verification of ship hydrodynamic performance.

[0005] Regarding measurement methods for ship model resistance tests, the ITTC (International Towing Tank Conference) requires that test results be collected over a longer time series as much as possible during ship model tests to improve the stability of the results. However, due to limitations imposed by experimental conditions, and different methods used by different testers to extract time series, even time series from the same test can produce inconsistent test results, sometimes with significant deviations. These inconsistencies can negatively impact the accuracy of actual ship performance predictions and assessments.

[0006] The ITTC procedure does not explicitly specify a method for interpreting results, but the common approach used in different pools is to collect a time series as long as possible and directly average it. The obvious drawbacks of this method are: 1) the results are highly susceptible to human error; 2) it is difficult to assess the reliability of the results; and 3) it does not fully utilize the availability of the experimental time series. Summary of the Invention

[0007] To address the issue that the reliability of averaging time series data for ship model resistance testing is low due to limitations in experimental conditions, which in turn affects the accuracy of real-world performance prediction and evaluation, and the existing method of directly averaging a long time series data point is highly susceptible to human factors, making it difficult to assess the reliability of the results, thus failing to fully utilize the usability of the test time series, this invention proposes a resistance test result evaluation method based on random truncation of the time series. By randomly truncating the time series and adding a benchmark reference value, the usability of the time series is enhanced. Multiple sets of effective time series samples are obtained within a limited time series length. Through program control, the efficiency of analysis and calculation is improved, the influence of human factors is reduced, and the reliability and accuracy of ship model resistance test results are enhanced.

[0008] The specific plan is as follows:

[0009] A method for evaluating the resistance test results of a ship model based on random time series truncation:

[0010] S1: Acquisition steps of the original time series TS0: Acquire the resistance data of each time point in the time series TS0 under steady operation in the ship model resistance experiment, establish a rectangular coordinate system with time as the horizontal axis and resistance data as the vertical axis, and obtain the resistance-time curve corresponding to the time series TS0.

[0011] S2: Steps for obtaining time series TS1: Calculate the resistance mean R0 of the time series TS0. Draw a straight line parallel to the horizontal axis in the rectangular coordinate system corresponding to the time series TS0 with R0 as the vertical axis. This line will intersect the resistance-time curve at a series of points. Discard the time series before the first intersection point and the time series after the last intersection point. Record the first time point that crosses the R0 reference line upward as T0. Record the time points from the second time point that crosses the R0 reference line upward to the last time point that crosses the R0 reference line upward as T1, T2, ..., Tmax, respectively. Then, record the time series intervals T0-T1, T1-T2, ..., Tmax-1-Tmax as period segments N1, N2, ..., Nmax, that is, divide the time series into max period segments to obtain a new time series TS1. Calculate the resistance mean R1 and the root mean square error SD1 of the time series TS1. Use R1 and SD1 as benchmark reference values.

[0012] S3: Steps for obtaining the sample time series TSi: Plot a straight line parallel to the horizontal axis on the resistance-time curve corresponding to the time series TS1 with R1 as the vertical axis. Record the first time point that crosses the R1 reference line upward as t0. Sequentially record the time points from the second time point that crosses the R1 reference line upward to the last time point that crosses the R1 reference line upward as t1, t2, ... tmax, respectively. Each period is formed between every two time points that cross the R1 reference line upward. Randomly select any two time points that cross the R1 reference line upward to cut off the time series TS1. The time series between any two time points that cross the R1 reference line upward constitutes a continuous time series with a different number of periods, which is regarded as the sample time series TSi, where i represents the i-th sample time series. Calculate the resistance mean Ri and the root mean square error SDi of the sample time series TSi.

[0013] S4: Steps for determining valid sample time series: Compare the resistance mean Ri and SDi of each sample time series TSi in S3 with R1 and SD1 respectively to determine whether TSi is a valid sample time series.

[0014] S5: Convergence Evaluation: Calculate the mean value of Ri corresponding to the effective sample time series obtained in S4 to obtain RF, evaluate the deviation of RF from R1. If the deviation of RF from R1 is less than 1%, the result is considered to be converged and RF is output. If the deviation of RF from R1 is greater than 1%, return to S3, replace R1 with RF, and repeat steps S3-S5 until the result converges.

