A one-dimensional large-scale optimized material cutting method for intelligent construction

Abstract

This invention relates to the field of digital and intelligent construction, and discloses a one-dimensional large-scale optimized material cutting method for intelligent construction. The main steps include the parameterized automatic generation of a linear component digital model to form the basis of material cutting data, determining the optimization objective, designing an intelligent swarm algorithm for large-scale one-dimensional material cutting problems to obtain the current optimal cutting combination and usage frequency, and finally determining the material cutting scheme and exporting the material cutting form. Based on BIM technology and intelligent optimization algorithms, this invention provides a new method for large-scale one-dimensional material cutting problems in the construction process. This invention can significantly reduce the surplus material rate and the number of cutting combinations in the material cutting process, effectively improve the solution speed of large-scale one-dimensional material cutting problems, while maintaining a certain level of solution accuracy, meeting actual production needs and possessing significant economic and social benefits.

Description

Technical Field

[0001] This invention relates to the field of building engineering, and in particular to a method based on BIM technology and intelligent optimization algorithms to provide a data foundation for large-scale one-dimensional material cutting problems and decompose and downgrade the material cutting problems, thereby achieving high-precision and high-efficiency optimization of material cutting problems. Background Technology

[0002] In the field of construction engineering, there is a problem of how to cut linear components such as steel pipes, reinforcing bars, and pipelines. This problem belongs to one-dimensional cutting problems. One-dimensional cutting problems are the most common problems in modern industrial production. They refer to finding a cutting scheme that minimizes the waste material generated when cutting raw materials, given the required part data. Its essence is the efficient allocation of limited resources.

[0003] The one-dimensional cutting problem is a classic NP-hard problem, and numerous experts and scholars have explored it and proposed various algorithms. The first category is exact algorithms, such as linear programming, dynamic programming, and the cutting plane algorithm. These algorithms require listing all possible cutting combinations and theoretically can obtain the optimal solution. However, as the cutting scale increases, using these algorithms becomes extremely time-consuming. The second category is heuristic algorithms, such as genetic algorithms, simulated annealing, and ant colony algorithms. These algorithms are faster and suitable for solving large-scale one-dimensional cutting problems, but they cannot guarantee the quality of the solution. Incorporating search strategies into heuristic algorithms can improve the quality of the solution while maintaining speed.

[0004] Currently, BIM technology has been gradually applied in the field of building engineering, but its application in one-dimensional material cutting is still lacking. Research on one-dimensional material cutting problems is still limited to algorithm optimization, and there is no relevant research that combines BIM with intelligent optimization algorithms to propose a more effective optimization method for solving one-dimensional material cutting problems. Summary of the Invention

[0005] The purpose of this invention is to provide a one-dimensional large-scale optimized material cutting method for intelligent construction, addressing the technical problems of low efficiency and low accuracy in existing methods for optimizing large-scale one-dimensional material cutting problems. The method involved in this invention can improve the solution accuracy and efficiency of large-scale one-dimensional material cutting problems, reduce waste material generation, and lower engineering costs.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A one-dimensional large-scale optimized material cutting method for intelligent construction includes the following steps:

[0008] Step 1: Create a BIM model of the main structure. In the BIM design platform, create a BIM model of the main structure. This main structure serves as the object for the linear component layout, used to determine the layout boundaries of the linear components.

[0009] Step 2: Optimize the layout of linear components and establish a BIM model of the linear components. Through the interaction between the BIM design platform and the secondary development platform, the parametric automatic generation of the linear components is completed.

[0010] Step 3: Create a material cutting data form to establish the foundation for material cutting data. Based on the linear component BIM model, determine the material cutting information and summarize it into a material cutting data form. The material cutting information includes the raw material length of the linear component, the length and required quantity of the parts to be cut, the number of part specifications, and the total number of parts, etc.

[0011] Step 4: Define the optimization objective and establish a mathematical model. Establish a mathematical model for the one-dimensional material cutting problem with the objectives of minimizing the residual material rate and the number of cutting combinations.

