Method for dynamically evaluating bow strength of icebreaker based on regression analysis

Abstract

The invention relates to the technical field of polar region ships, in particular to an icebreaker bow strength dynamic evaluation method based on regression analysis. According to the method, corresponding simulation databases are constructed on the basis of three ice condition scenes, the structural stress response extreme value and the measuring point stress extreme value of the icebreaker bow are set as safety evaluation and early warning indexes, and on the basis of simulation data of the icebreaker bow stress response, the safety evaluation and early warning of the icebreaker bow stress response are achieved. A regression analysis method is applied to optimization of a simulation fitting model, polynomial fitting optimization is carried out on a structural stress response extreme value and a measuring point stress extreme value, the fitting precision and effectiveness are greatly improved, an optimal polynomial mapping model of the structural stress response extreme value and the measuring point stress extreme value is obtained, and the structural stress response extreme value and the measuring point stress extreme value are optimized. Safety evaluation and early warning can be carried out more quickly, and the safety and high efficiency of the icebreaker working in the ice area are ensured.

Description

Technical Field

[0001] The present application relates to the technical field of polar ships, and in particular to a method for dynamically evaluating the strength of an icebreaker bow based on regression analysis. Background Art

[0002] With the gradual warming of the global climate and the melting of Arctic glaciers, countries around the world are actively exploring Arctic routes in order to shorten the transportation distance between Asian countries and European and American countries and reduce the cost of maritime transportation. However, ships sailing in the Arctic will inevitably encounter sea ice problems, and polar icebreakers play an important role in ensuring the continuity and smoothness of shipping, and are also indispensable for polar expeditions. During the polar navigation of icebreakers, the types, thickness, and forms of sea ice vary greatly in actual navigation, and the stress conditions of the bow structure are extremely complex. For this reason, studying ship-ice interactions plays an extremely important role in guiding polar ship navigation.

[0003] At present, the main numerical analysis method for studying ship-ice interaction is the finite element method (FEM). The finite element method has important applications in ship-ice interaction simulation, especially in hull structure strength analysis and dynamic response simulation. However, when dealing with complex ship-ice interaction problems, the finite element method cannot fully and quickly reflect the ice loads on the bow of the icebreaker during actual navigation. The parameter selection for the finite element simulation of ship-ice interaction is not comprehensive and accurate enough, and the fitting accuracy of the component simulation model is not high. It cannot accurately perform safety assessment and early warning, and cannot effectively ensure the safety and efficiency of icebreaker operations in ice areas. Summary of the invention

[0004] In view of the above technical problems in the prior art, the present invention provides a dynamic evaluation method for the bow strength of an icebreaker based on regression analysis, which is characterized by comprising the following steps: S1. Under a set ice condition scenario, construct a simulation database of stress response of the bow of an icebreaker, wherein the simulation database includes input variables and output variables, wherein the input variables include a first ice condition parameter variable and a second ice condition parameter variable in the set ice condition scenario, and the output variables include an extreme value of structural stress response of the bow of the icebreaker corresponding to the set ice condition scenario and an extreme value of stress at a measuring point; S2. Based on the regression analysis method, multiple fittings are performed on the input variables and the extreme values ​​of the structural stress response under the ice condition scenario set in the simulation database to obtain multiple structural response regression models, and the best fitting model is selected from the multiple structural response regression models to serve as the final structural response mapping model; S3, based on the regression analysis method, fitting the input variables and the extreme values ​​of the stress of the measuring points under the ice condition scenario set in the simulation database, obtaining multiple measuring point response regression models, and selecting the best fitting model from the multiple measuring point response regression models to serve as the final measuring point response mapping model; S4. Inputting the real-time parameters of ice conditions of the current route into the structural response mapping model and the measuring point response mapping model to dynamically evaluate the strength of the bow of the icebreaker.

[0005] Furthermore, the set ice condition scenario includes a layer ice scenario and an ice ridge scenario; in the layer ice scenario and the ice ridge scenario, the first ice condition parameter variable is the sea ice thickness, and the second ice condition parameter variable is the ship speed.

[0006] Furthermore, the set ice condition scenario also includes a crushed ice scenario; in the crushed ice scenario, the input variable also includes a third ice condition parameter variable, wherein the third ice condition parameter variable is sea ice density.

[0007] Further, the structural response regression model or the measuring point response regression model is a polynomial function fitting model, wherein in step S2 or step S3, selecting the best fitting model from the multiple structural response regression models or the multiple measuring point response regression models comprises the following steps: A1. Define variables: define the input variable as an independent variable, and define the output variable as a dependent variable; A2. Determine the cross-validation fold and fitting times: Based on the multivariate function variation relationship between the structural stress response extreme value or the measuring point stress extreme value and the input variable, set the cross-validation fold K value and the maximum number of polynomial fitting times; A3. Polynomial fitting optimization: According to the maximum number of polynomial fitting, different polynomial degrees are selected from low to high, the independent variables are respectively introduced into each polynomial for cross-fitting verification, and the best fitting model is selected from all fitting results.

