Optimization design method and optimization design device for joint return period

Abstract

Provided are an optimization design method and an optimization design device for a joint recurrence period. The optimization design method includes: obtaining a multidimensional environmental variable data set of a target area and determining a joint recurrence period to be solved; determining the extreme value of the marginal distribution variable corresponding to a first set in the multidimensional environmental variable data set under the joint recurrence period; sorting the first set and dividing the sorted first set into an extreme part and a non-extreme part based on the extreme value of the marginal distribution variable; performing probability distribution fitting on the extreme part and the non-extreme part respectively to obtain a first fitting result of the first set, and evaluating the first fitting result using a first environmental contour evaluation index; in response to the evaluation result of the first fitting result not meeting the standard, optimizing the first fitting result to obtain a first fitting optimization result of the first set. The optimization design method according to an embodiment of the present disclosure can improve the marginal distribution fitting accuracy of multidimensional environmental variable data.

Description

Technical Field

[0001] The present application relates to the field of marine engineering and marine environment, and more specifically, to an optimization design method and an optimization design device for a joint recurrence period. Background Art

[0002] Determining joint return periods is crucial for marine engineering, water resources management, urban planning, and infrastructure design. It enables engineers to accurately respond to situations where multiple environmental variables, such as storm surges and waves, reach extreme values ​​simultaneously. This is crucial for ensuring the safety and functionality of offshore platforms, seawalls, flood control measures, and other critical structures during extreme climate events, thereby protecting lives and property and significantly reducing economic losses and casualties caused by disasters.

[0003] The Environmental Contour Method (ECM) has become an effective tool for predicting these extreme ocean conditions. Recommended by the International Electrotechnical Commission (IEC), this method analyzes limited historical meteorological and oceanographic data to predict possible future extreme conditions. Compared to traditional extreme value analysis based on a single parameter, the ECM more carefully analyzes the interrelationships between various environmental parameters, constructing multidimensional spatial contours that reflect the same return period. These contours more accurately represent the joint probability of factors such as wind speed and wave height, thereby comprehensively capturing the extreme conditions that may affect structural design and safety.

[0004] The environmental contour method derived by the traditional IEC method has the following shortcomings:

[0005] When fitting marginal distributions, the commonly used single fitting function is unable to effectively capture the distribution characteristics of data under extreme conditions. This approach often leads to large deviations in the results when extrapolating the marginal distribution function to a specific return period, thus affecting overall accuracy. Summary of the Invention

[0006] One of the objectives of the present disclosure is to provide an optimization design method that can effectively capture the distribution characteristics of multidimensional environmental variable data under extreme conditions.

[0007] According to a first aspect of the present disclosure, a method for optimizing a joint recurrence period is provided, the method comprising: obtaining a multidimensional environmental variable data set of a target area and determining a joint recurrence period to be solved; determining a marginal distribution variable corresponding to a first set in the multidimensional environmental variable data set under the joint recurrence period; sorting the first set and dividing the sorted first set into an extreme part and a non-extreme part based on the marginal distribution variable; performing probability distribution fitting on the extreme part and the non-extreme part, respectively, to obtain a first fitting result of the first set, and evaluating the first fitting result using a first environmental contour evaluation index; in response to an evaluation result of the first fitting result not meeting the standard, optimizing the first fitting result to obtain a first fitting optimization result of the first set.

[0008] According to an embodiment of the present disclosure, optimizing the first fitting result to obtain the first set of first fitting optimization results may include: determining the first fitting optimization result based on a connection point between the fitting results of the extreme part and the fitting results of the non-extreme part.

[0009] According to an embodiment of the present disclosure, the step of determining the first fitting optimization result based on the connection point of the fitting result of the extreme part and the fitting result of the non-extreme part may include: fitting the non-extreme part to obtain the non-extreme part fitting result, and fitting the extreme part to obtain the extreme part fitting result; determining the connection point of the non-extreme part fitting result and the extreme part fitting result; determining the first fitting optimization result of the first set based on the non-extreme part fitting result, the extreme part fitting result and the connection point.

[0010] According to an embodiment of the present disclosure, the step of determining the first fitting optimization result of the first set based on the non-extreme part fitting results, the extreme part fitting results and the connection point may include: taking the combination of the non-extreme part fitting results before the connection point and the extreme part fitting results after the connection point as the first fitting optimization result.

[0011] According to an embodiment of the present disclosure, the connection point may be a point where the extreme part fitting goodness is the highest among the intersections of the non-extreme part fitting result and the extreme part fitting result.

[0012] According to an embodiment of the present disclosure, the step of fitting the extreme part to obtain the extreme part fitting result may include: using the log-normal distribution as the fitting model and fitting the extreme part with the weighted least squares method, and the weight of the weighted least squares method may be determined by a genetic algorithm.

[0013] According to an embodiment of the present disclosure, the optimization design method may also include: dividing the sorted first set into multiple segments, and obtaining the conditional probability distribution of the second set in the multidimensional environmental variable data set in each segment; solving the first environmental contour line results of the marginal distribution variables corresponding to the first set and the second variables corresponding to the second set under the joint recurrence period based on the conditional probability distribution and the first fitting optimization result.

[0014] According to an embodiment of the present disclosure, the optimization design method may further include: performing a second environmental contour index evaluation on the first environmental contour result; in response to the second environmental contour index evaluation meeting the standard, using the first environmental contour result as the final environmental contour result.