[0012] Step 5: Obtain the optimal cutting combination under the current material cutting scale. This involves an improved particle swarm optimization algorithm, which uses the remaining material ratio as a metric and only seeks the optimal cutting combination under the current material cutting scale at each step, thereby continuously decomposing and reducing the one-dimensional material cutting problem into simpler subproblems. An automatic adjustment rule is added to the particle swarm optimization algorithm to constrain the inertia coefficient, improving the algorithm's optimization ability while avoiding the occurrence of local optima.

[0013] Step Six: Optimize the frequency of use of the best cutting combination. Within the constraints, increase the frequency of use of the current best cutting combination as much as possible;

[0014] Step 7: Update the data of the parts to be cut, reduce the scale of the cutting problem, and set the iteration judgment condition. Use the demand quantity of the parts to be cut as the judgment condition. When the demand quantity of the parts to be cut is 0, the algorithm ends; otherwise, return to step 5 to continue iterating.

[0015] Step 8: Determine the final cutting scheme. After the iteration terminates, the optimal cutting combination and its corresponding usage frequency obtained from solving each sub-problem are combined to form the final cutting scheme, which is the optimized solution to the large-scale one-dimensional cutting problem;

[0016] Step 9: Output data and export the material cutting optimization form. The material cutting optimization form includes the complete material cutting plan, the effective material utilization length for each cutting combination, the residual material length for each cutting combination, the residual material rate for each cutting combination, and the total residual material rate of the material cutting plan, etc.

[0017] Step 10: Case Comparison. Using known material cutting data, compare the algorithms of a one-dimensional large-scale optimized material cutting method for intelligent construction to verify its feasibility and advancement.

[0018] When creating the material cutting data form in step three above, the material cutting information required in the steps is included, but is not limited to.

[0019] The mathematical model in step four above is created with the goal of minimizing the residual material rate and the number of cutting methods, but it is not limited to this and can be adjusted according to actual engineering needs.

[0020] The particle swarm optimization algorithm used in step five above aims to find the optimal cutting combination under the current material feeding scale, but it is not limited to the particle swarm optimization algorithm; other intelligent optimization algorithms can also achieve the same purpose.

[0021] When exporting the material cutting optimization form in step nine above, including but not limited to the information mentioned in the steps, the material cutting optimization form can be designed by the user.

[0022] Compared with the prior art, the beneficial effects of the present invention are:

[0023] This invention provides a more optimized approach for solving large-scale one-dimensional material cutting problems. Based on a BIM design platform and its corresponding secondary development platform, it automatically generates parametric BIM models of linear components. Using a data programming language, it designs an improved particle swarm optimization algorithm for one-dimensional material cutting problems. Computer-written program code automatically processes material cutting data and optimizes the cutting scheme. This invention has the following advantages compared to other existing technologies:

[0024] 1. This invention optimizes the design of linear component BIM models through interaction with a BIM design platform and a secondary development platform, ensuring the accuracy of material cutting data.

[0025] 2. This invention addresses the problem of large-scale one-dimensional material cutting by constructing a mathematical model with the goal of minimizing the waste material rate and the number of cutting combinations. This reduces waste material while also decreasing the number of cutting combinations.

[0026] 3. This invention proposes an improved particle swarm optimization algorithm for large-scale one-dimensional cutting problems. It calculates only the optimal cutting combination for the current cutting scale each time, continuously decomposing the large-scale one-dimensional cutting problem into smaller subproblems, thus shortening the processing scale of a single problem and the algorithm's solution time.

[0027] 4. This invention incorporates an automatic adjustment rule into the particle swarm optimization algorithm to constrain the inertia coefficient, thereby addressing the problem of premature convergence and failure to find the optimal solution.

[0028] 5. This invention optimizes the frequency of use of the optimal cutting combination. Under the premise of not exceeding the required quantity of parts to be cut, the current optimal cutting combination is used as much as possible, thereby reducing the cost of switching cutting combinations in actual engineering. Attached Figure Description

[0029] The present invention will now be described in further detail with reference to the accompanying drawings.

[0030] Figure 1 This is the overall flowchart of the present invention.