[0008] Furthermore, in step S2 or step S3, before obtaining the structural response regression model or the measuring point response regression model, it also includes: fitting the functional change relationship between the structural stress response extreme value or the measuring point stress extreme value and each ice condition parameter variable under the set ice condition scenario, and using a cross-validation method to perform polynomial fitting optimization, so as to obtain the optimal polynomial equation and prediction curve of the structural stress response extreme value or the measuring point stress extreme value corresponding to each ice condition parameter variable, and then, integrate the functional relationship of each ice condition parameter variable to the structural stress response extreme value or the measuring point stress extreme value, so as to obtain the multivariate functional change relationship between the structural stress response extreme value or the measuring point stress extreme value and the input variable.

[0009] Furthermore, in the step A3, it further includes: for each polynomial fitting, the residual between the fitting value and the true value of each data point is calculated according to the current fitting surface, and the data point with the largest residual is dynamically eliminated; re-fitting is performed based on the data points after eliminating the residual, and then, the optimal fitting model is selected from the re-fitting results.

[0010] Furthermore, in the step S3, before inputting the real-time parameters of ice conditions of the current route into the structural response mapping model and the measuring point response mapping model, it also includes: verifying the validity of the structural response mapping model and the measuring point response mapping model, inputting ice condition data that meets the actual ship voyage into the structural response mapping model and the measuring point response mapping model, and the relative error between the real-time calculation result and the model calculation result does not exceed a set error threshold; if it exceeds the set error threshold, returning to the step A2, re-determining the value of the cross-validation fold K and the maximum number of polynomial fittings, and performing polynomial fitting optimization.

[0011] Furthermore, before the step S1, it also includes: selecting the actual icebreaker bow structure part to establish a numerical simulation model, and determining the main verification area of ​​the ship-ice collision; sorting out the physical and mechanical properties of sea ice, determining the main behaviors of sea ice in the ship-ice collision problem, selecting materials from the software material library to simulate sea ice, setting material parameters, combining additional failure criteria to improve the compressive failure characteristics, considering the strain rate effect, simulating the ice ball impact rigid wall test, and comparing with the pressure-area isoparametric curve, so as to obtain the sea ice material model; then, establishing the set ice condition scenario and performing ship-ice coupled finite element calculation.

[0012] Furthermore, when the ice condition scenario is set to a layer ice scenario, the structural response mapping model obtained by step S2 is: σ 1层 =899.9-7.796x-2309y+18.48xy+1971y 2 -7.029xy 2 -477.8y 3 Among them, σ 1层 is the extreme value of the structural stress response under the ice layer scenario, in MPa; x is the ship speed, in kn; y is the sea ice thickness, in m; When the ice condition scene is set to a crushed ice scene, the structural response mapping model obtained by step S2 is: σ 1碎 =-231.995+10.9312x+210.5y+166.15z where σ 1碎is the extreme value of structural stress in the broken ice scenario, unit: MPa; x is the ship speed, unit: kn; y is the sea ice thickness, unit: m; z is the sea ice density.

[0013] Furthermore, when the ice condition scenario is set to a layer ice scenario, the measurement point response mapping model obtained by step S3 is: σ 2层 =50.09-0.4997x-119.8y+1.209xy+99.6y 2 -0.3277xy 2 -23.58y 3 Among them, σ 2层 is the extreme stress value at the measuring point, in MPa; x is the ship speed, in kn; y is the ice thickness, in m.

