A pressure-bearing sealing and partitioning device for pressure culverts and its intelligent design method

Abstract

This invention relates to a pressure-bearing sealing and separating device for pressure culverts and its intelligent design method. The device includes two side steel plates, multiple circumferential steel plates, multiple corner bracing steel plates, two edge sealing steel plates, an inlet steel pipe, and a flange. The method includes: determining the optimization objective and design variables, with the dual objectives of maximizing the safety factor of the separating device and minimizing the amount of steel used; constructing a three-dimensional model and performing finite element analysis to obtain sample data; constructing and training a Kriging surrogate model; using the MOPSO algorithm for multi-objective optimization to obtain a Pareto optimal solution set; selecting a scheme from the Pareto optimal solution set and performing finite element verification to determine the final design scheme. This invention has a simple structure, significantly reduces the computational cost of high-dimensional parameter optimization, avoids local optimum traps, can output a complete Pareto optimal solution set, and is adaptable to the diverse needs of different engineering conditions, belonging to the field of hydraulic engineering structural design and intelligent optimization technology.

Description

Technical Field

[0001] This invention relates to the field of hydraulic engineering structural design and intelligent optimization technology, specifically to a separation device suitable for pressure culverts, and also to a smart design method for the device based on the Kriging surrogate model and the MOPSO algorithm. It can be widely applied in hydraulic engineering projects such as water resource allocation projects and pump station renovations, where channel separation and multi-objective structural optimization are required. Background Technology

[0002] In the design of complex water conservancy systems, multi-objective optimization is a core element in improving structural performance and engineering benefits. As a key structure in water resource allocation projects and pumping station systems, the functional integrity and operational efficiency of pressure culverts directly affect the stable operation of the entire water conservancy system. However, existing pressure culverts often suffer from single-channel design flaws. When supplying water to multiple main lines with different head requirements, independent scheduling cannot be achieved, leading to energy waste in low-head main lines. Furthermore, mutual interference of water flows within the culvert generates undesirable flow patterns, increasing head loss and equipment load. Long-term operation can also accumulate unbalanced structural stress, posing a safety hazard.

[0003] Therefore, it is necessary to modify the pressure culvert by adding a separation device to divide the original single flow channel into more independent flow channels, and to set up independent inlets and outlets for each flow channel to connect to water pumps with different head and users respectively.

[0004] Traditional pressure box culvert modification design relies on engineering experience and single-objective optimization approaches, which have many limitations: First, traditional multi-objective optimization methods have high computational costs and are prone to getting trapped in local optima in high-dimensional design parameter spaces, making it difficult to effectively search for the global optimum; Second, the output results are singular and cannot provide a complete Pareto optimal solution set, making it difficult to adapt to diverse application needs under different working conditions; Third, they lack the ability to handle high-dimensional, nonlinear, and strongly coupled optimization variables and constraints, and the ideal mathematical model is out of touch with complex engineering realities, limiting the engineering applicability of the optimization results.

[0005] To overcome the shortcomings of traditional methods, intelligent optimization has become the development direction for multi-objective optimization of steel structures in water conservancy projects. The combination of surrogate models and intelligent optimization algorithms has been proven to be an effective approach. However, existing technologies lack dedicated design schemes for pressure box culvert separation devices, and existing intelligent optimization methods are mostly developed based on general civil engineering structures, making it difficult to directly adapt to the special mechanical properties and multi-objective design requirements of pressure box culverts. This results in problems such as low efficiency and difficulty in balancing safety and economy in the design of separation devices.

[0006] Therefore, there is an urgent need for a dedicated pressure culvert separation device that is easy to implement intelligent design. At the same time, it is necessary to develop an efficient intelligent structural performance design method to achieve synergistic optimization of maximizing the safety factor of the separation device and minimizing the amount of steel used, thereby addressing the shortcomings of existing technologies. Summary of the Invention

[0007] In view of the technical problems existing in the prior art, the purpose of this invention is to provide a pressure tank culvert pressure-bearing sealing and separation device that is easy to implement with intelligent design.

[0008] Another objective of this invention is to provide a smart design method for a pressure box culvert pressure-bearing sealing separation device, which efficiently achieves the synergistic optimization of maximizing the safety factor of the separation device and minimizing the amount of steel used.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] A pressure-bearing sealing and partitioning device for a pressure box culvert includes two side steel plates, multiple circumferential steel plates, multiple corner bracing steel plates, two sealing steel plates, an inlet steel pipe, and a flange. The multiple circumferential steel plates are evenly distributed along the circumferential direction, fixing the inlet steel pipe to the center of the pressure box culvert's sealing section. Two annular side steel plates are fixed parallel to each other on both sides of the circumferential steel plates, with the inner ring connected to the inlet steel pipe and the outer ring connected to the inner wall of the pressure box culvert. An annular sealing steel plate is obliquely fixed between the inner wall of the pressure box culvert and the outer side of one side steel plate, and the three are reinforced by a circumferentially distributed corner bracing steel plate. The flange is located at one end of the inlet steel pipe.