[0031] Figure 2 This is a flowchart for creating a BIM model of a linear component.

[0032] Figure 3 This is a flowchart for obtaining the optimal cutting combination under the current material cutting scale.

[0033] Figure 4 This is a flowchart of the constraint inertia coefficient in the particle swarm optimization algorithm. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] The above-mentioned invention utilizes the Revit BIM design platform and its secondary development platform Dynamo released by Autodedsk for optimized design. Simultaneously, Matlab is used for algorithm programming to complete the program code implementation. It should be understood that the relevant software platforms and computer languages ​​used herein are merely tools for implementing the method described in this invention and are not intended to limit the invention. The operation steps are as follows (see...). Figure 1 ):

[0036] Step 1: Create a BIM model of the main structure. In the BIM design platform, create a BIM model of the main structure. This main structure serves as the object for the linear component layout, used to determine the layout boundaries of the linear components.

[0037] Step Two: Optimize the layout of linear components and establish a BIM model for the linear components. Through the interaction between the BIM design platform and the secondary development platform, the parametric automatic generation of the linear components is completed (see...). Figure 2 );

[0038] 1. Create a parametric linear component family in the BIM design platform;

[0039] 2. Based on the construction drawings and structural drawings, create a layout information table for linear components;

[0040] 3. In the secondary development platform, call the corresponding nodes or custom function nodes to form a visual function script. The visual script includes:

[0041] 1) Pick the main structure created in step one in the secondary development platform;

[0042] 2) Process the layout information table of the created linear components and extract the layout information;

[0043] 3) To address common issues such as overlapping and collisions in linear components, custom function nodes can be called in the secondary development platform to set the layout rules for linear components;

[0044] 4) In the secondary development platform, call the corresponding nodes and custom function nodes to complete the position change of the linear component;

[0045] 4. The BIM design platform and the secondary development platform interact and collaborate. Visual function scripts run in the secondary development platform to drive parameterizable linear component families, and parameterized automatic generation of linear components is completed in the BIM design platform.

[0046] Step 3: Create a material cutting data form to establish the foundation for material cutting data. Based on the linear component BIM model, determine the material cutting information and summarize it into a material cutting data form. The material cutting information includes the raw material length of the linear component, the length and required quantity of the parts to be cut, the number of part specifications, and the total number of parts, etc.

[0047] 1. Split and number the BIM model of linear components, and extract the quantities.

[0048] 2. Determine the material cutting information, including the length of the raw material, the length and quantity of the parts to be cut, the number of parts by specifications, and the total number of parts.

[0049] 3. Generate a material cutting data form.

[0050] Step 4: Define the optimization objective and establish a mathematical model. Establish a mathematical model for the one-dimensional material cutting problem with the objectives of minimizing the residual material rate and the number of cutting combinations.

[0051] 1. Define the material cutting data information as follows:

[0052] The raw material length of the linear component is L, the number of part specifications is m, and the length and required quantity of each specification are l respectively. j d j (j = 1, 2, ..., m). Let the final material cutting scheme be X = {x1, x2, ..., xm}. n There are n possible cutting combinations, and each cutting combination has x. i ={x i1 ,x i2 ,...x im}, where x ij f represents the quantity of the j-th type of part in the i-th cutting combination, and f is the number of times the i-th cutting combination is used. i The remaining material length for each cutting combination is...

[0053] 2. A mathematical model is established with the objective functions of minimizing the overall residual material rate and minimizing the number of cutting combinations in the material cutting scheme, as shown below:

[0054]

[0055] minZ2=n

[0056]

[0057]

[0058] x ij ∈N,f i ∈N +

[0059] Step 5: Obtain the optimal cutting combination under the current material cutting scale. This involves an improved particle swarm optimization (PSO) algorithm. This algorithm uses the remaining material ratio as a metric and only seeks the optimal cutting combination under the current material cutting scale at each step, thereby continuously decomposing and reducing the one-dimensional material cutting problem into simpler subproblems. An automatic adjustment rule is added to the PSO algorithm to constrain the inertia coefficient, improving the algorithm's optimization ability while avoiding the occurrence of local optima (see [link to PSO algorithm]). Figure 3 );