[0015] According to an embodiment of the present disclosure, the first environmental contour line evaluation index may include at least one of the following indicators: the determination coefficient corresponding to the extreme part, the root mean square error corresponding to the extreme part, and the extreme value envelope rate of the environmental contour line; the second environmental contour line evaluation index may include the relative error between the maximum value of the marginal distribution variable based on the extrapolation of the environmental contour line and the extreme value obtained by the EVA method.

[0016] According to an embodiment of the present disclosure, the step of dividing the sorted first set into extreme parts and non-extreme parts based on marginal distribution variables may include: calculating the cumulative exceedance probability based on the marginal distribution variables, and dividing the first set into extreme parts and non-extreme parts according to the cumulative exceedance probability.

[0017] According to an embodiment of the present disclosure, the multidimensional environmental variable data set may be a multidimensional ocean environmental variable data set, and may include a significant wave height data set and a spectrum peak period data set, and the joint recurrence period may be a two-dimensional variable joint recurrence period.

[0018] According to a second embodiment of the present disclosure, an optimization design device for a joint recurrence period is provided, the optimization design device comprising: a first calculation unit, which obtains a multidimensional environmental variable data set of a target area and determines a joint recurrence period to be solved; a second calculation unit, which determines a marginal distribution variable corresponding to a first set in the multidimensional environmental variable data set under the joint recurrence period; an extreme event separation unit, which sorts the first set and divides the sorted first set into an extreme part and a non-extreme part based on the marginal distribution variable; a fitting unit, which performs probability distribution fitting on the extreme part and the non-extreme part respectively to obtain a first fitting result of the first set, and evaluates the first fitting result using a first environmental contour evaluation index; an optimization unit, which optimizes the first fitting result in response to an evaluation result of the first fitting result not meeting the standard, to obtain a first fitting optimization result of the first set.

[0019] According to a third aspect of the present disclosure, a computing device is provided, comprising a memory and a processor, wherein the memory stores instructions or programs, and when the instructions or programs are executed by the processor, the processor is prompted to execute the above-mentioned optimization design method.

[0020] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor is prompted to execute the above-mentioned optimization design method.

[0021] According to the optimization design method of the embodiment of the present disclosure, the accuracy of marginal distribution fitting of multidimensional environmental variable data can be improved by combining the connection points for marginal distribution fitting, and the result deviation can be effectively reduced.

[0022] The optimization design method according to the embodiment of the present disclosure can enhance the objective evaluation capability of environmental contour results by evaluating the contour results using preset rating indicators, thereby improving accuracy and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a flowchart illustrating a method for optimizing design of joint return periods according to a first embodiment of the present disclosure.

[0024] Figure 2 is a flowchart illustrating a method for optimizing design of joint return periods according to a second embodiment of the present disclosure.

[0025] Figure 3 FIG. 4 is a flowchart illustrating specific optimization steps of a first fitting optimization result according to an embodiment of the present disclosure.

[0026] Figure 4 is a flow chart illustrating an overall optimization design method according to an embodiment of the present disclosure.

[0027] Figure 5 It is the hourly significant wave height and spectral peak period hindcast result based on the numerical simulation of the ocean model of the offshore wind farm according to the embodiment of the present disclosure.

[0028] Figure 6 It is the comparison result between the probability density obtained by different fitting models and fitting methods and the probability density obtained by numerical simulation of ocean models.

[0029] Figure 7 It is the comparison result between the probability density obtained based on the non-extreme part and the extreme part and the probability density obtained by numerical simulation of the ocean model.

[0030] Figure 8 1 is a graph showing the comparison results between the probability density obtained by using the three-parameter Weibull distribution and the method disclosed herein and the probability density obtained by model numerical simulation.

[0031] Figure 9 1 shows the environmental contour lines obtained by the optimization design method according to the embodiment of the present disclosure and the environmental contour lines obtained by the IEC method.

[0032] Figure 10 is a block diagram illustrating an apparatus for optimizing design of joint return periods according to the present disclosure. DETAILED DESCRIPTION

[0033] The following detailed description is provided to assist in gaining a comprehensive understanding of the methods, apparatuses, and / or systems described herein. However, the order of operations described herein is merely exemplary and is not limited to the order set forth herein. Equivalent substitutions or variations are possible, except for operations that must occur or be performed in a specific order. Furthermore, for the sake of clarity and conciseness, descriptions of matters known in the art may be omitted or simplified.

[0034] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present disclosure pertains after understanding the present disclosure. Unless expressly defined as such herein, terms (such as those defined in commonly used dictionaries) should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and the present disclosure, and should not be interpreted in an idealized or overly formal manner.

[0035] Unless otherwise specified, the same reference numerals generally refer to the same elements (e.g., components, steps, and methods). Reference numerals described in previous embodiments that appear again in later embodiments may be omitted. Furthermore, technical features described in different or the same embodiments may be combined in any manner, as long as the combined embodiment or technical solution is complete and can solve the technical problems of this application or achieve technical effects described or not described in this disclosure but determinable based on the above complete technical solution. The following describes the terms used in this disclosure.

[0036] Environmental contour line: also called environmental outline, is a probability contour line drawn in a multidimensional parameter space where one or more environmental variables (such as wind speed, wave height, flow velocity, wave period, etc.) reach or exceed a certain threshold.