[0014] The dynamic evaluation method for the strength of the bow of an icebreaker based on regression analysis provided by the present invention has the following beneficial effects: The present invention constructs corresponding simulation databases based on three ice condition scenarios (layer ice scenario, ice ridge scenario, and broken ice scenario), and achieves comprehensive coverage of typical ice condition scenarios; and different input variable parameters are set for each ice condition scenario, so as to establish a fitting model corresponding to each ice condition scenario, making the fitting model more refined and accurate; The present invention sets two output variables as safety assessment and early warning indicators, namely, the extreme value of the structural stress response of the bow of the icebreaker and the extreme value of the stress at the measuring point, so as to monitor and evaluate the structural stress of the bow as a whole, and monitor the stress from the local fixed position unit, and take the maximum stress extreme value as the most dangerous situation of the measuring point unit under the current working condition, so as to understand the influence of different ice conditions on the fixed position structure; therefore, the present invention combines the extreme value of the structural stress response and the extreme value of the stress at the measuring point for comprehensive evaluation, and can achieve a more comprehensive and accurate evaluation compared with a single evaluation indicator; Based on the simulation data of the icebreaker bow stress response, the present invention applies the regression analysis method to the optimization of the simulation fitting model, performs polynomial fitting optimization on the structural stress response extreme values ​​and the measuring point stress extreme values, and obtains the optimal polynomial mapping model of the structural stress response extreme values ​​and the measuring point stress extreme values. Compared with the existing bow stress response fitting model, the fitting accuracy and effectiveness are greatly improved, so that safety assessment and early warning can be performed more quickly and accurately, ensuring the safety and efficiency of the icebreaker operating in the ice area. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is the installation position of the actual ship ice load measurement point in the embodiment of the present invention; Figure 2A comparison diagram of pressure-area curves of different materials obtained by simulating the sea ice model in an embodiment of the present invention; Figure 3 A numerical simulation collision model for a layer of ice condition scenario in an embodiment of the present invention; Figure 4 It is the numerical simulation of ice layer destruction of the ice layer scene in the embodiment of the present invention; Figure 5 Schematic diagram of the numerical simulation collision model of the ice ridge ice scene in the embodiment of the present invention Figure 1 ; Figure 6 Schematic diagram of the numerical simulation collision model of the ice ridge ice scene in the embodiment of the present invention Figure 2 ; Figure 7 Schematic diagram of the numerical simulation collision model of the crushed ice scene in the embodiment of the present invention Figure 1 ; Figure 8 Schematic diagram of the numerical simulation collision model of the crushed ice scene in the embodiment of the present invention Figure 2 ; Fig. 9 The extreme value changes of structural response stress of each density working condition with the ship speed when the ice thickness is 0.5m in the broken ice scenario in the embodiment of the present invention; Fig.10 The extreme value changes of structural response stress of each density working condition with the ship speed when the ice thickness is 0.8m in the broken ice scenario in the embodiment of the present invention; Fig.11 The extreme value changes of structural response stress of each density condition with the speed at 8kn in the broken ice scenario in the embodiment of the present invention; Fig.12 The structural response stress extreme value changes of each density condition with the speed at 12kn in the broken ice scenario in the embodiment of the present invention; Fig.13 It is a schematic diagram of a bow structure response mapping model for a layer ice condition scenario in an embodiment of the present invention; Fig.14 It is a schematic diagram of a bow measurement point response mapping model and measured data points in a layer ice scenario in an embodiment of the present invention; Fig.15 It is a schematic diagram of a bow structure response mapping model and a warning value plane in a layer ice condition scenario in an embodiment of the present invention. DETAILED DESCRIPTION

[0016] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0017] In the existing technology, when the finite element method is used to deal with complex ship-ice interaction problems, it is unable to comprehensively and quickly reflect the ice loads exerted on the bow of an icebreaker during actual navigation. The parameter selection for the finite element simulation of ship-ice interaction is not comprehensive and accurate enough, and the fitting accuracy of the component simulation model is not high. It is also unable to accurately perform safety assessment and early warning, and is unable to effectively ensure the safety and efficiency of icebreaker operations in ice areas.

[0018] In view of the above technical problems, this embodiment proposes a dynamic evaluation method for the strength of the bow of an icebreaker based on regression analysis, which mainly includes the following steps: S1. Under a set ice condition scenario, construct a simulation database of stress response of the bow of an icebreaker, wherein the simulation database includes input variables and output variables, wherein the input variables include a first ice condition parameter variable and a second ice condition parameter variable in the set ice condition scenario, and the output variables include an extreme value of structural stress response of the bow of the icebreaker corresponding to the set ice condition scenario and an extreme value of stress at a measuring point; S2. Based on the regression analysis method, multiple fittings are performed on the input variables and the extreme values ​​of the structural stress response under the ice condition scenario set in the simulation database to obtain multiple structural response regression models, and the best fitting model is selected from the multiple structural response regression models to serve as the final structural response mapping model; S3, based on the regression analysis method, fitting the input variables and the extreme values ​​of the stress of the measuring points under the ice condition scenario set in the simulation database, obtaining multiple measuring point response regression models, and selecting the best fitting model from the multiple measuring point response regression models to serve as the final measuring point response mapping model; S4. Inputting the real-time parameters of ice conditions of the current route into the structural response mapping model and the measuring point response mapping model to dynamically evaluate the strength of the bow of the icebreaker.

[0019] It should be noted that the measuring points refer to the test points set at the bow of the actual ship to measure the ice load, see Figure 1 , Figure 1The figure shows the installation position of the actual ship ice load measuring point, where Z represents the height of the measuring point from the ship bottom baseline, in meters; Fr represents the hull rib position, indicating the sequence number from the stern to the bow; the ice load measuring point is installed in the ice belt area on the side of the icebreaker where the bow and shoulder transition.

[0020] Through the above steps S1-S4, two output variables are set as safety assessment and early warning indicators, namely, the extreme value of the structural stress response of the bow of the icebreaker and the extreme value of the stress at the measuring point, so that the structural stress of the bow as a whole is monitored and evaluated, and the stress is monitored from the local fixed position unit. Compared with a single evaluation indicator, this embodiment can achieve a more comprehensive and accurate evaluation; and, based on the simulation data of the stress response of the bow of the icebreaker, this embodiment applies the regression analysis method to the optimization of the simulation fitting model, fits and optimizes the extreme value of the structural stress response and the extreme value of the stress at the measuring point, and obtains the optimal mapping model of the extreme value of the structural stress response and the extreme value of the stress at the measuring point. Compared with the existing bow stress response fitting model, the fitting accuracy and effectiveness are greatly improved, so that the safety assessment and early warning can be performed more accurately, ensuring the safety and efficiency of the icebreaker operating in the ice area.