[0060] 1. Import the blanking data form created in step 3 into the Matlab environment, and extract the raw material length L of the linear component and the length l of the part to be blanked. j and the demand for parts d j , where j = 1, 2, ..., m;

[0061] 2. Set algorithm parameters: particle population size is Q, maximum allowed number of iterations is K, and the inertia coefficient ω ranges from [ω...]. min ,ω max ];

[0062] 3. Randomly assign initial positions and velocities to all particles, and restrict particle velocity and position boundaries. Each particle's position corresponds to a specific cutting combination.

[0063] 4. Using the residual material rate as a metric, construct the fitness function as follows:

[0064]

[0065] Among them, Y i S represents the residual material rate of the cutting combination corresponding to the i-th particle. i The length of raw material used for the cutting combination corresponding to the i-th particle.

[0066] 5. In solving for the optimal cutting combination, due to the randomness of the particles, the usable length of a cutting combination corresponding to a certain particle may exceed the raw material length. To avoid this, the particles are optimized so that the usable length of the cutting combination corresponding to a particle continuously approaches the raw material length. The specific steps are as follows:

[0067] 1) Take S i =0, j=1;

[0068] 2) If S i +x ij l j If ≤L, then S i =S i +x ij l j .on the contrary, S i =S i +x ij l j ;

[0069] 3) Let j = j + 1, and repeat step 2; end when j is greater than m.

[0070] Where, x ij Let m be the number of parts of the j-th specification to be cut in the cutting combination corresponding to the i-th particle, and m be the number of specifications of the parts to be cut.

[0071] 6. Based on information sharing among particles in the population, compare the fitness values ​​of each particle and update the optimal position of the population and the optimal position of the individual.

[0072] 7. Establish automatic adjustment rules to constrain the inertia coefficient ω. During normal iteration, the coefficient decreases non-linearly. To prevent the population from getting trapped in local optima, the inertia coefficient of prematurely converged particles is increased (see [link to relevant documentation]). Figure 4 );

[0073] 1) To ensure that the particle has good global search capability in the early stage of iteration and good local optimization capability in the later stage, ω is made to decrease non-linearly during the iteration process, as shown below:

[0074]

[0075] in This represents the inertia coefficient of the i-th particle after the k-th iteration. When k is small, Close to ω max At this point, the algorithm has strong global optimization ability; as k increases, Gradually decrease to ω min The ability to find local optimization is enhanced.

[0076] 2) Simultaneously, to prevent the population from getting trapped in local optima, the inertia coefficient of prematurely converged particles is increased to help them escape local convergence. Given a particle population size of Q, assume the population size at the k-th iteration step... fitness value The fitness value of the individual's optimal position in the k-th iteration step is The fitness value of the optimal position in the population at the (k-1)th iteration step is Assume the population has been trapped in local convergence for t times, with an initial value of 0, and that the number of iterations after trapping in local convergence does not exceed T. The average fitness of all particles in the k-th iteration is... for The particles, taking the average value Y avg2 The specific steps to avoid the population getting trapped in local optima are as follows:

[0077] i)If The fitness value of the optimal position of the population in the k-th iteration step. At this point, t is reset to its initial value.

[0078] ii) If Let t = t + 1. When t ≥ T, for For particles that are selected for a given iteration, their inertia coefficient is increased in the next iteration step; for the remaining particles, their inertia coefficient continues to decrease nonlinearly. This is illustrated below:

[0079]

[0080] iii) For the next iteration step Apply conditional constraints, that is, when hour, otherwise

[0081] 8. Update the position and velocity of each particle in the population according to the following formula:

[0082]

[0083]

[0084] in, and These represent the j-th dimension components of the velocity and position of the i-th particle after k+1 iterations; The j-th dimension component represents the optimal position of the i-th particle after k iterations. c1 represents the j-th dimension component of the optimal position of the population after k iterations; c1 and c2 represent learning factors; r1 and r2 represent random functions that take values ​​in [0,1].