[0011] As a preferred embodiment, the pressure box culvert has a circular sealing section, an annular side steel plate, a rectangular circumferential steel plate, a right-angled triangular corner bracing steel plate, an annular sealing steel plate, and a circular inlet steel pipe; the flange includes an annular flange plate and a circular flange cover, the flange plate is fixed to one end of the inlet steel pipe, and the flange cover is detachably connected to the flange plate.

[0012] As a preferred embodiment, the length of the inlet steel pipe is greater than the horizontal distance between the two side steel plates; one end of the inlet steel pipe is flush with one side steel plate, the other end is connected to a flange, and the inner ring of the other side steel plate is fixed to the outer wall of the inlet steel pipe.

[0013] As a preferred embodiment, the inner wall, circumferential steel plate, and inlet steel pipe of the pressure box culvert are sequentially welded together; the inner wall, circumferential steel plate, and inlet steel pipe of the pressure box culvert are all welded together with the side steel plates; the inner wall, sealing steel plate, and side steel plate of the pressure box culvert are sequentially welded together; the inner wall, sealing steel plate, and side steel plate of the pressure box culvert are all welded together with the angle brace steel plate; the flange is welded to the inlet steel pipe, and the flange is bolted to the flange cover.

[0014] A smart design method for a pressure-bearing sealing and partitioning device for a pressure culvert includes the following steps: S1: Determine the optimization objective and design variables, with the dual objectives of maximizing the safety factor of the partitioning device and minimizing the amount of steel used; S2: Construct a three-dimensional model and perform finite element analysis to obtain sample data; S3: Construct a Kriging surrogate model and train and verify it; S4: Use the MOPSO algorithm for multi-objective optimization to obtain the Pareto optimal solution set; S5: Select a scheme from the Pareto optimal solution set and perform finite element verification to determine the final design scheme.

[0015] As a preferred option, in step S1, the design variables are steel plate thickness, horizontal spacing, circumferential fraction, and flange thickness; wherein, steel plate thickness includes the thickness of side steel plates, circumferential steel plates, corner brace steel plates, edge sealing steel plates, and inlet hole steel pipes; circumferential fraction refers to the number of circumferential steel plates; and flange thickness refers to the individual thickness of the flange and flange cover.

[0016] As a preferred option, in step S2, four parameters are selected from the four design variables within the empirical range, and 256 working conditions are generated using orthogonal experimental design. Finite element analysis is then performed by applying unilateral hydrostatic pressure.

[0017] As a preferred option, in step S3, a Kriging proxy model is constructed using a Python program; the sample data obtained in step S2 is divided into a training set and a validation set according to a certain ratio, the Kriging proxy model is trained, and the accuracy is verified.

[0018] As a preferred option, in step S4, a MOPSO program is written using MATLAB, the MOPSO algorithm parameters are set, the objective function value is calculated using the Kriging model, the optimized objective function is used as the fitness function, and the Pareto optimal solution set is searched by updating the individual optimal solution and the global optimal solution.

[0019] As a preferred option, in step S5, candidate solutions are selected from the Pareto optimal solution set and finite element verification is performed to ensure that the safety factor is ≥2.0.

[0020] The present invention has the following advantages:

[0021] 1. The separation device has a simple structure, clear and quantifiable design variables, and is easy to combine with intelligent design methods. It can quickly obtain the optimal design solution through parametric modeling and intelligent optimization, thereby significantly improving design efficiency and structural safety.

[0022] 2. The separation device has a simple structure, mainly composed of side steel plates, circumferential steel plates, corner bracing steel plates, edge sealing steel plates, entry hole steel pipes and flanges, etc. The overall structure is symmetrical, the stress is uniform, and it is easy to process and manufacture.

[0023] 3. The separating device of the present invention adopts a symmetrical structural design, and its mechanical response characteristics can be fully described by only four key design variables (steel plate thickness, horizontal spacing, circumferential equal fraction, and flange thickness), which greatly simplifies the complexity of the optimization problem and lays a good variable foundation for subsequent proxy model construction and intelligent optimization.