[0021] Furthermore, in order to establish a stress response simulation database under a set ice condition scenario, before step S1, it also includes: selecting an actual icebreaker bow structure part to establish a numerical simulation model, and determining the main verification area of ​​the ship-ice collision; sorting out the physical and mechanical properties of sea ice, determining the main behaviors of sea ice in the ship-ice collision problem, selecting materials from the software material library to simulate sea ice, setting material parameters, and improving the compressive damage characteristics in combination with additional damage criteria, considering the strain rate effect, simulating the ice ball impact on the rigid wall test, and comparing with the pressure-area isoparametric curve (i.e., ISO curve), so as to obtain the sea ice material model; then, establishing the set ice condition scenario and performing ship-ice coupled finite element calculation.

[0022] Specifically, see Figure 2 In the process of obtaining the sea ice material model, materials No. 13, No. 63 and No. 124 in the LS-DYNA software material library were used to simulate sea ice, and the material parameters were reasonably set. The compressive failure characteristics were improved in combination with the additional failure criterion. The strain rate effect was considered, and the ice ball impact on the rigid wall test was simulated. Compared with the pressure-area isoparametric curve (i.e., ISO curve), the rationality of the isotropic elastic fracture failure model of material No. 13 (MAT13) was verified, and thus material No. 13 was selected as the final sea ice material model.

[0023] In the analysis of ship-ice interaction under set ice conditions, multiple ship-ice coupling finite element calculations were established based on the set ice conditions. The measured sensor data of navigation in ice areas was compared with the finite element simulation results, and the feasibility of each ship-ice coupling method was analyzed until the simulated values ​​of ice-induced stress at each measuring point were consistent with the measured data, thereby constructing a simulation model and simulation database for the corresponding ice condition scenario.

[0024] Furthermore, the set ice condition scenario includes a layer ice scenario and an ice ridge scenario; in the layer ice scenario and the ice ridge scenario, the first ice condition parameter variable is the sea ice thickness, and the second ice condition parameter variable is the ship speed.

[0025] Furthermore, the set ice condition scenario also includes a crushed ice scenario; in the crushed ice scenario, the input variable also includes a third ice condition parameter variable, wherein the third ice condition parameter variable is sea ice density.

[0026] That is to say, this embodiment constructs corresponding simulation databases based on three ice condition scenarios (layer ice scenario, ice ridge scenario, and broken ice scenario), thereby achieving comprehensive coverage of typical ice condition scenarios; and, for each ice condition scenario, different input variable parameters are set respectively, so as to establish a fitting model corresponding to each ice condition scenario, making the fitting model more refined and accurate.

[0027] Specifically, see Figure 3-Figure 4 In the analysis of ship-ice interaction under layered ice conditions, the effects of ice thickness and speed are mainly considered, and continuous layered ice conditions are established for ship-ice coupling finite element calculations. The analysis found that when the ice thickness is small or the speed is high, the main failure mode of the layered ice is splitting damage, showing large radial cracks; when the ice thickness is large, the main failure mode is crushing and bending, with obvious circumferential cracks. The simulation phenomenon is consistent with the actual situation.

[0028] See also Figure 5-Figure 6 In the analysis of ship-ice interaction under ice ridge conditions, the influence of ice thickness and speed is mainly considered, and a continuous icebreaker ridge condition is established for ship-ice coupling finite element calculation. The analysis found that the main failure modes of ice ridges are crushing and bending, and obvious circumferential cracks are generated. The stress extreme value time history under this condition shows a state of first rising and then stable fluctuation. The area where the stress extreme value appears is mainly concentrated on the bottom ribs of the bow, and the stress extreme value also shows a trend of increasing with the increase of ice thickness and speed. Under the collision and extrusion of ice ridges, the stress on the bow column and both sides of the bow column is always the largest, that is, at the waterline where it contacts the ice layer.

[0029] See also Figure 7-Figure 8, mainly considering the influence of ice thickness, density and speed, and establishing the collision crushed ice field condition for fluid-solid coupling ALE method calculation. The fluid-solid coupling ALE method calculates the influence of fluid on the movement of ice. In the dynamic simulation of the bow and crushed ice, the crushed ice showed sliding, flipping, accumulation, sinking and other phenomena. Because of the sliding phenomenon of crushed ice along the hull outer plate, the extreme value of the ship-ice collision force and the area with high stress are no longer concentrated in the side area of ​​the front bottom and the bow shoulder, and the bottom structure of the bow is also an important load-bearing part.

[0030] Furthermore, in step S1, the input variables are determined to be the sea ice thickness, the sea ice density (the parameters under the ice breaking condition also include the sea ice density), and the speed of the icebreaker under the continuous icebreaking state; the output variables are the maximum stress value of the ice-resistant component of the bow of the icebreaker and the stress value of the measuring point unit.