[0085] 9. When the number of population iterations reaches K, the cutting combination corresponding to the optimal position of the population is the best cutting combination under the current material feeding scale; otherwise, jump to step 5 to continue iterating.

[0086] Step Six: Optimize the frequency of use of the best cutting combination. Within the constraints, increase the frequency of use of the current best cutting combination as much as possible;

[0087] 1. Without exceeding the required quantity of parts, utilize the current optimal cutting combination as much as possible using the following formula:

[0088]

[0089] Where F represents the frequency of use of the optimal cutting combination; d j x represents the required quantity of parts of specification j to be cut; j This represents the number of parts of specification j to be cut in the optimal cutting combination.

[0090] Step 7: Update the data of the parts to be cut, reduce the scale of the cutting problem, and set the iteration judgment condition. Use the demand quantity of the parts to be cut as the judgment condition. When the demand quantity of the parts to be cut is 0, the algorithm ends; otherwise, return to step 5 to continue iterating.

[0091] 1. Update the data of the part to be cut using the following formula:

[0092] d j =d j -x j j = 1, 2, ..., m

[0093] 2. When the demand for the parts to be cut is not zero, continue to solve for the optimal cutting combination under the new cutting scale.

[0094] Step 8: Determine the final cutting scheme. After the iteration terminates, the optimal cutting combination and its corresponding usage frequency obtained from solving each subproblem are combined to form the final cutting scheme, which is the optimal solution to the large-scale one-dimensional cutting problem.

[0095] Step 9: Output data and export the material cutting optimization form. The material cutting optimization form includes the complete material cutting plan, the effective material utilization length for each cutting combination, the residual material length for each cutting combination, the residual material rate for each cutting combination, and the total residual material rate of the material cutting plan, etc.

[0096] Step 10: Case Comparison. Using known material cutting data, compare the algorithms of a one-dimensional large-scale optimized material cutting method for intelligent construction to verify its feasibility and advancement.

[0097] 1. To verify the feasibility and advancement of the algorithm proposed in this invention, it is compared with algorithms proposed in relevant literature, and verified using corresponding test data. Relevant literature is as follows:

[0098] [1] Zhang Meng, Chen Shijun, Li Jiabin, Liu Chaoyang. Research on one-dimensional blanking based on simulated annealing algorithm [J]. Computer Era, 2017(12):1-4. DOI:10.16644 / j.cnki.cn33-1094 / tp.2017.12.001.

[0099] [2] Li Bin, He Fei. An improved hybrid genetic algorithm for solving the one-dimensional material cutting problem [J]. Journal of Inner Mongolia University (Natural Science Edition), 2014, 45(03):245-250. DOI:10.13484 / j.nmgdxxbzk.20140304.

[0100] [3] Wei Liangliang, Ye Jiawei. An improved adaptive genetic algorithm for one-dimensional material cutting problem [J]. Journal of South China University of Technology (Natural Science Edition), 2003(06):26-30.

[0101] 2. The algorithm proposed in this invention is programmed in Matlab and runs on a computer with Windows 11 system and AMD Ryzen 5 4600H processor. The particle population size is set to 50, the maximum number of iterations is allowed to be 300, and the inertia coefficient ω ranges from [0.4, 0.8].

[0102] 3. Example 1: Reference [1] proposes an improved simulated annealing algorithm, which solves the one-dimensional blanking problem by adjusting the mutation operator and proposing an improved decoding strategy. The experimental data are shown in Table 1, including 36 specifications of parts to be blanked, with a raw material length of 305m.

[0103] Table 1. Experimental data used in Example 1

[0104]

[0105]

[0106] The cutting scheme given in reference [1] is shown in Table 2. This cutting scheme includes 7 cutting combinations, requires 7 raw materials, and has a total leftover length of 31m and a leftover rate of 1.45%. The cutting scheme given by the algorithm proposed in this invention is shown in Table 3. This cutting scheme includes 7 cutting combinations, requires 7 raw materials, has a total leftover length of 31m, and a leftover rate of 1.45%. The difference is that in the cutting scheme obtained by the method proposed in this invention, the leftover of the first six cutting combinations is 0.