[0015] S6: Output Results: The Type A uncertainty UF corresponding to the output RF after the calculation results have converged is used as the final output result.

[0016] Preferably, the number of individual periods contained in the sample time series TSi in S3 is N. i The N i Greater than or equal to N min The N min The value is 15, and the sample time series TSi is a continuous truncation based on the time series TS1.

[0017] Preferably, the number of sample time series obtained in S3 is: (N1-N) min )+(N1-N min-1 )+…+1, which is equivalent to N cycles respectively. min 、(N min The time series of +1)…N are shifted on a time series of total length TS1.

[0018] Preferably, the formula for calculating the average resistance Ri of TSi in S4 is: Where Tsi(t) is the resistance monitoring value at any time, and Ts is the time series TS. i The total length.

[0019] Preferably, the formula for calculating the standard deviation SDi of TSi in S4 is: Where Tsi(t) is the resistance monitoring value at any time, and Ts is the time series TS. i The total length.

[0020] Preferably, the method for determining whether TSi is a valid sample time series as described in S4 is: if R i The deviation from R1 is within 2%, and SD i If the deviation from SD1 is within 10%, TSi is considered a valid sample time series; otherwise, TSi is discarded as an invalid series.

[0021] The present invention has the following beneficial effects:

[0022] This invention proposes a method for evaluating the test results of ship model resistance based on random truncation of time series. In particular, by adopting a random truncation method for the time series, multiple sets of effective time series samples can be obtained within a limited time series length, thereby improving the reliability of the test results and the accuracy of the evaluation of ship model resistance. Specifically, 1. This invention collects resistance data at various time points in a time series TS0 during a stable operation of a ship model resistance experiment. The original time series is truncated using the mean R0 of TS0 to obtain a time series TS1 with an integer number of periods. Then, the mean R1 and the added benchmark, standard deviation SD1, of TS1 are calculated. These two parameters, R1 and SD1, are used as benchmark reference values. Since sequences at the beginning and end of TS0 that do not belong to a complete period are removed, the large and unstable deviations in experimental data at the beginning and end of the experiment are avoided, thus improving the accuracy of the benchmark reference values. 2. This invention plots a straight line parallel to the horizontal axis on the resistance-time curve corresponding to the time series TS1 with R1 as the ordinate, generating a series of new intersection points. Each pair of intersection points forms a single period. Two intersection points are randomly selected to truncate the time series TS1. Random selection is reflected in two aspects: firstly, the starting time point of the sample time series is random; secondly, the length of the sample time series is random. Sample time series with the same starting point can have multiple lengths, and sample time series with the same length can have multiple starting points. (N1-N) was obtained from an original time series. min )+(N1-N min-1 This invention obtains multiple time series samples with different characteristics from a finite original time series. These time series samples are independent and random, significantly increasing the richness of the sample time series and improving the stability of the ship model resistance test results. 3. This invention combines the average resistance R of TSi from different time starting points and different numbers of periods. i and SD i The results are compared with R1 and SD1 respectively, and TSi with excessive deviation is removed. This allows the effective time sample to fluctuate within a small range. The mean value of Ri corresponding to the effective time sample TSi is then calculated and compared with the benchmark reference value R1 until the results converge. This greatly improves the accuracy of the test results of the ship model resistance test under the acquisition of finite time time series. Attached Figure Description

[0023] Figure 1 This is an overall flowchart of a method for evaluating the results of ship model resistance tests based on random truncation of time series.

[0024] Figure 2 This is the resistance-time curve corresponding to the original time series TS0 under steady-state conditions in the ship model resistance test.

[0025] Figure 3 This is the resistance-time curve of the original time series TS0 after removing the first and last sequences, TS1.

[0026] Figure 4 This is an example graph of the sample time series TSi extracted from the time series TS1.