[0031] It should be noted that after determining the input variables and output variables of each working condition, the multi-condition simulation sample space is set, and numerical simulation calculations are performed for each working condition. The input and output values ​​of each working condition are sorted out according to the stress extreme value reading method, and the read ship-ice coupling simulation results are standardized. The CSV result file is compiled into the database, thereby forming a ship-ice coupling simulation database. The specific steps are as follows: a1. Read CSV file: Use appropriate functions in MATLAB to read the data in the CSV file. Make sure the format of the CSV file is correct, including the meaning and data type of the columns.

[0032] a2. Connect to the database: Use MATLAB's database toolbox to connect to the generated database. You can choose to use SQLite, MySQL, PostgreSQL and other databases.

[0033] a3. Create a table: If there is no corresponding table in the database to store simulation data, you need to create a new table. You can use SQL statements to create a new table in the database.

[0034] a4. Insert data: Insert the data in the CSV file into the table of the database. This involves converting the data structure in MATLAB into a form acceptable to the database and finally performing the insertion operation.

[0035] Specifically in this embodiment, a CSV file containing simulation data for each operating condition is used, including fields such as speed, ice thickness, and stress output. First, use the readtable function to read the CSV file as a MATLAB table. Then, specify the path of the SQLite database file, build a JDBC connection string, connect to the SQLite database, and execute an SQL command to create a table named simulation_data. Finally, convert the MATLAB table to a cell array and insert the data into the database table, which can be expanded continuously in the future.

[0036] Furthermore, in step S2 and step S3, the input and output data in the simulation database are fitted using a regression analysis method, and the optimal combination of parameters and coefficients is determined in the mathematical model equation, so that the regression model finally reaches the optimal evaluation index. Since each ice condition scenario corresponds to more than two input variables (i.e., sea ice thickness, sea ice density, and ship speed), in order to obtain a more accurate structural response mapping model and a measuring point response mapping model, it is necessary to accurately determine the multivariate function change relationship between the structural stress response extreme value or the measuring point stress extreme value and the input variables. Therefore, before obtaining the multivariate function change relationship, it is particularly important to accurately determine the functional change relationship between the structural stress response extreme value or the measuring point stress extreme value and each ice condition parameter variable. For this reason, it is necessary to fit and optimize the functional change relationship between the structural stress response extreme value or the measuring point stress extreme value and each ice condition parameter variable.

[0037] Specifically, taking the layer ice condition as an example, in order to explore the functional relationship between the extreme stress value and the thickness of sea ice, the mapping equation is established by using polynomial fitting for the working conditions under different speeds. The main steps are: b1. Define several different sets of stress dependent variables and a set of independent variables, sea ice thickness; b2. Set the maximum number of polynomial fitting searches and use the cross-validation method to select the optimal polynomial fitting model to prevent overfitting; b3. In each run, different fitting times are selected from low to high, and polynomial fitting is performed cyclically. The root mean square error (RMSE), mean absolute error (MAE) and coefficient of determination (R2) are calculated based on the fitting results. The optimal fitting model and its performance evaluation index are recorded each time. b4. Select the fitting model with the smallest root mean square error from all runs as the final result, and display its polynomial order, coefficient, RMSE, MAE and R2 value; b5. Displays the prediction curve of stress extreme value for different ice thickness at various speeds. This prediction curve is the final fitting curve of the fitting optimization.

[0038] In addition, taking the fitting process of the extreme value of the structural stress response under broken ice conditions as an example, see Table 1, which shows the input and output data of the extreme value of the structural stress response with the change of sea ice thickness, speed, and sea ice density. According to these input and output data, the extreme value curves of the bow structural stress with the change of speed or ice thickness under different ice density conditions are drawn respectively; combined with Figure 9-12 The analysis of stress changes shows that the stress extremes of the bow structure show an overall upward trend as the ice thickness and speed change under various density conditions. At the same time, the stress results of each numerical analysis in each figure also increase with the increase of density. A small number of conditions do not meet the overall trend. This is because in the crushed ice condition, the extreme values ​​of the ship-ice collision force and the areas with high stress are no longer concentrated in the side areas of the front bottom and the bow shoulder, resulting in differences in stress extremes. Furthermore, based on the numerical simulation data of these crushed ice conditions, a polynomial response model covering the three inputs of speed, ice thickness and density and the output of the extreme value of the bow structure stress is established.

