A method for rapid optimization of ship mid-section structural dimensions

Abstract

The invention relates to a method for fast optimization of the structural dimensions of a mid-section of a ship. Firstly, a scheme in which all design variables take lower limit values ​​is selected as an initial scheme, a characteristic quantity with the largest degree of constraint violation of each type is determined, and an adjustment order of the design variables is determined according to a sensitivity analysis result of the characteristic quantity to the design variables; at the same time, the values ​​of the corresponding most sensitive design variables are adjusted upward in sequence until all constraints meet the design requirements; then, according to the descending order of the influence of the design variables on the objective function, a small range adjustment of the optimization scheme based on the coordinate rotation method is performed, a single design variable is taken in sequence, the values ​​of the other design variables are kept unchanged, the values ​​of the design variables are gradually reduced, and the feasibility of the scheme is calculated by using a specification checking software until the current scheme stops satisfying the constraints; finally, the value of the design variable is increased by a step length, and the value is used as the final value scheme of the design variable, and the design scheme at this time is the optimal scheme obtained.

Description

Technical Field

[0001] The invention relates to a ship mid-transverse section structure design method, in particular to a method for quickly optimizing the size of a ship mid-transverse section component under a determined frame spacing. Background Art

[0002] There are many design parameters for the mid-section structure of a ship. The optimization of the mid-section structure of a ship under the requirements of the regulations involves more complex constraints. Most existing related studies use intelligent optimization algorithms to directly embed the specification verification software for optimization calculation. Wang Yuan et al. [1] and Xu Puning et al. [2] both used the Isight platform to integrate the Mars2000 software and realized the optimization of the mid-section of a ship with 20 design variables under the CSR specification and the optimization of the mid-section of a ship with 51 design variables under the HCSR specification based on the simulated annealing algorithm. The former obtained a feasible mid-section solution under the premise that the original solution was not feasible, and the section area was reduced by 0.07%; the latter obtained an optimized solution with a section area reduction of 6.4% after 10,000 iterations. Feng Guoqing et al. [3] used Isight to integrate Excel and adopted the multi-island genetic algorithm, particle swarm algorithm, and simulated annealing algorithm to realize the optimization of the mid-section of 40 variables under the requirements of the HCSR specification. The section areas of the optimized solutions were reduced by 0.93%, 1.02%, and 1.35%, respectively. Liu Lang et al. [4] used MATLAB integrated with Excel and adopted an artificial bee colony algorithm to achieve cross-section optimization of 295 variables under the HCSR specification. After approximately 5,000 specification checks, the cross-section area of ​​the optimized solution decreased by approximately 5.1%.

[0003] Although the calculation time for a single mid-section check by standard checking software such as Mars2000 is relatively short, generally a few seconds to more than ten seconds, due to the large number of design variables in mid-section optimization, the number of scheme checks required by the intelligent optimization algorithm generally reaches 10. 4 ~10 5The total calculation time for a single optimization is several days, and the amount of calculation is still not to be underestimated. Establishing a proxy model for the midsection[5] to replace the standard software calculation in the optimization process is a relatively feasible way to reduce the amount of calculation. However, due to the large number of design variables and complex constraints in the midsection optimization, it is also difficult to establish a reliable proxy model[6]. Guo Xiaodong et al.[7] used the improved relative difference quotient method to optimize the midsection of a large oil tanker with 20 variables under the DNV specification. After 489 iterations, an optimization solution that met the constraints was obtained, and the weight was reduced by 5.54%. However, each iteration required 20 midsection checks (the number of design variables) to calculate the relative difference quotient under each design variable, and the total amount of calculation was still high. Chen Jing et al.[8] combined the characteristics of the cross-section strength constraint and adopted the idea of ​​hierarchical optimization, first considering the local constraint and then the total longitudinal strength constraint, and combined with the hull beam theory to adjust the variables according to the distance of the component from the neutral axis, making good use of the characteristics of the cross-section strength constraint, but the types and number of constraints considered were relatively small.