[0027] Figure 5 It is a curve showing the resistance mean versus the number of cycles for all valid sample time series TSi obtained in the implementation case. Detailed Implementation

[0028] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0029] A method for evaluating ship model resistance test results based on random truncation of time series data. The overall flowchart of time series processing and test result evaluation in this invention is as follows: Figure 1 As shown, the steps for the ship model resistance test are as follows:

[0030] S1: Acquisition steps of the original time series TS0: Acquire resistance data at various time points in a time series TS0 during the steady-state operation of the ship model resistance experiment. Establish a rectangular coordinate system with time as the horizontal axis and resistance data as the vertical axis, as follows: Figure 2 The figure shows the resistance-time curve corresponding to the time series TS0;

[0031] S2: Steps for obtaining time series TS1: Calculate the average resistance R0 of the time series TS0. Plot a straight line parallel to the horizontal axis in the Cartesian coordinate system corresponding to the time series TS0, with R0 as the ordinate. This line will intersect the resistance-time curve at a series of points. Discard the time series before the first intersection point and the time series after the last intersection point. Record the first time point that crosses the R0 reference line upwards as T0. Record the second time point that crosses the R0 reference line upwards to the last time point that crosses the R0 reference line upwards as T1, T2…Tmax, respectively. Then, record the time series intervals T0-T1, T1-T2,…, Tmax-1-Tmax as period segments N1, N2,…, Nmax, that is, divide the time series into max period segments to obtain a new time series TS1; the obtained new time series, as shown... Figure 3 As shown, the mean resistance value R1 = 4.64049 and the root mean square deviation SD1 = 0.54857 of the time series TS1 are used as the benchmark reference values.

[0032] S3: Steps for obtaining the sample time series TSi: Plot a straight line parallel to the horizontal axis on the resistance-time curve corresponding to the time series TS1 with R1 as the ordinate. Record the first time point that crosses the R1 reference line upwards as t0. Sequentially record the time points from the second to the last time point that crosses the R1 reference line upwards as t1, t2, ..., tmax, respectively. Each two times that cross the R1 reference line upwards form a period. Randomly select any two times that cross the R1 reference line upwards to truncate the time series TS1. The time series between these two times constitutes a continuous time series with a different number of periods, which is considered the sample time series TSi, where i represents the i-th sample time series. Calculate the mean resistance Ri and the root mean square deviation SD of the sample time series TSi. i ;like Figure 4 As shown, the time series between time point t1 and time point tmax is used as the sample time series selected in this embodiment. In this experiment, the mean resistance Ri of the sample time series TSi formed by time point t1 and time point tmax is 4.64778, and the root mean square error SDi is 0.53654.

[0033] S4: Steps for determining valid sample time series: Compare the resistance mean Ri and SDi of each sample time series TSi in S3 with R1 and SD1 respectively to determine whether TSi is a valid sample time series.

[0034] Based on combinations of continuous time series TSi with different numbers of periods, a series of sample calculation results Ri and SDi are generated. These are compared with R1 and SD1, and results with excessive deviations are eliminated. Finally, a series of calculation results that meet the requirements and fluctuate within a certain range are obtained. Figure 5 As shown, in this embodiment, the extracted period lengths Ni range from 15 to 21. The number of valid time samples obtained is 17. Figure 5 These are the number of periods and the average resistance value corresponding to these 17 different effective time samples;

[0035] S5: Convergence Evaluation: Calculate the mean value of Ri corresponding to the 17 valid sample time series obtained in S4 to obtain RF, evaluate the deviation of RF from R1. If the deviation of RF from R1 is less than 1%, the result is considered to be converged and RF is output. If the deviation of RF from R1 is greater than 1%, return to S3, replace R1 with RF, and repeat steps S3-S5 until the result converges.

[0036] S6: Output Results: The Type A uncertainty UF corresponding to the output RF after the calculation results have converged is used as the final output result.

[0037] Preferably, the number of individual periods contained in the sample time series TSi in S3 is N. i The N i Greater than or equal to N min The N min The value is 15, and the sample time series TSi is a continuous truncation based on the time series TS1.

[0038] Preferably, the number of sample time series obtained in S3 is: (N1-N) min )+(N1-N min-1 )+…+1, which is equivalent to N cycles respectively. min 、(N min The time series of +1)…N are shifted on a time series of total length TS1.

[0039] Preferably, the formula for calculating the average resistance Ri of TSi in S4 is: Where Tsi(t) is the resistance monitoring value at any time, and Ts is the total length of the time series TSi.

[0040] Preferably, the formula for calculating the standard deviation SDi of TSi in S4 is: Where Tsi(t) is the resistance monitoring value at any time, and Ts is the time series TS. i The total length.

[0041] Preferably, the method for determining whether TSi is a valid sample time series as described in S4 is: if R i The deviation from R1 is within 2%, and SD i If the deviation from SD1 is within 10%, TSi is considered a valid sample time series; otherwise, TSi is discarded as an invalid series.