[0039] Table 1 Input and output data of ice crushing condition Working conditions Sea ice thickness / m Speed / kn Density Stress extreme value / MPa Condition 1 0.5 8 40% 50.6 Condition 2 0.5 12 40% 74.1 Condition 3 0.8 8 40% 75.4 Condition 4 0.8 12 40% 119.6 Condition 5 0.5 8 50% 67.1 Condition 6 0.5 12 50% 76.7 Condition 7 0.8 8 50% 71.8 Condition 8 0.8 12 50% 124.7 Condition 9 0.5 8 60% 81.4 Condition 10 0.5 12 60% 78.2 Condition 11 0.8 8 60% 144.4 Condition 12 0.8 12 60% 256.6 Condition 13 0.5 8 70% 86.7 Condition 14 0.5 12 70% 76.8 Condition 15 0.8 8 70% 91.9 Condition 16 0.8 12 70% 212.4 Taking the layered ice condition as an example again, in order to explore the functional relationship between the extreme stress value and the speed, a polynomial fitting is used to establish a mapping equation for the working results under different ice thicknesses. However, when the ice thickness is large, the scatter line graphs under different speed conditions have a large degree of inflection points. If a polynomial fitting is performed on all the working results, overfitting or poor fitting results will inevitably occur. In order to avoid such situations, and in order to meet the simple and fast characteristics of the engineering application model, the method of eliminating abnormal points is adopted when optimizing the polynomial function. The main steps are: c1. Define several groups of different stress dependent variables and a group of independent variable speed respectively; c2. Set the maximum number of polynomial fitting searches and use the cross-validation method to select the optimal polynomial fitting model to prevent overfitting; c3. In each run, different fitting times are selected from low to high, and polynomial fitting is performed cyclically; in each optimal fitting process, the residual between the predicted value and the true value of each data point is calculated according to the current fitting curve; c4. Dynamically remove the data points with the largest residuals from the current curve, and then analyze the three evaluation indicators of root mean square error (RMSE), mean absolute error (MAE) and determination coefficient (R2), and select the fitting model with the smallest root mean square error (RMSE) from all runs as the final result; c5. Display the prediction curve of stress extreme value for different speeds under different ice thicknesses, and mark the excluded abnormal values. The prediction curve is the final fitting curve of the fitting optimization.

[0040] Therefore, the above-mentioned cross-validation fitting optimization method is used to perform polynomial fitting optimization, so as to obtain the optimal polynomial equation and prediction curve corresponding to each ice condition parameter variable of the structural stress response extreme value or the measuring point stress extreme value, and then, the functional relationship of each ice condition parameter variable to the structural stress response extreme value or the measuring point stress extreme value is integrated, so as to obtain the multivariate function change relationship between the structural stress response extreme value or the measuring point stress extreme value and the input variable, obtain multiple structural response regression models or multiple measuring point response regression models, and use the cross-validation fitting optimization method to select the best fitting model from the multiple structural response regression models or multiple measuring point response regression models, so as to serve as the final structural response mapping model or measuring point response mapping model, which specifically includes the following steps: A1. Define variables: define the input variable as an independent variable, and define the output variable as a dependent variable; A2. Determine the cross-validation fold and fitting times: Based on the multivariate function variation relationship between the structural stress response extreme value or the measuring point stress extreme value and the input variable, set the cross-validation fold K value and the maximum number of polynomial fitting times; A3. Polynomial fitting optimization: According to the maximum number of polynomial fitting, different polynomial degrees are selected from low to high, the independent variables are respectively introduced into each polynomial for cross-fitting verification, and the best fitting model is selected from all fitting results.

[0041] In the step A3, a method for eliminating abnormal points is further included: for each polynomial fitting, the residual between the fitting value and the true value of each data point is calculated according to the current fitting surface, and the data point with the largest residual is dynamically eliminated; re-fitting is performed based on the data points after eliminating the residual, and then the optimal fitting model is selected from the re-fitting results.

[0042] Furthermore, according to the above method, the structural response mapping model of the extreme values ​​of the structural stress response with the three ice condition scenarios and the measuring point response mapping model of the extreme values ​​of the measuring point stress with the three ice condition scenarios were finally obtained.

[0043] Specifically, taking the extreme value of the structural response stress under the crushed ice condition as an example, the optimal polynomial equation in the crushed ice condition is finally obtained, that is, the structural response mapping model under the crushed ice condition is: σ 1碎 =-231.995+10.9312x+210.5y+166.15z where σ 1碎 is the extreme value of structural stress under the broken ice scenario, in MPa; x is the ship speed, in kN; y is the sea ice thickness, in m; z is the sea ice density; the confidence limit of each coefficient is 95%.

[0044] The corresponding evaluation indicators are shown in Table 2: Table 2 Evaluation indicators of the structural response mapping model under broken ice conditions Evaluation indicators Root Mean Square Error (RMSE) Mean Absolute Error (MAE) Coefficient of determination (R2) Numeric 33.77 27.42 0.615 Taking the extreme value of the structural response stress under ice layer conditions as an example, the optimal polynomial equation in the ice layer condition is finally obtained, that is, the structural response mapping model under ice layer conditions is: σ 1层 =899.9-7.796x-2309y+18.48xy+1971y 2 -7.029xy 2 -477.8y 3 Among them, σ 1层 is the extreme value of the structural stress response under the layer ice scenario, in MPa; x is the ship speed, in kn; y is the sea ice thickness, in m; the confidence limit of each coefficient is 95%.