[0004] [1] Wang Yuan, Wu Jiameng. Optimization of the structural dimensions of the mid-section of an oil tanker based on the Excel-Mars2000-Isight platform [J]. Ships, 2019, 030(001): 119-127.

[0005] [2] Xu Puning, Zhang Qi, Bai Zekun. Optimization design of mid-section of oil tanker based on standard calculation [J]. Shipbuilding Technology, 2020, (06): 16-22.

[0006] [3] Feng Guoqing, Chang Qi, Wang Yuan, et al. Optimization of mid-section structure of large oil tankers under complex constraints [J]. Journal of Huazhong University of Science and Technology (Natural Science Edition), 2019, 47(10): 75-81.

[0007] [4] Liu Lang, Xia Lijuan. Optimization design of oil tanker midship section based on artificial bee colony algorithm [J]. Ship Engineering, 2019, 267(05): 9-12+62.

[0008] [5] Zhao Liuping, Zhan Dawei, Cheng Yuansheng, et al. Research progress in ship structure optimization design technology [J]. Chinese Journal of Ship Research, 2014, 9(04): 1-10.

[0009] [6]SHAN S,WANG G.Survey of modeling and optimization strategies to solve high-dimensional design problems with computationally-expensive black-box functions[J]. Structural&Multidisciplinary Optimization, 2010,41(2):219-241.

[0010] [7] Guo Xiaodong, Yu Minghua, Guan Yifeng, et al. Optimization of mid-section structure of large oil tanker based on relative difference quotient method [J]. Ship Mechanics, 2007, (04): 615-621.

[0011] [8] Chen Jing, Zhang Pan, Xiao Hanlin, et al. Multidisciplinary design optimization of hull structure subsystem design method for large ships [J]. Chinese Journal of Ship Research, 2013, 8(05): 27-32. Summary of the invention

[0012] In order to achieve rapid optimization of mid-section structural dimensions under complex code constraints, the present invention proposes a rapid optimization method for mid-section structural dimensions of ships based on sensitivity sorting, and develops a rapid optimization system for mid-section structural dimensions of ships under complex code constraints based on the general optimization platform nested code verification software. Compared with the intelligent optimization algorithm, the optimization method proposed by the present invention can achieve better optimization results, greatly reduce the amount of calculation, and improve the optimization efficiency by more than 80%.

[0013] To achieve the above-mentioned purpose, the technical solution of the present invention is: a method for rapid optimization of the structural dimensions of a cross-section of a ship, first, a solution in which all design variables take lower limit values ​​is selected as an initial solution, the characteristic quantity with the largest degree of constraint violation of each type is determined, and the adjustment order of the design variables is determined according to the sensitivity analysis results of the characteristic quantity to the design variables; at the same time, the values ​​of the corresponding most sensitive design variables are adjusted upward in sequence until all constraints meet the design requirements; then, according to the descending order of the influence of the design variables on the objective function, a small-scale adjustment of the optimization solution based on the coordinate rotation method is performed, a single design variable is taken in sequence, the values ​​of the remaining design variables are kept unchanged, the values ​​of the design variables are gradually reduced, and the feasibility of the solution is calculated using the specification verification software until the current solution does not meet the constraints and stops, and finally, the value of the design variable is increased by one step, and the value is used as the final value solution of the design variable. The design solution at this time is the optimal solution obtained.

[0014] A method for rapidly optimizing the structural dimensions of a ship's mid-section cross section, the specific steps of which are as follows:

[0015] Step 1: Input the name of the specification to be met, the mid-cross section structure data file established by the specification checking software, the upper and lower limits of the component size and the sensitivity update control parameter N, and perform a sensitivity update every N iterations;

[0016] Step 2: Select the corresponding optimization mathematical model and calculation command according to the selected specification;

[0017] Step 3: Use the general optimization platform to read the data file and complete the automatic definition of design variables, and calculate the upper and lower limits of each design variable based on the original size;

[0018] Step 4: Determine whether the solution with all design variables taking the lower limit meets the constraints. If so, directly output the lower limit solution as the optimization solution. If not, proceed to step 5.