[0042] Although this specification has described the invention in detail with reference to the accompanying drawings and embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the invention. In short, all technical solutions and improvements that do not depart from the spirit and scope of the invention should be covered within the protection scope of the patent of this invention.

Claims

1. A method for evaluating the results of ship model resistance tests based on time series random truncation, characterized in that, S1: Acquisition steps of the original time series TS0: Acquire the resistance data of each time point in the time series TS0 under steady operation in the ship model resistance experiment, establish a rectangular coordinate system with time as the horizontal axis and resistance data as the vertical axis, and obtain the resistance-time curve corresponding to the time series TS0. S2: Steps for obtaining time series TS1: Calculate the resistance mean R0 of the time series TS0. Draw a straight line parallel to the horizontal axis in the rectangular coordinate system corresponding to the time series TS0 with R0 as the vertical axis. This line will intersect the resistance-time curve at a series of points. Discard the time series before the first intersection point and the time series after the last intersection point. Record the first time point that crosses the R0 reference line upward as T0. Sequentially record the time points from the second time point that crosses the R0 reference line upward to the last time point that crosses the R0 reference line upward as T1, T2, ..., Tmax, respectively. Then, sequentially record the time series intervals T0-T1, T1-T2, ..., Tmax-1-Tmax as period segments N1, N2, ..., Nmax, that is, divide the finite time series into max period segments to obtain a new time series TS1. Calculate the resistance mean R1 and the root mean square error SD1 of the time series TS1. Use R1 and SD1 as benchmark reference values. S3: Steps for obtaining the sample time series TSi: Plot a straight line parallel to the horizontal axis on the resistance-time curve corresponding to the time series TS1, with R1 as the ordinate. Record the first time point that crosses the R1 reference line upwards as t0. Sequentially record the time points from the second to the last time point that crosses the R1 reference line upwards as t1, t2, ..., tmax, respectively. Each two times point that crosses the R1 reference line upwards forms a period. Randomly select any two times point that crosses the R1 reference line upwards to truncate the time series TS1. The time series between these two times point constitutes a continuous time series with a different number of periods, which is considered the sample time series TSi, where i represents the i-th sample time series. Calculate the mean resistance Ri and the root mean square error SDi of the sample time series TSi. The sample time series TSi contains N single periods. i The N i Greater than or equal to N min The N min The value is 15, and the sample time series TSi is continuously truncated based on the time series TS1; the number of sample time series obtained by S3 is: (N1-N) min )+(N1-N min-1 )+…+1, which is equivalent to N cycles respectively. min 、(N min The time series of +1)…N are shifted over a time series of total length TS1; S4: Steps for determining valid sample time series: Compare the resistance mean Ri and SDi of each sample time series TSi in S3 with R1 and SD1 respectively to determine whether TSi is a valid sample time series. S5: Convergence Evaluation: Calculate the mean value of Ri corresponding to the effective sample time series obtained in S4 to obtain RF, evaluate the deviation of RF from R1. If the deviation of RF from R1 is less than 1%, the result is considered to be converged and RF is output. If the deviation of RF from R1 is greater than 1%, return to S3, replace R1 with RF, and repeat steps S3-S5 until the result converges. S6: Output Results: The Type A uncertainty UF corresponding to the output RF after the calculation results have converged is used as the final output result.

2. The method for evaluating ship model resistance test results based on time series random truncation as described in claim 1, characterized in that, The formula for calculating the average resistance Ri of TSi in S4 is as follows: Where Tsi(t) is the resistance monitoring value at any time, and Ts is the time series TS. i The total length.

3. The method for evaluating ship model resistance test results based on time series random truncation as described in claim 1, characterized in that, The formula for calculating the standard deviation SDi of TSi in S4 is as follows: Where Tsi(t) is the resistance monitoring value at any time, and Ts is the time series TS. i The total length.

4. The method for evaluating ship model resistance test results based on time series random truncation as described in claim 1, characterized in that, The method described in S4 for determining whether TSi is a valid sample time series is as follows: if the deviation between Ri and R1 is within 2%, and the deviation between SDi and SD1 is within 10%, then TSi is considered a valid sample time series; otherwise, TSi is discarded as an invalid series.

Images