[0045] Taking the extreme stress value of the measuring point under the layer ice condition as an example, the optimal polynomial equation in the layer ice condition is finally obtained, that is, the measuring point response mapping model under the layer ice condition is: σ 2层 =50.09-0.4997x-119.8y+1.209xy+99.6y 2 -0.3277xy 2 -23.58y 3 Among them, σ 2层 is the extreme stress value at the measuring point, in MPa; x is the ship speed, in kn; y is the ice thickness, in m; the confidence limit of each coefficient is 95%.

[0046] like Fig.13 As shown in the figure, it is an extreme value polynomial model of the bow structure stress response under layered ice conditions. The corresponding root mean square error (RMSE) and mean absolute error (MAE) are small, and the determination coefficient is higher than 0.9, indicating that the model can well explain the variance of the data and has a good fitting effect.

[0047] Furthermore, after obtaining the structural response mapping model and the measuring point response mapping model, it also includes: verifying the validity of the structural response mapping model and the measuring point response mapping model, inputting ice condition data that meets the actual ship voyage into the structural response mapping model and the measuring point response mapping model, and the relative error between the real-time calculation result and the model calculation result does not exceed the set error threshold; if it exceeds the set error threshold, returning to the step A2, re-determining the value of the cross-validation fold K, and the maximum number of polynomial fitting, and performing polynomial fitting optimization. Optionally, the set error threshold is 5%.

[0048] like Fig.14 As shown in the figure, it is the stress extreme value polynomial model of the measuring point under layer ice conditions and the measured data points. The measured data in the figure are relatively evenly distributed on both sides of the response surface, which proves that the mathematical model can better reflect the mean level of the measured data and can predict the stress trend with the change of working conditions, thereby proving the effectiveness of the mapping model establishment method and providing reliable analysis for stress prediction of actual voyages.

[0049] Furthermore, after verifying the effectiveness of the structural response mapping model and the measuring point response mapping model, the response mapping model and the measuring point response mapping model can be applied to actual evaluation and early warning. The real-time parameters of the ice conditions on the current route are input into the structural response mapping model and the measuring point response mapping model to dynamically evaluate the strength of the bow of the icebreaker, thereby realizing real-time evaluation and short-term forecasting of the ship's structural strength.

[0050] Furthermore, this embodiment sets a dangerous stress warning value. After the ice condition parameters are input into the mapping model, if the output stress extreme value is higher than the dangerous stress warning value, it means that the set structural dangerous state has been reached, and warning measures will be taken at this time; if it is lower than the dangerous stress warning value, it means that the ice condition ahead is not enough to have a significant impact on the structure, and normal navigation can be carried out. Optionally, the dangerous stress warning value is set to 80% of the threshold value of 355MPa, that is, 284MPa. Fig.15 As shown in the figure, the Z plane represents the dangerous stress warning value plane, thus realizing the establishment of a structural safety assessment and warning indicator parameter system, providing important support for the navigation safety and design basis of polar ships.

[0051] Although the present invention has been described in detail by referring to the accompanying drawings and in combination with the preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, a person of ordinary skill in the art may make various equivalent modifications or substitutions to the embodiments of the present invention, and these modifications or substitutions shall be within the scope of the present invention or any person skilled in the art who is familiar with the present invention may easily think of changes or substitutions within the technical scope disclosed by the present invention, and shall be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A dynamic evaluation method for the strength of an icebreaker bow based on regression analysis, characterized in that: The steps include: S1. Under a set ice condition scenario, construct a simulation database of stress response of the bow of an icebreaker, wherein the simulation database includes input variables and output variables, wherein the input variables include a first ice condition parameter variable and a second ice condition parameter variable in the set ice condition scenario, and the output variables include an extreme value of structural stress response of the bow of the icebreaker corresponding to the set ice condition scenario and an extreme value of stress at a measuring point; S2. Based on the regression analysis method, multiple fittings are performed on the input variables and the extreme values ​​of the structural stress response under the ice condition scenario set in the simulation database to obtain multiple structural response regression models, and the best fitting model is selected from the multiple structural response regression models to serve as the final structural response mapping model; S3, based on the regression analysis method, fitting the input variables and the extreme values ​​of the stress of the measuring points under the ice condition scenario set in the simulation database, obtaining multiple measuring point response regression models, and selecting the best fitting model from the multiple measuring point response regression models to serve as the final measuring point response mapping model; S4. Inputting the real-time parameters of ice conditions of the current route into the structural response mapping model and the measuring point response mapping model to dynamically evaluate the strength of the bow of the icebreaker.

2. The dynamic evaluation method for the strength of the bow of an icebreaker based on regression analysis according to claim 1 is characterized in that: The set ice condition scenario includes a layer ice scenario and an ice ridge scenario; in the layer ice scenario and the ice ridge scenario, the first ice condition parameter variable is the sea ice thickness, and the second ice condition parameter variable is the ship speed.

3. The dynamic evaluation method for the strength of the bow of an icebreaker based on regression analysis according to claim 2 is characterized in that: The set ice condition scenario also includes a crushed ice scenario; in the crushed ice scenario, the input variable also includes a third ice condition parameter variable, wherein the third ice condition parameter variable is sea ice density.