[0019] Step 5: Determine whether the solution with all design variables taking the upper limit does not meet the constraints. If so, it means that there is no feasible solution within the range of the design variable value, and the program ends directly. If not, proceed to step 6;

[0020] Step 6: Set the benchmark scheme and calculate the response when a single variable changes one by one, that is, select a design variable, change the design variable based on the benchmark scheme, and calculate the response using the code checking software; then select another design variable and repeat the above steps until all design variables are traversed;

[0021] Step 7: Based on the results of step 6, calculate the rate of change of the characteristic quantity with respect to each design variable, i.e., sensitivity;

[0022] Step 8: According to the sensitivity calculation results of step 7, determine whether each type of feature quantity is only related to the local size:

[0023] Step 9: Select the scheme with all variables taking the lower limit as the initial scheme, calculate the structural response, and set i = 1. The initial scheme here refers to the scheme at the beginning of the iterative optimization calculation, which is different from the benchmark scheme calculated for the characteristic quantity change rate in step 6;

[0024] Step 10: Determine whether i is divisible by N, that is, whether sensitivity update is required. If yes, perform sensitivity calculation and update variable adjustment order (same as step 6 and step 7), and then proceed to step 11. If no, proceed directly to step 11;

[0025] Step 11: Check the constraint violation of the scheme according to the calculation results, and adjust the variables according to the constraint violation using the obtained variable adjustment order;

[0026] Step 12: Calculate the adjusted solution. If the constraint is satisfied, proceed to step 13. Otherwise, set i=i+1 and return to step 10.

[0027] Step 13: Make small-scale adjustments to the optimization solution based on the coordinate rotation method and output the final optimization solution.

[0028] In step 1, the sensitivity update control parameter N has a great influence on the optimization effect of the method, and the recommended setting is: N=5.

[0029] In step 3, the general optimization platform used is the MATLAB platform.

[0030] In steps 1, 6, and 13, the standard verification software used is mars2000 software.

[0031] In step 7, the calculation formula for the characteristic quantity change rate is In the formula, Z is any characteristic quantity, area is the total area of ​​the mid-cross section, Δz is the change of the characteristic quantity caused by the change of a certain design variable, and Δarea is the corresponding change of the total cross-sectional area. According to the size of the characteristic quantity change rate, the adjustment order of the design variables under the constraints of each characteristic quantity that does not meet the constraints is determined from high to low.

[0032] In step 8, the judgment method is: if the change rate of a certain characteristic quantity to the design variables except its own size variable is 0, it means that the characteristic quantity is only related to the local size.

[0033] In step 11, for each feature quantity related only to the local size, when it does not meet the constraint, the corresponding local size design variable is adjusted, and for the remaining feature quantities related to the global size, only the constraint with the largest degree of non-satisfaction is extracted for each category, and the variable is adjusted according to the sensitivity sorting, and only one step is increased for each step. The theoretical basis of this approach is: when there is a feature quantity related only to the local size that does not meet the constraint, in order to make it meet the constraint, the only way is to increase its corresponding local size variable, so there is no risk of over-adjustment when adjusting this type of local feature quantity.

[0034] In step 11, for the feature quantities related to the global size, if all feature quantities that do not meet the constraints are adjusted at the same time in each iteration step, there is a risk of over-adjustment because the increase of each design variable has a positive impact on all global feature quantities. Therefore, each step only adjusts the constraints that do not meet the greatest degree of constraints, and ultimately a lighter design solution can be obtained.