[0037] EVA, also known as traditional extreme value statistics, is a statistical method used to evaluate and model rare events, particularly extreme events in fields such as meteorology, hydrology, engineering, and financial risk management. EVA focuses on understanding extremely rare data points that deviate significantly from the average, rather than the central tendency of a distribution. Extreme value analysis first involves selecting samples from time series data and then fitting a line to the sample to obtain the optimal extreme value distribution function. This function is then used to analyze extreme values ​​under different return periods.

[0038] For the probability of occurrence of a specific event, the extreme value analysis uses empirical probability calculation, as shown in the following formula (1):

[0039] (1)

[0040] Where n is the total number of extreme value samples, m is the sample number arranged in descending order, and the value of m ranges from 1 to n. The observed extreme value sequence is fitted to the theoretical extreme value probability distribution function to establish the extreme value model.

[0041] R 2 : Coefficient of determination is a statistical indicator used in regression analysis to measure the degree of fit between the original sample and the fitted sample. Its calculation formula is: , where n is the number of samples, and Represent the cumulative probability density of the original sample and the fitted distribution function, respectively. Represents the mean of the original sample cumulative probability density.

[0042] RMSE: Root mean square error, a standard measure of the difference between the original sample and the fitted sample. Its calculation formula is , where the symbols have the same meaning as above.

[0043] ELR: Envelope rate, used to measure the degree of envelope of the environmental contour line of a specific return period to the data point distance within the return period. Its calculation formula is , where and They represent the number of data points enclosed by the environmental contour line and the total number of data points, respectively.

[0044] RE: Relative error, used to determine the degree of deviation between the extreme value derived from the environmental contour line of a specific recurrence period and the extreme value derived from the traditional extreme value statistical method. The calculation formula is: , where and They represent the extreme values ​​of the extrapolated environmental contours and the extreme values ​​of the extrapolated EVA method respectively.

[0045] Environmental Contour Method: Based on the inverse first-order reliability method (IFORM), environmental probability contours are defined in the environmental parameter space according to the relevant probability distribution. Short-term load and response calculations are performed along these contours for a limited number of critical sea conditions. The specific return period response of the structure is estimated by selecting appropriate short-term response quantiles, significantly reducing simulation time and costs. The solution process meets the requirements of IEC 61400-3-1-2019.

[0046] The optimization design method of the present disclosure can be executed or implemented by various computing devices such as computers.

[0047] The present disclosure addresses the aforementioned deficiencies in the prior art and provides an optimization design method for a joint return period. Furthermore, the optimization design method of the present disclosure can accurately estimate the joint return period of two-dimensional variables based on environmental contours, optimize the marginal distribution fitting function, and utilize the IFORM principle and Rosenblatt transformation to obtain environmental contour results for a specified joint return period. These results are then evaluated using a constructed uncertainty evaluation scheme, thereby achieving accurate output of environmental operating conditions for a specified joint return period. This is described in detail below in conjunction with the accompanying drawings of the present disclosure.

[0048] Figure 1 is a flowchart illustrating a method for optimizing a joint return period according to a first embodiment of the present disclosure, Figure 2 is a flowchart illustrating a method for optimizing a joint return period according to a second embodiment of the present disclosure, Figure 3 is a flowchart showing specific optimization steps of the first fitting optimization result according to an embodiment of the present disclosure, Figure 4 is a flow chart illustrating an overall optimization design method according to an embodiment of the present disclosure, Figure 5 is the hourly significant wave height and spectrum peak period hindcast result based on the numerical simulation of the offshore wind farm ocean model according to an embodiment of the present disclosure, Figure 6 is the comparison result between the probability density obtained by different fitting models and fitting methods and the probability density obtained by numerical simulation of ocean models. Figure 7 It is the result of comparing the probability density obtained based on the non-extreme part and the extreme part with the probability density obtained by numerical simulation of the ocean model. Figure 8 : is a comparison result showing the probability density obtained by using the three-parameter Weibull distribution and the method disclosed in the present invention and the probability density obtained by the numerical simulation of the model, Figure 9 1 shows the environmental contour lines obtained by the optimization design method according to the embodiment of the present disclosure and the environmental contour lines obtained by the IEC method.

[0049] Reference Figure 1The optimization design method of the joint return period according to an embodiment of the present disclosure may include step S110, step S120, step S130, step S140 and step S150.

[0050] In step S110 , a multi-dimensional environmental variable data set of a target area is obtained and a joint recurrence period to be solved is determined.

[0051] The multidimensional environmental variable dataset here can be a multidimensional marine environmental variable dataset, for example, including a significant wave height dataset and a spectral peak period dataset. Based on this, the joint return period can be a two-dimensional variable joint return period, that is, the multidimensional marine environmental variable includes both significant wave height and spectral peak period. However, this is merely an example; the multidimensional environmental variable dataset can also include wind speed, current velocity, and so on. This disclosure only uses significant wave height and spectral peak period as examples for description.

[0052] The multidimensional environmental variable data set here can be obtained from a third party, such as an ocean data center, or from an ocean observation station equipped with various measuring instruments. For example, nearly 40 years of multidimensional environmental variable data can be obtained, and the joint return period to be solved can be set to 50 years. However, this is merely an example.

[0053] In step S120 , the marginal distribution variables corresponding to the first set in the multidimensional environmental variable data set under the joint return period are determined.

[0054] Here, the marginal distribution variable is the significant wave height and the first set is the significant wave height data set. However, this is only an example, and other variables such as wind speed can also be used as marginal distribution variables.