4. The dynamic evaluation method for the strength of the bow of an icebreaker based on regression analysis according to claim 3 is characterized in that: The structural response regression model or the measuring point response regression model is a polynomial function fitting model, wherein in step S2 or step S3, selecting the best fitting model from the multiple structural response regression models or the multiple measuring point response regression models comprises the following steps: A1. Define variables: define the input variable as an independent variable, and define the output variable as a dependent variable; A2. Determine the cross-validation fold and fitting times: Based on the multivariate function variation relationship between the structural stress response extreme value or the measuring point stress extreme value and the input variable, set the cross-validation fold K value and the maximum number of polynomial fitting times; A3. Polynomial fitting optimization: According to the maximum number of polynomial fitting, different polynomial degrees are selected from low to high, the independent variables are respectively introduced into each polynomial for cross-fitting verification, and the best fitting model is selected from all fitting results.

5. The dynamic evaluation method for the strength of the bow of an icebreaker based on regression analysis according to claim 4 is characterized in that: In step S2 or step S3, before obtaining the structural response regression model or the measuring point response regression model, it also includes: fitting the functional change relationship between the structural stress response extreme value or the measuring point stress extreme value and the various ice condition parameter variables under the set ice condition scenario, and using a cross-validation method to perform polynomial fitting optimization, so as to obtain the optimal polynomial equation and prediction curve of the structural stress response extreme value or the measuring point stress extreme value corresponding to each ice condition parameter variable, and then, integrating the functional relationship of each ice condition parameter variable to the structural stress response extreme value or the measuring point stress extreme value, so as to obtain the multivariate functional change relationship between the structural stress response extreme value or the measuring point stress extreme value and the input variable.

6. The dynamic evaluation method for the strength of the bow of an icebreaker based on regression analysis according to claim 5 is characterized in that: In the step A3, it further includes: for each polynomial fitting, the residual between the fitting value and the true value of each data point is calculated according to the current fitting surface, and the data point with the largest residual is dynamically eliminated; re-fitting is performed based on the data points after eliminating the residual, and then, the optimal fitting model is selected from the re-fitting results.

7. The dynamic evaluation method for the strength of an icebreaker bow based on regression analysis according to claim 6 is characterized in that: In the step S3, before inputting the real-time parameters of ice conditions of the current route into the structural response mapping model and the measuring point response mapping model, it also includes: verifying the validity of the structural response mapping model and the measuring point response mapping model, inputting ice condition data that meets the actual ship voyage into the structural response mapping model and the measuring point response mapping model, and the relative error between the real-time calculation result and the model calculation result does not exceed the set error threshold; if it exceeds the set error threshold, returning to the step A2, re-determining the value of the cross-validation fold K and the maximum number of polynomial fittings, and performing polynomial fitting optimization.

8. The dynamic evaluation method for the strength of the bow of an icebreaker based on regression analysis according to claim 7 is characterized in that: Before the step S1, it also includes: selecting the actual icebreaker bow structure part to establish a numerical simulation model, and determining the main verification area of ​​the ship-ice collision; sorting out the physical and mechanical properties of sea ice, determining the main behaviors of sea ice in the ship-ice collision problem, selecting materials from the software material library to simulate sea ice, setting material parameters, combining additional failure criteria to improve the compressive failure characteristics, considering the strain rate effect, simulating the ice ball impact rigid wall test, and comparing with the pressure-area isoparametric curve, so as to obtain the sea ice material model; then, establishing the set ice condition scenario and performing ship-ice coupled finite element calculation.

9. The dynamic evaluation method for the strength of the bow of an icebreaker based on regression analysis according to claim 8 is characterized in that: When the ice condition scenario is set to a layer ice scenario, the structural response mapping model obtained by step S2 is: σ 1层 =899.9-7.796x-2309y+18.48xy+1971y 2 -7.029xy 2 -477.8y 3 ; Among them, σ 1层 is the extreme value of the structural stress response under the ice layer scenario, in MPa; x is the ship speed, in kn; y is the sea ice thickness, in m; When the ice condition scene is set to a crushed ice scene, the structural response mapping model obtained by step S2 is: s 1碎 =-231.995+10.9312x+210.5y+166.15z; where σ 1碎 is the extreme value of structural stress in the broken ice scenario, unit: MPa; x is the ship speed, unit: kn; y is the sea ice thickness, unit: m; z is the sea ice density.

10. The dynamic evaluation method for the strength of an icebreaker bow based on regression analysis according to claim 9 is characterized in that: When the ice condition scenario is set to a layer ice scenario, the measurement point response mapping model obtained by step S3 is: σ 2层 =50.09-0.4997x-119.8y+1.209xy+99.6y 2 -0.3277xy 2 -23.58y 3 ; Among them, σ 2层 is the extreme stress value at the measuring point, in MPa; x is the ship speed, in kn; y is the ice thickness, in m.

Images