[0035] In step 13, the specific method is as follows: according to the objective function brought about by increasing a single design variable by one step in the sensitivity analysis calculation result, that is, the change in the total cross-sectional area, obtain the descending order of the influence of the design variables on the objective function; take a single design variable in turn, keep the values ​​of the other design variables unchanged, gradually reduce the value of the design variable, and use the specification verification software to calculate the feasibility of the scheme until the current scheme does not meet the constraints and stops; increase the value of the design variable by one step, and use this value as the final value scheme of the design variable, repeat the above method until all design variables are traversed, and then output the final optimization scheme; if the value of the design variable has reached the lower limit, there is no need to further calculate the design variable.

[0036] The beneficial effects of the present invention are:

[0037] The present invention proposes a method for rapid optimization of the structural dimensions of a mid-section of a ship based on sensitivity sorting, and develops a rapid optimization system for the structural dimensions of a mid-section of a ship under complex specification constraints based on the general optimization platform nested specification verification software. First, the sensitivity of each design variable is calculated for each constraint characteristic quantity, and the sensitivity result is used to obtain the adjustment order of the design variables when each characteristic quantity does not meet the constraint. At the same time, it is judged whether each characteristic quantity is only related to the local size according to the sensitivity calculation result. Then, the current scheme is subjected to specification verification calculation, and targeted variable adjustment is performed in combination with the sensitivity information according to the constraint violation of the current scheme, and the sensitivity is updated once every certain number of iteration steps. After the iterative calculation is completed, a small-range adjustment method of the optimization scheme based on the coordinate rotation method is introduced to further improve the optimization effect. Compared with the intelligent optimization algorithm of the directly nested specification verification software, the optimization method proposed by the present invention can obtain better optimization effects, and the amount of calculation is greatly reduced, and the optimization efficiency is improved by more than 80%. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a flowchart of the sensitivity-sort-based rapid optimization method for the mid-cross-section of the present invention. DETAILED DESCRIPTION

[0039] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0040] The basic concept of the mid-section rapid optimization method based on sensitivity sorting of the present invention is: if a certain characteristic quantity constraint does not meet the requirements, choosing to increase the size of the design variable that is most sensitive to it is the most efficient and lowest weight cost method. The specific idea is: select the scheme in which all design variables take the lower limit value as the initial scheme, determine the characteristic quantity with the largest degree of constraint violation of each type, and determine the adjustment order of the design variables according to the sensitivity analysis results of the characteristic quantity on the design variable; at the same time, adjust the corresponding most sensitive design variable values ​​upward in turn until all constraints meet the design requirements. According to the descending order of the influence of the design variables on the objective function, a small range adjustment of the optimization scheme based on the coordinate rotation method is performed, and a single design variable is taken in turn, and the values ​​of the remaining design variables are kept unchanged, and the values ​​of the design variables are gradually reduced, and the feasibility of the scheme is calculated using the specification verification software until the current scheme does not meet the constraints and stops. The value of the design variable is increased by one step, and the value is used as the final value scheme of the design variable. The design scheme at this time is the optimal scheme obtained.

[0041] like Figure 1 As shown, the method for rapid optimization of ship mid-section structural dimensions based on sensitivity sorting of the present invention comprises the following specific steps:

[0042] Step 1: Input the name of the specification to be met, the mid-cross-section structure data file established by the specification verification software (mars2000), the upper and lower limits of the component size, and the sensitivity update control parameter N (which means: a sensitivity update is performed every N iterations). The sensitivity update control parameter N has a great influence on the optimization effect of this method. The recommended setting is: N = 5;

[0043] Step 2: Select the corresponding optimization mathematical model and calculation command according to the selected specification;

[0044] Step 3: Use a general optimization platform (matlab) to read the data file and complete the automatic definition of design variables, and calculate the upper and lower limits of each design variable based on the original size;

[0045] Step 4: Determine whether the solution with all design variables taking the lower limit meets the constraints. If so, directly output the lower limit solution as the optimization solution. If not, proceed to step 5.

