A method for optimizing the stiffness of blocking timbers based on the docking strength of a hull

Abstract

The present invention relates to the technical field of hull structure design, and particularly relates to a method for optimizing the stiffness of blocking timbers based on the strength of the hull during docking. The method for optimizing the stiffness of blocking timbers based on the strength of the hull during docking proposed by the present invention, based on the one-dimensional beam finite element model of the hull during docking, can efficiently obtain a relatively reasonable distribution of the stiffness coefficients of the blocking timbers within a limited time through technical means such as experimental design, approximate models, and optimization algorithms, greatly shortening the time for designing the stiffness coefficients of the blocking timbers, and helping to improve the reaction force level of the target blocking timbers, ensuring the safety of the hull structure during docking, and effectively improving the efficiency of formulating the ship docking plan.

Description

Technical Field

[0001] The present invention relates to the technical field of hull structure design, and particularly relates to a method for optimizing the stiffness of dock blocks based on the strength of the hull during docking. Background Art

[0002] The dock building method is a common construction method for large ships. The hull is usually located on the dock blocks in the dock. The self-weight of the hull will exert a certain force on the dock blocks, and at the same time, the reaction force of the dock blocks will also act on the hull. During the docking process, it is necessary to ensure not only that the dock blocks are not damaged, but also that the hull structure has sufficient strength. Therefore, considering the safety of the docking state, it is necessary to calculate and check the strength of the hull during docking.

[0003] In engineering, due to different hull structure types, the ability of the hull structure at different dock block positions to resist the action of the reaction force is also different. This ability can be obtained by comprehensively using theoretical calculations, simulation analyses, experimental studies and other means. Since the sum of the reaction forces of each dock block is equal to the weight of the hull, on the premise that the weight of the hull remains unchanged, it is often desired that the weaker structure bears a smaller reaction force, while the stronger structure bears a larger reaction force, so as to achieve a balance between the structural ability and the reaction force of the dock blocks.

[0004] The stiffness coefficient of the dock blocks affects the reaction force of the dock piers, and thus affects the strength of the hull during docking. Changing the stiffness coefficient of the dock blocks at a certain position will have different degrees of influence on the reaction forces at all dock block positions. The large number of dock blocks in engineering will increase the difficulty of evaluating this influence. Repeated and inefficient iterative operations will take a lot of time, resulting in difficulty in accurately evaluating the influence degree of the stiffness coefficients of each dock block on the reaction forces at different positions in the existing design methods. Therefore, how to efficiently obtain a reasonable distribution of the stiffness coefficients of the dock blocks within a limited time is of great significance to the safety and economy of ship docking. Summary of the Invention

[0005] The technical problem to be solved by the present invention is: to propose a method for optimizing the stiffness of dock blocks based on the strength of the hull during docking, which can efficiently obtain and optimize the distribution of the stiffness coefficients of the dock blocks, greatly shorten the time for designing the stiffness coefficients of the dock blocks, help improve the safety of the hull during docking, and can effectively improve the efficiency of formulating the ship docking plan.

[0006] To solve the above technical problem, the technical solution adopted by the present invention is:

[0007] A method for optimizing the stiffness of dock blocks based on the strength of the hull during docking, comprising the following steps:

[0008] S1, establish a one-dimensional beam finite element model of the hull and the dock blocks, and analyze the reaction forces of each dock block under the action of the self-weight of the hull;

[0009] S2, selecting sample points by using an optimized Latin hypercube design method, and establishing a neural network approximate model between the hull docking response and the pier stiffness coefficient according to the sample point data;

[0010] S3, based on the neural network approximate model, taking the minimum support reaction force of the target pier as the design goal, and taking the support reaction force of all piers less than the preset allowable value as the constraint condition, a multi-island genetic algorithm is used to perform global optimization design on the pier stiffness coefficient.

[0011] Furthermore, in step S1, the one-dimensional beam finite element model of the hull and the piers is established, specifically, based on the weight distribution of the whole ship, the profile characteristics, the pier distribution and the pier stiffness coefficient, a finite element model of the hull beam elastically supported on the dock pier is established in the finite element software, and the solver provided in the finite element software is used to calculate the support reaction data of each pier under the action of the hull's own weight.