[0055] The significant wave height corresponding to the significant wave height data set in the two-dimensional environmental variable data set under this joint return period can be determined as the marginal distribution variable. Furthermore, the extreme value of the significant wave height under this joint return period can be inferred using the existing EVA method. The extreme value of the marginal distribution variable extrapolated by the EVA method can be used to evaluate subsequent evaluation indicators. Specifically, based on the characteristics of the significant wave height data, an appropriate probability distribution model can be selected to fit the data. Once the optimal extreme value distribution function is obtained, the extreme value under this return period can be determined based on this function.

[0056] Reference Figure 5 , Figure 5 The long series of hindcast data of an offshore wind farm simulated by ocean model is shown. The results include 44 years of ocean hydrological environmental parameters such as wind speed, significant wave height (Hs) and spectral peak period (Tp) with an interval of 1 hour. The joint recurrence period to be solved is determined to be once in 50 years. Figure 5The 50-year Hs obtained by the EVA method is 11.23m, and the environmental contour line category is the Hs-Tp environmental contour line.

[0057] In step S130 , the first set is sorted and the sorted first set is divided into an extreme part and a non-extreme part based on the marginal distribution variable.

[0058] As an example, the significant wave height data set as the first set can be sorted from large to small, and then the cumulative exceedance probability of each Hs as the marginal distribution variable is calculated, and the Hs with a cumulative exceedance probability of 99.8% is determined to be 5.4m. That is, the part with a cumulative exceedance probability greater than 99.8% is defined as an extreme part or an extreme event, and the part that does not exceed 99.8% is defined as a non-extreme part or a non-extreme event. In other words, the data with a significant wave height greater than 5.4m accounts for 0.2%, that is, the non-extreme part accounts for 99.8%, and the extreme part accounts for 0.2%. Therefore, the step of dividing the sorted first set into extreme parts and non-extreme parts based on the marginal distribution variables includes: calculating the cumulative exceedance probability based on the marginal distribution variables, and dividing the first set into extreme parts and non-extreme parts according to the cumulative exceedance probability.

[0059] As mentioned above, features with percentiles exceeding the 99.8th percentile are defined as extreme events. For example, if the 99.8th percentile corresponds to a significant wave height of 5.40 m, then any moment with a wave height exceeding 5.40 m is considered an extreme event. This method allows for a more accurate assessment and analysis of the impact of different fitting methods and optimization correction approaches on environmental contours.

[0060] As described above, according to the embodiments of the present disclosure, attention is paid to the fitting of extreme events and their data envelopment characteristics. 2 If the dataset used for the RMSE and ELR metrics includes all events, the results obtained by different fitting methods may differ slightly, thereby weakening the sensitivity of the fitting method to extreme events. Therefore, embodiments of the present disclosure can only consider datasets of extreme events when calculating evaluation metrics.

[0061] In step S140 , probability distribution fitting is performed on the extreme part and the non-extreme part respectively to obtain a first fitting result of a first set, and the first fitting result is evaluated using a first environmental contour evaluation index.

[0062] The multidimensional environmental variables described above can be marine hydrological environmental parameters or variables. Commonly used distribution functions for these parameters or variables include the three-parameter Weibull, lognormal, exponential-Weibull, and Rayleigh distributions. Commonly used fitting methods include maximum likelihood (MLE), least squares (LS), and genetic algorithms (GA). Depending on the data characteristics, different combinations of distribution functions and fitting methods can yield significantly different results. The specific fitting process using fitting models such as the three-parameter Weibull, lognormal, exponential-Weibull, and Rayleigh distributions, as well as fitting methods such as maximum likelihood (MLE), least squares (LS), and genetic algorithms (GA), can all be based on existing methods.

[0063] According to an embodiment of the present disclosure, the same data source can first be fitted based on a combination of different distribution functions and different fitting methods by cross-combining different distribution functions in the above-mentioned distribution functions with different fitting methods, and then the optimal combination method can be determined based on a preset objective function to achieve the optimized combination function of the fitting functions and methods.

[0064] As an example, we can first fit 12 different combinations of the same data source by cross-combining 4 distribution functions with 3 fitting methods. Then, with the minimization of overall error and extreme value error as the objective function, we can determine the optimal combination method and realize the optimized combination function of fitting functions and methods.

[0065] The fitting results can be seen in Table 1 and Figure 6 shown.

[0066] Table 1 Probability distribution fitting results of Hs by different fitting models and fitting methods

[0067]

[0068] The first environmental contour evaluation index may include at least one of the following indexes: the determination coefficient R corresponding to the extreme part 2 (Evaluation indicator 1), the root mean square error RMSE corresponding to the extreme part (Evaluation indicator 2), and the extreme envelope rate ELR of the environmental contour line (Evaluation indicator 3).

[0069] As an example, it can be determined whether R 2The three constraints of >0.90, RMSE<0.05, and RE<1% are met. If they are met, the marginal distribution result as the first fitting result meets the standard. If they are not met, the marginal distribution result as the first fitting result does not meet the standard.

[0070] In step S150 , in response to the evaluation result of the first fitting result not meeting the standard, the first fitting result may be optimized to obtain a first set of first fitting optimization results.

[0071] As an example, optimizing the first fitting result to obtain the first set of first fitting optimization results may include: determining the first fitting optimization result based on a connection point (or anchor point) between the fitting results of the extreme part and the fitting results of the non-extreme part.