[0107] Table 2. Material cutting scheme obtained from reference [1]

[0108]

[0109] Table 3. Material cutting scheme obtained by the algorithm proposed in this invention

[0110]

[0111] Example 2: Reference [2] proposes an improved hybrid genetic algorithm, which constructs a local search algorithm to assist the genetic algorithm in solving the large-scale material cutting problem. The experimental data comes from reference [3], as shown in Table 4, including 20 specifications of parts to be cut, with a raw material length of 12m.

[0112] Table 4. Experimental data used in Example 2

[0113]

[0114] The cutting scheme given in reference [2] is shown in Table 5. This cutting scheme includes 24 cutting combinations, requires a total of 24 raw materials, and the total length of the leftover material is 7468 mm, with a leftover rate of 2.59%. The cutting scheme given by the algorithm proposed in this invention is shown in Table 6. This cutting scheme includes 14 cutting combinations, requires a total of 24 raw materials, and the total length of the leftover material is 6982 mm, with a leftover rate of 2.42%.

[0115] Table 5. Material cutting scheme derived from reference [2]

[0116]

[0117]

[0118] Table 6. Material cutting scheme obtained by the algorithm proposed in this invention

[0119]

[0120] This invention is applicable to the solution process of material cutting optimization problems for all linear components. Regardless of the structural form of the linear component or the value of the relevant parameters, this invention can be used to optimize the material cutting scheme and guide the material cutting process.

[0121] The above description merely illustrates preferred embodiments of the present invention, and while the description is relatively specific and detailed, it should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications, improvements, and substitutions without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A one-dimensional large-scale optimization cutting method for smart construction, characterized in that: The method comprises the following steps: Step one: establishing a main structure BIM model; creating a BIM model for the main structure in the BIM design platform, which is used as an object for linear component arrangement to determine the arrangement boundary of linear components; Step two: optimizing linear component arrangement and establishing a linear component BIM model; through the interaction between the BIM design platform and the secondary development platform, the parametric automatic generation of linear components is completed; Step three: creating a cutting data form and establishing a cutting data basis; based on the linear component BIM model, cutting information is determined and summarized to form a cutting data form; the cutting information includes the raw material length of linear components, the length and demand quantity of parts to be cut, the part specification number, and the total number of parts; Step four: determining an optimization target and establishing a mathematical model; a mathematical model of one-dimensional cutting problem is established with the minimum scrap rate and the minimum cutting combination number as the target; Step five: get the best cutting combination under the current blanking scale; involves an improved particle swarm algorithm, which adds automatic adjustment rules to constrain the inertia coefficient in the particle swarm algorithm, and takes nonlinear decrease as the change trend in the iteration process, while increasing the inertia coefficient of the particle for premature convergence. The algorithm takes the scrap rate as the measurement standard and only gets the best cutting combination under the current blanking scale each time. In this way, the one-dimensional blanking problem is continuously decomposed and degraded into simple sub-problems, and the inertia coefficient calculation meets the following formula with nonlinear decrease as the change trend: , denotes the inertia coefficient of the denotes the maximum number of iterations​​​​ Step six: optimizing the use frequency of the best cutting combination; within the constraint condition range, the use frequency of the current best cutting combination is increased as much as possible; Step seven: updating the data of parts to be cut, reducing the size of the cutting problem, and setting an iteration judgment condition; when the demand quantity of parts to be cut is 0, the algorithm ends, otherwise, it returns to step five for iteration; Step eight: determining the final cutting scheme; after iteration termination, the best cutting combination and the corresponding use frequency solved by each sub-problem are combined to form the final cutting scheme, which is the optimization solution of the large-scale one-dimensional cutting problem; Step nine: outputting data and exporting a cutting optimization form; the cutting optimization form includes the complete cutting scheme, the effective raw material length of each cutting combination, the scrap length of each cutting combination, the scrap rate of each cutting combination, and the total scrap rate of the cutting scheme.

Images