[0012] Furthermore, in step S2, the optimized Latin hypercube design method includes:

[0013] S21, divide the whole dock into m n design space cells, and then generate sample points cell by cell;

[0014] S22, randomly generate the first sample point P1 in the first spatial cell, define the coordinates of the sample point P1 as (i1, j1, ..., g1, 1), and put the sample point P1 into the sample set P = {P1}, where i1, j1, ..., g1∈{1, 2, ..., m};

[0015] S23, generating a second sample point P2 at a position farthest from the first sample point P1 in the second spatial cell, defining the coordinates of the sample point P2 as (i2, j2, ..., g2, 2), and updating the sample set P = {P1, P2};

[0016] S24, for the kth sample point P k , where k ≥ 3, calculate the positions of each pier in the k-th spatial cell and the sample set P = {P1, P2, ..., P k-1}, and take the pier position corresponding to the maximum distance value to generate the kth sample point P k , define the sample point P k The coordinates of (i k , j k , …, g k , k), and update the sample set P = {P1, P2, ..., P k-1 , P k};

[0017] S25, repeat step S24 until m sample points are generated.

[0018] Further, in step S2, the hull docking response in the neural network approximation model specifically refers to the reaction force data of each sample point in the sample set P = {P1, P2, …, P m}.

[0019] Further, in step S2, the fender stiffness coefficient in the neural network approximation model specifically refers to the fender stiffness coefficient at the position of each sample point in the sample set P = {P1, P2, …, P m}.

[0020] Further, in step S3, the multi-island genetic algorithm includes:

[0021] S31, dividing the sample points in the sample set P = {P1, P2, …, P m} into several sub-populations, each sub-population randomly containing a certain number of individuals, and converting the reaction force data corresponding to each individual into binary coding form;

[0022] S32, evaluating the reaction force data corresponding to the stiffness coefficient of each individual in the neural network approximation model. The smaller the reaction force, the higher the fitness of the corresponding individual;

[0023] S33, selecting according to the fitness of the individuals. The higher the fitness of an individual, the higher the probability of being selected into the next generation;

[0024] S34, performing crossover operation and mutation operation on the selected individuals;

[0025] S35, repeating steps S32 to S34 until the optimal stiffness coefficient distribution is reached.

[0026] Further, the crossover operation specifically is: randomly selecting the stiffness coefficients of any individuals from the selected individuals and taking the average value to generate new individuals.

[0027] Further, the mutation operation specifically is: randomly selecting the stiffness coefficient of any individual from the selected individuals and multiplying it by a preset correction coefficient to generate new individuals.

[0028] Further, the optimal stiffness coefficient distribution specifically is: after performing crossover operation and mutation operation on the selected individuals, when the change rate of the population is less than a preset value, the corresponding fender stiffness coefficient distribution.

[0029] The present invention has the following main advantages compared with the prior art:

[0030] An optimization method for the stiffness of blocking timbers based on the strength of a hull during docking, based on a one-dimensional beam finite element model of the hull during docking, can efficiently obtain a relatively reasonable distribution of the stiffness coefficients of the blocking timbers within a limited time through technical means such as experimental design, approximate models, and optimization algorithms. It can significantly shorten the time for designing the stiffness coefficients of the blocking timbers, help improve the reaction force level of the target blocking timbers, ensure the safety of the hull structure during docking, and effectively improve the efficiency of formulating the ship docking plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 Simplified schematic diagram of the hull and blocking timbers in an embodiment of the present invention;

[0032] Figure 2 Schematic diagram of the distribution of sample points in the experimental design in an embodiment of the present invention;

[0033] Figure 3 Schematic diagram of the neural network model between F1 and the change rate between K1 and K3 in an embodiment of the present invention;

[0034] Figure 4 Flowchart of the method for establishing an approximate model and optimizing design in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0036] It should be noted that according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, or two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the objectives of the present invention.

[0037] Embodiment 1. This embodiment provides an optimization method for the stiffness of blocking timbers based on the strength of a hull during docking, mainly including the following steps:

[0038] S1. Establish a one-dimensional beam finite element model of the hull and the blocking timbers, and analyze the reaction forces of each blocking timber under the action of the hull's own weight;

[0039] S2. Select sample points using the optimized Latin hypercube design method, and establish a neural network approximate model between the hull docking response and the stiffness coefficients of the blocking timbers according to the sample point data;

[0040] S3, based on the neural network approximate model, taking the minimum support reaction force of the target pier as the design goal, and taking the support reaction force of all piers less than the preset allowable value as the constraint condition, a multi-island genetic algorithm is used to perform global optimization design on the pier stiffness coefficient.

[0041] Furthermore, in step S1, the one-dimensional beam finite element model of the hull and the piers is established, specifically, based on the weight distribution of the whole ship, the profile characteristics, the pier distribution and the pier stiffness coefficient, a finite element model of the hull beam elastically supported on the dock pier is established in the finite element software, and the solver provided in the finite element software is used to calculate the support reaction data of each pier under the action of the hull's own weight.