[0072] When there are multiple connection points, the point with the highest extreme portion goodness of fit among the intersections of the non-extreme portion fitting result and the extreme portion fitting result may be selected as the connection point (i.e., intersection point) for determining the first fitting optimization result. For example, if the non-extreme portion fitting curve (which is the non-extreme portion fitting result) and the extreme portion fitting curve (which is the extreme portion fitting result) have two intersections (i.e., connection points), the point with the smallest deviation from the true data among these two intersections may be selected as the connection point for selecting the fitting curve.

[0073] Reference Figure 3 The step of determining the first fitting optimization result based on the connection point between the fitting result of the extreme part and the fitting result of the non-extreme part includes step S1511, step S1512 and step S1513.

[0074] In step S1511 , fitting is performed on the non-extreme portion to obtain a non-extreme portion fitting result, and fitting is performed on the extreme portion to obtain an extreme portion fitting result.

[0075] As an example, for the fitting of non-extreme events, different fitting models and fitting methods can be combined. There are four fitting models to choose from, namely three-parameter Weibull, Exponential-Weibull, Lognormal and Rayleigh distribution. There are three fitting methods to choose from, namely least squares (LS), maximum likelihood (MLE) and artificial intelligence genetic algorithm (GA). The combination that best fits the non-extreme part of the variable is selected as the fitting result of that part.

[0076] The step of fitting the extreme part to obtain the extreme part fitting result includes: using lognormal distribution as a fitting model and fitting the extreme part with weighted least squares method, and the weight of the weighted least squares method is determined by genetic algorithm.

[0077] As an example, in order to ensure that the extrapolated return period extreme value is as close as possible to the result of the EVA method while effectively improving the goodness of fit of the extreme value part, the optimization design method according to the embodiment of the present disclosure can compare the extrapolated return period extreme value with the return period extreme value obtained by the EVA method, and use the minimum error as the objective function, the weight of each sample as the decision variable, and automatically optimize the parameters based on the NSGA-Ⅱ algorithm (non-dominated sorting genetic algorithm with elite strategy) to find the weight under the global optimal state.

[0078] That is to say, Lognormal can be selected for fitting extreme events. The fitting method adopts weighted least squares method and sets the weight value of each extreme event sample. The larger the weight value, the more important the information corresponding to the sample. The fitting goals are as follows: to ensure that the fitting results are as close as possible to the distribution characteristics of extreme event data; the extrapolated return period extreme value is as close as possible to the result obtained by the EVA method.

[0079] As an example, a genetic algorithm can be used for weight optimization, with the R 2 The three objective functions of maximum, RMSE and RE minimum are used. The constraint range of each weight value is limited to between 0 and 1. Five intervals or bins can be set. The final parameters can be: when Hs∈[5.40,6.35), weight = 0.111; when Hs∈[6.35,7.30), weight = 0.137; when Hs∈[7.30,8.26), weight = 0.089; when Hs∈[8.26,9.21), weight = 0.107; when Hs≥9.21, weight = 0.555.

[0080] In step S1512 , the connection point between the non-extreme part fitting result and the extreme part fitting result is determined.

[0081] Reference Figure 7 The fitting functions of the non-extreme part and the extreme part form two intersection points (connection points). The point with higher fitting goodness of extreme events is selected as the connection point of the marginal distribution function, that is, Hs=3.5m. The Hs probability distribution function obtained is shown in the attached figure. Figure 8 shown.

[0082] In step S1513, a first set of first fitting optimization results is determined according to the non-extreme part fitting results, the extreme part fitting results, and the connection points.

[0083] As an example, the step of determining the first fitting optimization result of the first set based on the non-extreme part fitting results, the extreme part fitting results and the connection point includes: taking the combination of the non-extreme part fitting results before the connection point and the extreme part fitting results after the connection point as the first fitting optimization result.

[0084] That is, before the connection point, the fitting results of the non-extreme part are used, and after the connection point, the fitting results of the extreme part are used. The connection point is selected from the intersection of the two fitting results. If there are multiple intersection points, the intersection point with the higher goodness of fit of the extreme event can be selected as the connection point of the marginal distribution function as described above.

[0085] The above optimization method can make up for the lack of understanding of extreme events when fitting the marginal distribution function, improve the extrapolation accuracy of the marginal distribution function, and thus improve the accuracy of solving the joint working conditions by environmental contour lines.

[0086] Reference Figure 2 , the optimization design method according to the present disclosure may further include step S160 and step S170.

[0087] In step S160 , the sorted first set is divided into a plurality of segments, and the conditional probability distribution of the second set in the multidimensional environmental variable data set within each segment is obtained.

[0088] As an example, the significant wave height data set can be divided into several segments with equal intervals, the number of samples of another set (for example, spectral peak period) can be counted within each segment, and the conditional probability distribution within each segment can be obtained using the log-normal distribution. However, this is only an example, and the conditional probability distribution can be obtained using other methods.

[0089] In step S170, first environmental contour results of marginal distribution variables corresponding to the first set and second variables corresponding to the second set under the joint return period are solved based on the conditional probability distribution and the first fitting optimization result.

[0090] Here, the method of obtaining environmental contours based on conditional probability distribution, marginal distribution, etc. can adopt existing methods, and environmental contours can be obtained based on the IRORM principle and Rosenblatt transformation. The brief process is as follows: according to the failure probability, the set in the standard normal space (U space) is converted into the environmental variables in the physical space (X space) through the inverse Rosenblatt transformation, thereby obtaining environmental contours. The following set Figure 4 The overall optimization design method disclosed in the present invention is described.