[0046] Step 5: Determine whether the solution with all design variables taking the upper limit does not meet the constraints. If so, it means that there is no feasible solution within the range of the design variable value, and the program ends directly. If not, proceed to step 6;

[0047] Step 6: Set the benchmark scheme and calculate the response when a single variable changes one by one. That is, select a design variable, change the design variable based on the benchmark scheme, and calculate the response using the specification verification software (mars2000); then select another design variable and repeat the above steps until all design variables are traversed;

[0048] Step 7: Based on the results of step 6, calculate the change rate (i.e. sensitivity) of the characteristic quantity with respect to each design variable. The calculation formula for the characteristic quantity change rate is In the formula, Z is any characteristic quantity, area is the total area of ​​the mid-cross section, ΔZ is the characteristic quantity change caused by a certain design variable change, and Δarea is the corresponding total cross-sectional area change. According to the size of the characteristic quantity change rate, the design variable adjustment order of each characteristic quantity that does not meet the constraint is determined from high to low;

[0049] Step 8: Determine whether each type of feature quantity is only related to the local size based on the sensitivity calculation results of step 7. The judgment method is: if the change rate of a feature quantity to the design variables except its own size variable is 0, it means that the feature quantity is only related to the local size;

[0050] Step 9: Select the scheme with all variables taking the lower limit as the initial scheme, calculate the structural response, and set i = 1. The initial scheme here refers to the scheme at the beginning of the iterative optimization calculation, which is different from the baseline scheme calculated for the characteristic quantity change rate in step 6;

[0051] Step 10: Determine whether i is divisible by N, that is, whether sensitivity update is required. If yes, perform sensitivity calculation and update variable adjustment order (same as step 6 and step 7), and then proceed to step 11. If no, proceed directly to step 11;

[0052] Step 11: Check the constraint violation of the scheme according to the calculation results, and adjust the variables according to the constraint violation using the obtained variable adjustment order. Among them, for each feature quantity related only to the local size, when it does not meet the constraint, the corresponding local size design variable is adjusted, and for the remaining feature quantities related to the global size, only the constraint with the largest degree of non-satisfaction is extracted for each category, and the variable is adjusted according to the sensitivity sorting, and only one step is increased for each step. The theoretical basis of this approach is: when there is a feature quantity related only to the local size that does not meet the constraint, in order to make it meet the constraint, the only way is to increase its corresponding local size variable, so there is no risk of over-adjustment when adjusting this type of local feature quantity. For the feature quantities related to the global size, if all feature quantities that do not meet the constraints are adjusted at the same time in each iteration step, there is a risk of over-adjustment because the increase of each design variable has a positive impact on all global feature quantities. Therefore, each step only adjusts the constraint with the largest degree of non-satisfaction, and finally a lighter design scheme can be obtained;

[0053] Step 12: Calculate the adjusted solution. If the constraint is satisfied, proceed to step 13. Otherwise, set i=i+1 and return to step 10.

[0054] Step 13: Make small-scale adjustments to the optimization scheme based on the coordinate rotation method and output the final optimization scheme. The specific approach is: according to the change in the objective function (total cross-sectional area) caused by increasing a single design variable by one step in the sensitivity analysis calculation results, obtain the descending order of the influence of the design variable on the objective function. Take a single design variable in turn, keep the values ​​of the other design variables unchanged, gradually reduce the value of the design variable, and use the specification verification software (mars2000) to calculate the feasibility of the scheme until the current scheme does not meet the constraints and stops. Increase the value of the design variable by one step, and use this value as the final value scheme of the design variable. Repeat the above steps until all design variables are traversed, and then output the final optimization scheme. It should be noted that if the value of the design variable has reached the lower limit, there is no need to further calculate the design variable.