[0042] Furthermore, in step S2, the optimized Latin hypercube design method includes:

[0043] 1) Divide the design space evenly into m n Space units, and then generate sample points grid by grid. For the first sample point P1, it can be randomly generated in the first grid. Define the position of point P1 (i1, j1, ..., g1, 1) (i1, j1, ..., g1∈{1, 2, ..., m}), and put point P1 into the sample set P = {P1};

[0044] 2) For the second sample point P2, it will be generated in the second grid. The position with coordinates (i1, j1, ..., g1, 1) is already occupied by P1, so the sample point P2 can only be placed in the remaining space unit of the second grid. The coordinates of the position of point P2 are defined as (i2, j2, ..., g2, 2). The maximum distance between the position of the remaining space unit of the second grid and the existing sample point P1 can be expressed as max(d((i2, j2, ..., g2, 2), P1))(i2∈{r i |r i =1, 2, ..., m; r i ≠i1}, j2∈{r j |r j =1, 2, ..., m; r j ≠j1},…,g2∈{r g |r g =1, 2, ..., m; r g ≠g1}). Calculate the distance between the remaining spatial unit position of the second grid and the sample point P1, and the sample point P2 will be placed in the spatial unit farthest from point P1. Update the sample set P = {P1, P2};

[0045] 3) For the kth sample point P k , {k|k=3,4,…,m-1}, indicating that there are sample points {P1,P2,…,P k-1} has occupied k-1 grids. Therefore, the kth sample point can only appear in the remaining space unit of the kth grid, that is, Pk (i k , j k , …, g k , k)(i k ∈ {r i | r i =1, 2, …, m; r i ≠ i1, i2, …, i k-1}, j k ∈ {r j | r j =1, 2, …, m; r j ≠ j1, j2, …, j k-1}, …, g k ∈ {r g | r g =1, 2, …, m; r g ≠ g1, g2, …, g k-1})。In the remaining space unit of the k-th grid, calculate the distances between the positions of each pier timber and the existing sample points {P1, P2, …, P k-1}, and take the maximum distance value as the eigenvalue. The position of the pier timber corresponding to the maximum eigenvalue is the coordinate position (i k , j k , …, g k , k) of the k-th sample point. The maximum and minimum distances between the k-th sample point P k and the existing sample points P can be expressed as max(d((i k , j k , …, g k , k), P)). Update the sample set P = {P1, P2, …, P k-1 , P k};

[0046] 4) Repeat step 3) until m - 1 sample points are generated;

[0047] 5) The m-th sample point is placed in the last remaining space unit.

[0048] Furthermore, in step S2, the hull docking response in the neural network approximation model is specifically the reaction force data of each sample point in the sample set P = {P1, P2, …, P m}}; the pier timber stiffness coefficient in the neural network approximation model is specifically the pier timber stiffness coefficient at the position of each sample point in the sample set P = {P1, P2, …, P m}}.

[0049] Furthermore, in step S3, the multi-island genetic algorithm includes:

[0050] S31, Take the sample set P = {P1, P2, …, Pm The sample points in} are divided into several sub-populations, each of which contains a certain number of individuals at random, and the support reaction data corresponding to each individual is converted into binary coding form;

[0051] S32, evaluating the support reaction force data corresponding to the stiffness coefficient of each individual in the neural network approximate model, the smaller the support reaction force, the higher the fitness of the corresponding individual;

[0052] S33, selection is made based on the fitness of the individual. Individuals with higher fitness have a higher probability of being selected to enter the next generation.

[0053] S34, performing crossover and mutation operations on the selected individuals;

[0054] S35, repeat steps S32 to S34 until the optimal stiffness coefficient distribution is reached.

[0055] The crossover operation is specifically: randomly selecting the stiffness coefficient of any individual from the selected individuals and taking the average value to generate a new individual.

[0056] The mutation operation is specifically as follows: randomly selecting a stiffness coefficient of any individual from the selected individuals, and multiplying it by a preset correction coefficient to generate a new individual.

[0057] Furthermore, the optimal stiffness coefficient distribution is specifically: after the selected individuals are subjected to crossover and mutation operations, when the rate of change of the population is less than a preset value, the corresponding stiffness coefficient distribution of each pier.

[0058] Embodiment 2: This embodiment provides a method for optimizing the stiffness of pier wood based on the docking strength of the hull, which mainly includes:

[0059] 1) A one-dimensional beam finite element model of the hull and piers is established to analyze the reaction force of the piers under the deadweight of the hull.

[0060] Based on the information of the whole ship weight distribution, profile characteristics, dock pier distribution, dock pier stiffness coefficient and so on, a finite element model of the hull beam elastically supported on the dock pier is established in the finite element software, and the support reaction force value of each dock pier under the action of the hull's deadweight is calculated by the solver provided by the finite element software.