[0091] Reference Figure 4 In step S10, a long sequence data set of significant wave height and spectral peak period is obtained, the joint recurrence period is determined, and the extreme value under the joint recurrence period is deduced based on the EVA method.

[0092] In step S20 , extreme events are separated. As described above, the extreme part and the non-extreme part can be separated by cumulative exceedance probability.

[0093] In step S30, marginal distribution fitting can be performed using the above-mentioned fitting model and fitting method.

[0094] When the evaluation index 1, the evaluation index 2 and the evaluation index 3 all fail to meet the standards, marginal distribution optimization may be performed in step S40 , and the specific optimization method is as described above.

[0095] In step S50, conditional distribution fitting is performed. As an example, the significant wave height data set can be divided into several segments at equal intervals, the number of samples of another set (for example, spectral peak period) is counted within each segment, and the conditional probability distribution within each segment is obtained using the log-normal distribution. However, this is only an example, and the conditional probability distribution can be obtained using other methods.

[0096] S70: Environmental contours are solved based on IFORM and Rosenblatt transformation. The set in standard normal space (U space) can be converted into environmental variables in physical space (X space) through the inverse Rosenblatt transformation, thereby obtaining environmental contours.

[0097] The comparison of the probability distribution fitting results of Hs by the IEC method and the optimization design method according to the present disclosure is shown in Table 2 below.

[0098] Table 2 Probability distribution fitting results of Hs by traditional IEC method and this method

[0099]

[0100] As shown in Table 2, the traditional IEC method using the combination of Weibull and maximum likelihood methods does not provide an ideal fitting effect in this embodiment, especially in the extreme part. Figure 7 Based on the distribution trend of the data points, the traditional IEC method significantly underestimates the specified return period, which undoubtedly affects the design safety of offshore structures. However, the present disclosure uses a piecewise function approach, which achieves ideal fitting results in both extreme and non-extreme regions. The extrapolated 50-year extreme value is consistent with the EVA method, meeting the requirements of the environmental contour evaluation index.

[0101] In addition, refer to Figure 9The red dots represent the long-series hindcast data of the ocean model numerical simulation, and the black line represents the 50-year environmental contour line obtained by the traditional IEC method. From the above analysis, it can be seen that the marginal distribution function fitted by the traditional method is not ideal in the extreme value area, and the extrapolated 50-year Hs has a significant underestimate. Therefore, the environmental contour line obtained based on this method omits a large number of large Hs data points. For the extreme events defined in this disclosure, the ELR index is even only 0%. Therefore, according to the environmental contour line evaluation index defined in this disclosure, the results obtained by the traditional method do not meet the standards. The results obtained by using the method of the present invention to deduce the environmental contour line are shown in the attached figure. Figure 9 As shown by the blue line. As can be seen from the above analysis, the cumulative probability distribution of Hs obtained by the method of the present disclosure can effectively capture extreme events, and the extrapolated 50-year extreme value is completely consistent with the EVA method. Furthermore, the envelope rate of the environmental contour lines obtained based on this method for extreme events also meets the evaluation indicators set by the present disclosure. Therefore, the results can be considered to meet the standards and can be used in the design and construction of offshore wind turbines. Therefore, the optimization design method according to the present disclosure can also include the step of designing and / or constructing offshore wind turbines based on the environmental contour lines.

[0102] According to an embodiment of the present disclosure, the optimization design method may further include: performing a second environmental contour index evaluation on the first environmental contour result; in response to the second environmental contour index evaluation meeting the standard, using the first environmental contour result as the final environmental contour result.

[0103] The second environmental contour evaluation index may include a relative error between a maximum value of a marginal distribution variable based on environmental contour extrapolation and an extreme value obtained by an EVA method.

[0104] As an example, determine whether the relative errors of the maximum values ​​of the two variables extrapolated by the environmental contour and the extreme values ​​derived by the EVA method both meet the condition of RE < 1%. If so, the environmental contour result is valid; if not, the result is invalid.

[0105] If there is more than one environmental contour line result that meets the conditions, the result with the smallest relative error RE is selected as the optimal conclusion.

[0106] The requirements for the first and second environmental contour indicators can be described as follows:

[0107] Evaluation indicator 1: R of extreme events 2 Should be greater than 0.90;

[0108] Evaluation indicator 2: RMSE of extreme events should be less than 0.05;

[0109] Evaluation indicator 3 compliance conditions: If the time length of the input data series is less than 20% of the recurrence period length, the ELR of the extreme value part of the environmental contour line should be greater than 95%.

[0110] Evaluation indicator 4 compliance conditions: Under the same recurrence period, the RE of the maximum value of the variable extrapolated by the environmental contour line and the extreme value derived by the EVA method should be less than 1%.

[0111] The use of the above evaluation indicators can generate more accurate environmental contour lines, thereby obtaining more accurate specified joint return period environmental conditions, which can further reduce design and construction costs while ensuring structural safety, and is applicable to water resources management, flood risk assessment, urban planning and marine engineering.

[0112] In addition, the optimization design method disclosed in the present invention has high universality and is applicable not only to marine hydrological environment parameters, but also to multidimensional variables to be solved in other fields.