[0055] From the basic principle point of view, the method of the present invention is similar to the tanker mid-section optimization based on the relative difference quotient method in the literature [7]. Both adopt a gradient-like index (relative difference quotient / sensitivity) as the classic optimization idea for variable adjustment in the iterative process. However, there are also important differences: the sensitivity of the present method is established for each single constraint feature quantity, and the sensitivity result is used to obtain the adjustment order of the design variables when each feature quantity does not meet the constraint. At the same time, the sensitivity calculation result is used to judge whether each feature quantity is only related to the local size. During iteration, the variable adjustment is carried out in a targeted manner according to the constraint violation of the current solution combined with the sensitivity information, which makes full use of the characteristics of each constraint feature quantity. However, the literature [7] calculates the relative difference quotient for all valid constraint feature quantities as a whole, and does not consider the characteristic information of the constraint feature quantity. In addition, this method also introduces a small-scale adjustment strategy for the optimization scheme based on the coordinate rotation method, and develops a general rapid optimization system for mid-sections under specification constraints based on the general optimization platform (matlab) nested specification verification software (Mars2000). This system can realize the rapid optimization of mid-sections of different types of ships under different specifications (Mars2000 software supports BV / CSR / HCSR specification calculations), while reference [7] only focuses on the optimization of mid-sections of oil tankers under DNV specification constraints.

Claims

1. A method for rapid optimization of ship mid-section structural dimensions, characterized in that: First, the scheme with all design variables taking lower limit values ​​is selected as the initial scheme, the characteristic quantity with the largest degree of constraint violation of each type is determined, and the change rate of the characteristic quantity with respect to each design variable, that is, the sensitivity, is calculated. The formula for calculating the change rate of the characteristic quantity is: In the formula, Z is any characteristic quantity, area is the total area of ​​the mid-cross section, ΔZ is the change in characteristic quantity caused by the change of a certain design variable, and Δarea is the corresponding change in total cross-sectional area. The adjustment order of design variables is determined based on the sensitivity analysis results of the characteristic quantity on the design variables. At the same time, the corresponding most sensitive design variable values ​​are adjusted upwards in sequence until all constraints meet the design requirements; Then, according to the descending order of the impact of the design variables on the objective function brought about by increasing a single design variable by one step in the sensitivity analysis calculation results, a small range adjustment of the optimization scheme based on the coordinate rotation method is performed. A single design variable is taken in turn, and the values ​​of the other design variables are kept unchanged. The value of the design variable is gradually reduced, and the feasibility of the scheme is calculated using the specification verification software until the current scheme does not meet the constraints and stops. Finally, the value of the design variable is increased by one step, and this value is used as the final value scheme of the design variable. The design scheme at this time is the optimal scheme.

2. According to claim 1, the method for rapid optimization of ship mid-section structural dimensions is characterized by: The specific steps are as follows: Step 1: Input the name of the specification to be met, the mid-cross section structure data file established by the specification checking software, the upper and lower limits of the component size and the sensitivity update control parameter N, and perform a sensitivity update every N iterations; Step 2: Select the corresponding optimization mathematical model and calculation command according to the selected specification; Step 3: Use the general optimization platform to read the data file and complete the automatic definition of design variables, and calculate the upper and lower limits of each design variable based on the original size; Step 4: Determine whether the solution with all design variables taking the lower limit meets the constraints. If so, directly output the lower limit solution as the optimization solution. If not, proceed to step 5. Step 5: Determine whether the solution with all design variables taking the upper limit does not meet the constraints. If so, it means that there is no feasible solution within the range of the design variable value, and the program ends directly. If not, proceed to step 6; Step 6: Set the benchmark scheme and calculate the response when a single variable changes one by one, that is, select a design variable, change the design variable based on the benchmark scheme, and calculate the response using the code checking software; then select another design variable and repeat the above steps until all design variables are traversed; Step 7: Based on the results of step 6, calculate the rate of change of the characteristic quantity with respect to each design variable, i.e., sensitivity; Step 8: According to the sensitivity calculation results of step 7, determine whether each type of feature quantity is only related to the local size: Step 9: Select the scheme with all variables taking the lower limit as the initial scheme, calculate the structural response, and set i = 1. The initial scheme here refers to the scheme at the beginning of the iterative optimization calculation, which is different from the benchmark scheme calculated for the characteristic quantity change rate in step 6; Step 10: Determine whether i is divisible by N, that is, whether sensitivity update is required. If yes, perform sensitivity calculation and update variable adjustment order, same as step 6 and step 7, and then proceed to step 11. If no, proceed directly to step 11; Step 11: Check the constraint violation of the scheme according to the calculation results, and adjust the variables according to the constraint violation using the obtained variable adjustment order; Step 12: Calculate the adjusted solution. If the constraint is satisfied, proceed to step 13. Otherwise, set i=i+1 and return to step 10. Step 13: Make small-scale adjustments to the optimization solution based on the coordinate rotation method and output the final optimization solution.