[0061] 2) The optimized Latin hypercube design method is used to select sample points, and a neural network approximate model between the hull docking response and the pier stiffness coefficient is established based on the sample data.

[0062] The optimized Latin hypercube design method generates more uniform sampling points compared to the traditional Latin hypercube method, and obtains relatively reasonable test data at a limited time cost. An approximate neural network model between the hull docking response and the stiffness coefficient of the blocking timbers is established based on the test data, and the credibility of the approximate model is verified by means of error analysis and other methods. The neural network model has the ability to approximate complex non-linear functions and has a strong fault tolerance function.

[0063] 3) Based on the approximate neural network model, with the minimum reaction force of the target blocking timbers as the design objective and the condition that the reaction forces of all blocking timbers are less than the allowable value as the constraint condition, a multi-island genetic algorithm is used to globally optimize the stiffness coefficient of the blocking timbers.

[0064] The invocation of the approximate neural network model reduces the number of times of invoking the finite element software and shortens the optimization time. The multi-island genetic algorithm can better achieve global optimization and prevent falling into local optimal solutions.

[0065] 4) Comprehensively apply the above technologies to form an intelligent optimization design method for the stiffness coefficient of the blocking timbers based on the hull docking strength.

[0066] Embodiment 3: In this embodiment, the total length of the hull is 60 m, the total weight is 600 t, and the weight is evenly distributed. The hull is divided into 6 theoretical stations along the ship length direction, and the length of each theoretical station is 10 m. The blocking timbers are arranged in the middle between two adjacent station numbers and are evenly distributed along the ship length direction. The hull is located on the blocking timbers. The stiffness coefficients of each blocking timber are K1, K2,... K6, and the reaction forces of each blocking timber are F1, F2,... F6. A simplified schematic diagram of the hull docking is as Figure 1 shown.

[0067] In this embodiment, the stiffness coefficients of K1, K2,... K6 are all 10000 kN / mm, and the value range of the change rate is -30% - 30%. Taking K1 and K3 as design variables (the design variables can be selected according to the actual engineering situation), and taking the minimum of F1 and F3 responses as the optimization objective (the optimization objective can be selected according to the actual engineering situation), the ability of the hull structure at different blocking timber positions to resist the action of the reaction force is different, and the acquisition of this ability does not belong to the content of this patent and will not be elaborated here. Assuming that the reaction forces at all blocking timber positions do not exceed 20% of the original level as the constraint condition, a set of optimal stiffness coefficient distributions is given. The optimization mathematical expression is as follows:

[0068] Find K1 and K3

[0069] Min F1 and F3

[0070] s.t.F2,F4,F5,F6≤[F]

[0071] 1) Through the experimental design process, the numerical combinations of the change rates of K1 and K3 are simulated using the optimized Latin hypercube design method, as shown in Figure 2 shown.

[0072] 2) The reaction forces of the dock blocks under the above inputs are calculated through a one-dimensional beam finite element model. A neural network approximation model is established based on the sample data, as shown in Figure 3 shown. The credibility of the approximation model is verified through error analysis.

[0073] 3) Based on this approximation model, a multi-island genetic algorithm is used to carry out the optimization analysis, as shown in Figure 4 shown.

[0074] After optimization, when K1 = 7030579.08 N / mm and K3 = 7075532.158 N / mm, the structure satisfies the above mathematical expression. Table 1 shows the comparison between the optimized approximation model and the actual model, and the maximum error value is 0.31%.

[0075] Table 1 Comparison of Responses between the Approximation Model and the Actual Model

[0076]

[0077] Table 2 shows the comparison of the reaction forces of each dock block before and after optimization. It can be found that the reaction force of the target dock block is reduced by up to 17.02% after optimization, and the reaction force levels of the dock blocks at other positions are within the allowable range, meeting the design requirements preset in this embodiment.

[0078] Table 2 Comparison of Reaction Force Responses before and after Optimization

[0079]

[0080]

[0081] Furthermore, the parts not described in detail in this application are the same as or implemented using the prior art.

[0082] To sum up:

[0083] A method for optimizing the stiffness of dock blocks based on the hull docking strength proposed by the present invention, based on a one-dimensional beam finite element model of the hull docking, through technical means such as experimental design, approximation model, and optimization algorithm, can efficiently obtain a relatively reasonable distribution of dock block stiffness coefficients within a limited time, greatly shortening the time for designing dock block stiffness coefficients, and helping to improve the reaction force level of the target dock block, ensuring the safety of the hull structure during docking, and effectively improving the efficiency of formulating the ship docking plan.