[0113] Figure 10 is a block diagram illustrating an apparatus for optimizing design of joint return periods according to the present disclosure.

[0114] The device 1000 for optimizing the joint return period according to an embodiment of the present disclosure includes a first calculation unit 1100 , a second calculation unit 1200 , an extreme event separation unit 1300 , a fitting unit 1400 , and an optimization unit 1500 .

[0115] The first calculation unit 1100 may obtain a multidimensional environmental variable data set of a target area and determine a joint return period to be solved. In addition, the first calculation unit 1100 may also determine an extreme value of the first variable under the joint return period.

[0116] The second calculation unit 1200 may determine the marginal distribution variable corresponding to the first set in the multidimensional environmental variable data set for the joint return period under the joint return period. As described above, the significant wave height may be determined as the marginal distribution variable.

[0117] The extreme event separation unit 1300 may sort the first set for the joint return period and divide the sorted first set into an extreme part and a non-extreme part based on extreme values ​​of marginal distribution variables for the joint return period.

[0118] Fitting unit 1400 may perform probability distribution fitting on the extreme portion and the non-extreme portion of the joint return period to obtain a first fitting result for the first set of joint return periods, and evaluate the first fitting result for the joint return period using the first environmental contour evaluation index. The fitting model and fitting method sampled by fitting unit 1400 may be as described above and are not further described here.

[0119] In response to the evaluation result of the first fitting result not meeting the criteria, the optimization unit 1500 may optimize the first fitting result for the joint return period to obtain a first fitting optimization result for the first set of joint return periods.

[0120] As an example, the optimization unit 1500 may determine the first fitting optimization result based on a connection point between the fitting result of the extreme part and the fitting result of the non-extreme part.

[0121] The optimization unit 1500 can fit the non-extreme part to obtain the non-extreme part fitting result, and fit the extreme part to obtain the extreme part fitting result, determine the connection point between the non-extreme part fitting result and the extreme part fitting result, and determine the first fitting optimization result of the first set based on the non-extreme part fitting result, the extreme part fitting result and the connection point.

[0122] In one example, the optimization unit 1500 may use a combination of the non-extreme portion fitting result before the connection point and the extreme portion fitting result after the connection point as the first fitting optimization result. The connection point is the point where the extreme portion fitting goodness of fit is the highest among the intersections of the non-extreme portion fitting result and the extreme portion fitting result.

[0123] In addition, the first calculation unit 1100 , the second calculation unit 1200 , the extreme event separation unit 1300 , and the fitting unit 1400 may perform the corresponding steps described above, which will not be described in detail here.

[0124] When the systems, units or modules shown in the drawings are implemented in software, firmware, middleware or microcode, the program code or code segments for performing the corresponding operations may be stored in a computer-readable medium such as a storage medium, so that at least one processor or at least one computing device may perform the corresponding operations by reading and running the corresponding program code or code segments. In addition, the computer-readable medium or storage medium may cause the processor to execute the above-mentioned optimization design method when the computer program is executed by the processor.

[0125] For example, according to an exemplary embodiment of the present disclosure, a computer device including a readable medium storing computer program instructions may be provided, wherein the instructions, when executed by at least one computing device, prompt the at least one computing device to perform at least one of the above steps.

[0126] The computing device according to an embodiment of the present disclosure includes a memory and a processor. The memory stores instructions or programs, and when the instructions or programs are executed by the processor, the processor is prompted to perform the above-mentioned optimization design method.

[0127] According to an embodiment of the present disclosure, a computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the processor is prompted to execute the above-mentioned optimization design method.

[0128] The instructions stored in the computer-readable storage medium can be executed in an environment deployed in a computer device such as a client, a host, an agent device, a server, etc. It should be noted that the instructions can also be used to perform additional steps in addition to the above steps or perform more specific processing when performing the above steps. The contents of these additional steps and further processing have been described in detail in the referenced embodiment. Figures 1 to 10 It is mentioned in the description of related systems and methods, so it will not be repeated here to avoid repetition.

[0129] It should be noted that the control method according to the exemplary embodiment of the present disclosure can completely rely on the operation of computer programs or instructions to realize the corresponding functions, that is, each device corresponds to each step in the functional architecture of the computer program, so that the entire system is called through a special software package (for example, lib library) to realize the corresponding function.

[0130] The optimization design method for the joint return period according to the embodiment of the present disclosure can make up for the shortcomings of the traditional IEC method in fitting the marginal distribution function for extreme events, improve the extrapolation accuracy of the marginal distribution function, and thus improve the accuracy of solving the joint working conditions by environmental contour lines.

[0131] The optimization design method for the joint recurrence period according to the embodiment of the present disclosure has high universality and is applicable not only to marine hydrological environmental parameters, but also to multidimensional environmental variables to be solved in other fields.

[0132] The optimization design method for the joint return period according to the embodiment of the present disclosure can generate more accurate environmental contour lines, thereby obtaining more accurate environmental conditions for the specified joint return period. It can further reduce design and construction costs while ensuring structural safety, and is of great significance in the fields of water resource management, flood risk assessment, urban planning, and marine engineering.

[0133] The above describes in detail the specific implementation methods of the present disclosure. Although some embodiments have been shown and described, those skilled in the art should understand that these embodiments can be combined, modified and improved without departing from the principles and spirit of the present disclosure, the scope of which is defined by the claims and their equivalents. These combinations, modifications and improvements should also be within the scope of protection of the present disclosure.