3. The method for rapid optimization of ship mid-section structural dimensions according to claim 2 is characterized in that: In step 1, the sensitivity update control parameter N has a great influence on the optimization effect of the method, and the recommended setting is: N=5.

4. The method for rapid optimization of ship mid-section structural dimensions according to claim 2 is characterized in that: In step 3, the general optimization platform used is the MATLAB platform.

5. The method for rapid optimization of ship mid-section structural dimensions according to claim 2 is characterized in that: In steps 1, 6, and 13, the standard verification software used is mars2000 software.

6. The method for rapid optimization of ship mid-section structural dimensions according to claim 2 is characterized by: In step 7, the order of adjusting the design variables when each feature quantity does not meet the constraints is determined from high to low according to the size of the feature quantity change rate.

7. The method for rapid optimization of ship mid-section structural dimensions according to claim 2 is characterized by: In step 8, the judgment method is: if the change rate of a certain characteristic quantity to the design variables except its own size variable is 0, it means that the characteristic quantity is only related to the local size.

8. The method for rapid optimization of ship mid-section structural dimensions according to claim 2 is characterized by: In step 11, for each feature quantity related only to the local size, when it does not meet the constraint, the corresponding local size design variable is adjusted. For the remaining feature quantities related to the global size, only the constraint with the greatest degree of non-satisfaction is extracted for each category, and the variable is adjusted according to the sensitivity sorting, and only one step is increased in each step. The theoretical basis of this approach is: when there is a feature quantity related only to the local size that does not meet the constraint, the only way to make it meet the constraint is to increase its corresponding local size variable, so there is no risk of over-adjustment when adjusting this type of local feature quantity.

9. The method for rapid optimization of ship mid-section structural dimensions according to claim 8 is characterized by: In step 11, for the feature quantities related to the global size, if all feature quantities that do not meet the constraints are adjusted at the same time in each iteration step, there is a risk of over-adjustment because the increase of each design variable has a positive impact on all global feature quantities. Therefore, each step only adjusts the constraints that do not meet the greatest degree of constraints, and ultimately a lighter design solution can be obtained.

10. The method for rapid optimization of ship mid-section structural dimensions according to claim 2, characterized in that: In step 13, the specific method is as follows: according to the objective function brought by increasing a single design variable by one step in the sensitivity analysis calculation result, that is, the change in the total cross-sectional area, the descending order of the influence of the design variables on the objective function is obtained; a single design variable is taken in turn, the values ​​of the other design variables are kept unchanged, the value of the design variable is gradually reduced, and the feasibility of the scheme is calculated using the specification verification software until the current scheme does not meet the constraints and stops; the value of the design variable is increased by one step, and the value is used as the final value scheme of the design variable, and the above method is repeated until all design variables are traversed, and then the final optimization scheme is output; If the value of the design variable has reached the lower limit, no further calculation is required for the design variable.

Images