[0084] Those skilled in the art can easily understand that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for optimizing the stiffness of blocking timbers based on the strength of the hull when in dry dock, characterized in that, The steps include: S1, establish a one-dimensional beam finite element model of the hull and piers, and analyze the support reaction force of each pier under the deadweight of the hull; S2, selecting sample points by using an optimized Latin hypercube design method, and establishing a neural network approximate model between the hull docking response and the pier stiffness coefficient according to the sample point data; S3, based on the neural network approximate model, taking the minimum support reaction force of the target pier as the design goal, and taking the support reaction force of all piers less than the preset allowable value as the constraint condition, a multi-island genetic algorithm is used to perform global optimization design on the pier stiffness coefficient.

2. The method for optimizing the stiffness of blocking timbers based on the docking strength of a hull according to claim 1, wherein In step S1, the one-dimensional beam finite element model of the hull and the pier is established, specifically, based on the weight distribution of the whole ship, the profile characteristics, the pier distribution and the pier stiffness coefficient, a finite element model of the hull beam elastically supported on the dock pier is established in the finite element software, and the solver provided by the finite element software is used to calculate the support reaction force data of each pier under the action of the hull's own weight.

3. A method for optimizing the stiffness of blocking timbers based on the docking strength of a hull, as claimed in claim 1, wherein In step S2, the optimized Latin hypercube design method includes: S21, evenly divide the dock pier as a whole into m n design space cells, and then generate sample points cell by cell; S22, randomly generate the first sample point P1 in the first spatial cell, define the coordinates of the sample point P1 as (i1, j1, ..., g1, 1), and put the sample point P1 into the sample set P = {P1}, where i1, j1, ..., g1∈{1, 2, ..., m}; S23, generating a second sample point P2 at a position farthest from the first sample point P1 in the second spatial cell, defining the coordinates of the sample point P2 as (i2, j2, ..., g2, 2), and updating the sample set P = {P1, P2}; S24. For the k-th sample point P k , where k ≥ 3, calculate the distance values between each pier position in the k-th spatial cell and each sample point in the sample set P = {P1, P2, …, P k-1}, and generate the k-th sample point P k by taking the pier position corresponding to the maximum distance value. Define the coordinates of the sample point P k as (i k , j k , …, g k , k), and update the sample set P = {P1, P2, …, P k-1 , P k}; S25, repeat step S24 until m sample points are generated.

4. A method for optimizing the stiffness of blocking timbers based on the docking strength of a hull according to claim 3, characterized in that In step S2, the hull docking response in the neural network approximation model is specifically the reaction force data of each sample point in the sample set P = {P1, P2, …, P m}.

5. The method for optimizing the stiffness of blocking timbers based on the hull docking strength according to claim 4, wherein In step S2, the stiffness coefficient of the pier timber in the neural network approximation model is specifically the stiffness coefficient of the pier timber at the positions of the respective sample points in the sample set P = {P1, P2, …, P m}.

6. The optimization method for the stiffness of blocking timbers based on the hull docking strength according to claim 5, characterized in that In step S3, the multi-island genetic algorithm includes: S31. Divide the sample points in the sample set P = {P1, P2,..., P m} into several subpopulations. Each subpopulation randomly contains a certain number of individuals, and convert the reaction force data corresponding to each individual into binary coding form; S32, evaluating the support reaction force data corresponding to the stiffness coefficient of each individual in the neural network approximate model, the smaller the support reaction force, the higher the fitness of the corresponding individual; S33, selection is made based on the fitness of the individual. Individuals with higher fitness have a higher probability of being selected to enter the next generation. S34, performing crossover and mutation operations on the selected individuals; S35, repeat steps S32 to S34 until the optimal stiffness coefficient distribution is reached.

7. A method for optimizing the stiffness of blocking timbers based on the hull docking strength according to claim 6, characterized in that The crossover operation is specifically: randomly selecting the stiffness coefficient of any individual from the selected individuals and taking the average value to generate a new individual.

8. A method for optimizing the stiffness of blocking timbers based on the hull docking strength according to claim 6, characterized in that, The mutation operation is specifically as follows: randomly selecting the stiffness coefficient of any individual from the selected individuals, and multiplying it by a preset correction coefficient to generate a new individual.

9. The method for optimizing the stiffness of blocking timbers based on the docking strength of a hull according to claim 6, wherein The optimal stiffness coefficient distribution is specifically: after the selected individuals are subjected to crossover and mutation operations, when the population change rate is less than a preset value, the corresponding stiffness coefficient distribution of each pier.

Images