Claims

1. A method for optimizing the joint return period, characterized in that: The optimization design method comprises: Obtaining a multidimensional environmental variable data set for a target area and determining a joint return period to be solved; determining a marginal distribution variable corresponding to a first set of the multidimensional environmental variable data sets under the joint return period; sorting the first set and dividing the sorted first set into an extreme part and a non-extreme part based on the marginal distribution variable; Performing probability distribution fitting on the extreme part and the non-extreme part respectively to obtain a first fitting result of the first set, and evaluating the first fitting result using a first environmental contour evaluation index; In response to an evaluation result of the first fitting result not meeting a standard, optimizing the first fitting result to obtain a first fitting optimization result of the first set, Optimizing the first fitting result to obtain a first fitting optimization result of the first set includes: determining the first fitting optimization result based on a connection point between the fitting result of the extreme part and the fitting result of the non-extreme part, The step of determining the first fitting optimization result based on the connection point between the fitting result of the extreme part and the fitting result of the non-extreme part includes: Fitting the non-extreme part to obtain a non-extreme part fitting result, and fitting the extreme part to obtain an extreme part fitting result; Determining a connection point between the non-extreme part fitting result and the extreme part fitting result; Determine a first fitting optimization result of the first set according to the non-extreme part fitting result, the extreme part fitting result and the connection point, The step of determining a first fitting optimization result of the first set according to the non-extreme part fitting result, the extreme part fitting result, and the connection point comprises: A combination of the non-extreme part fitting result before the connection point and the extreme part fitting result after the connection point is used as the first fitting optimization result.

2. The optimization design method for joint return period according to claim 1, characterized in that: The connection point is a point where the extreme part fitting goodness of fit is the highest among the intersections of the non-extreme part fitting result and the extreme part fitting result.

3. The optimization design method for joint return period according to claim 1, characterized in that: The step of fitting the extreme part to obtain the extreme part fitting result includes: using lognormal distribution as a fitting model and fitting the extreme part by weighted least squares method, and the weight of the weighted least squares method is determined by genetic algorithm.

4. The optimization design method for joint return period according to claim 1, characterized in that: The optimization design method further includes: Dividing the sorted first set into a plurality of segments, and obtaining a conditional probability distribution of a second set in the multidimensional environmental variable data set within each segment; Based on the conditional probability distribution and the first fitting optimization result, first environmental contour results of marginal distribution variables corresponding to the first set and second variables corresponding to the second set under the joint recurrence period are solved.

5. The optimization design method for joint return period according to claim 4, characterized in that: The optimization design method further includes: Performing a second environmental contour index evaluation on the first environmental contour result; In response to the second environmental contour index evaluation being up to standard, the first environmental contour result is used as the final environmental contour result.

6. The optimization design method for joint return period according to claim 5, characterized in that: The first environmental contour line evaluation index includes at least one of the following indicators: the determination coefficient corresponding to the extreme part, the root mean square error corresponding to the extreme part, and the extreme value envelope rate of the environmental contour line; the second environmental contour line evaluation index includes the relative error between the maximum value of the marginal distribution variable based on the extrapolation of the environmental contour line and the extreme value obtained by the traditional extreme value statistical method.

7. The optimization design method for joint return period according to claim 1, characterized in that: The step of dividing the sorted first set into extreme parts and non-extreme parts based on the marginal distribution variables includes: calculating cumulative exceedance probabilities based on the marginal distribution variables, and dividing the first set into extreme parts and non-extreme parts according to the cumulative exceedance probabilities.

8. The optimization design method for joint return period according to claim 1, characterized in that: The multidimensional environmental variable data set is a multidimensional ocean environmental variable data set, and includes a significant wave height data set and a spectrum peak period data set, and the joint recurrence period is a two-dimensional variable joint recurrence period.

9. An optimization design device for joint recurrence period, characterized in that: The optimization design device comprises: A first computing unit acquires a multidimensional environmental variable data set of a target area and determines a joint recurrence period to be solved; a second computing unit, determining a marginal distribution variable corresponding to a first set in the multidimensional environmental variable data set under the joint return period; an extreme event separation unit, which sorts the first set and divides the sorted first set into an extreme part and a non-extreme part based on the marginal distribution variable; a fitting unit, performing probability distribution fitting on the extreme part and the non-extreme part respectively to obtain a first fitting result of the first set, and evaluating the first fitting result using a first environmental contour evaluation index; an optimization unit, in response to an evaluation result of the first fitting result not meeting a standard, optimizing the first fitting result to obtain a first fitting optimization result of the first set, The optimization unit determines the first fitting optimization result based on the connection point between the fitting result of the extreme part and the fitting result of the non-extreme part. wherein the optimization unit fits the non-extreme part to obtain a non-extreme part fitting result, and fits the extreme part to obtain an extreme part fitting result; determines a connection point between the non-extreme part fitting result and the extreme part fitting result; and determines a first fitting optimization result of the first set according to the non-extreme part fitting result, the extreme part fitting result, and the connection point. The optimization unit uses a combination of the non-extreme part fitting result before the connection point and the extreme part fitting result after the connection point as the first fitting optimization result.

10. A computing device, characterized in that: The device comprises a memory and a processor, wherein the memory stores instructions or programs and when the instructions or programs are executed by the processor, the processor is prompted to execute the optimization design method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed by a processor, causes the processor to execute the optimization design method according to any one of claims 1 to 8.

Images