[0024] 4. This invention uses a Kriging model to construct a surrogate model, fully considering the strong nonlinear characteristics of the mechanical response of the pressure tank culvert separation device and the small sample size of 256 sets of finite data. The Kriging model not only achieves accurate interpolation at sample points but also provides estimations of prediction uncertainties. Compared to traditional response surface methodology or neural network models, it is more suitable for the data characteristics of this invention, effectively ensuring the reliability of the optimization results.

[0025] 5. This invention uses the MOPSO algorithm for multi-objective optimization. Taking advantage of its fast convergence speed and uniform Pareto front distribution, it can obtain a high-quality optimal solution set in a short time, allowing engineers to flexibly choose according to engineering preferences, thus achieving the best balance between safety and economy.

[0026] 6. The intelligent design method integrates the Kriging surrogate model and the MOPSO algorithm, which significantly reduces the computational cost of high-dimensional parameter optimization, avoids local optimum traps, and can output a complete Pareto optimal solution set, adapting to the diverse needs of different engineering conditions. The method is trained and validated with 256 sets of working condition data. The model has high prediction accuracy, with a safety factor prediction accuracy of ≥80% and a steel quantity prediction accuracy of ≥97%, providing precise guidance for the design of separation devices.

[0027] 7. Independent scheduling of multiple flow channels in the pressure culvert is realized, which solves the problem of energy waste in water supply from main lines with different head, while optimizing the water flow pattern, reducing head loss and equipment load, and reducing the safety hazards of unbalanced structural stress. Attached Figure Description

[0028] Figure 1 This is a plan view of the separating device of the present invention.

[0029] Figure 2a yes Figure 1 Section II.

[0030] Figure 2b yes Figure 1 Section II-II.

[0031] Figure 3 This is a flowchart of the intelligent design method of the present invention.

[0032] Figure 4 This is a flowchart of the Kriging agent model construction and training process.

[0033] Figure 5 This is the flowchart of the MOPSO multi-objective optimization algorithm.

[0034] Figure 6a This is a comparison chart of the calculated and predicted values ​​of the safety factor during training.

[0035] Figure 6b This is a comparison chart of the calculated and predicted values ​​of steel quantity training.

[0036] Figure 6c This is a comparison chart of the calculated and predicted values ​​for the safety factor verification.

[0037] Figure 6d This is a comparison chart of the calculated and predicted values ​​for steel quantity verification.

[0038] Figure 7 This is the Pareto front plot obtained by optimizing the MOPSO algorithm.

[0039] 1-Side steel plate, 2-Circumferential steel plate, 3-Angle brace steel plate, 4-Edge sealing steel plate, 5-Inlet steel pipe, 6-Flange, 61-Flange, 62-Flange cover, 7-Pressure box culvert. Detailed Implementation

[0040] The present invention will now be described in further detail with reference to specific embodiments.

[0041] Example 1

[0042] Figure 1 , Figure 2a , Figure 2b The image shows a pressure-bearing sealing and partitioning device for a pressure box culvert, which is installed in an existing pressure box culvert for retrofitting existing pressure box culverts.

[0043] This embodiment of the device is applied to the renovation of a pressure culvert in a water resource allocation project in eastern Guangdong. This project needs to supply water to two branch lines with a head difference exceeding 10 meters. The original pressure culvert was a single-channel system and could not be independently managed. After installing the device of this invention, the pressure culvert is divided into... Figure 1 The left and right flow channels are shown.

[0044] A pressure-bearing sealing and partitioning device for a pressure box culvert includes two side steel plates, multiple circumferential steel plates, multiple corner bracing steel plates, two sealing steel plates, an inlet steel pipe, and a flange. The multiple circumferential steel plates are evenly distributed along the circumferential direction, fixing the inlet steel pipe to the center of the pressure box culvert's sealing section. Two annular side steel plates are fixed parallel to each other on both sides of the circumferential steel plates, with the inner ring connected to the inlet steel pipe and the outer ring connected to the inner wall of the pressure box culvert. An annular sealing steel plate is obliquely fixed between the inner wall of the pressure box culvert and the outer side of one side steel plate, and the three are reinforced by a circumferentially distributed corner bracing steel plate. The flange is located at one end of the inlet steel pipe.

[0045] The pressure box culvert has a circular sealing section, an annular side steel plate, a rectangular circumferential steel plate, a right-angled triangular corner bracing steel plate, an annular sealing steel plate, and a circular inlet pipe. The flange includes an annular flange plate and a circular flange cover. The flange plate is fixed to one end of the inlet pipe, and the flange cover is detachably connected to the flange plate.

[0046] The length of the steel pipe entering the hole is greater than the horizontal distance between the two side steel plates; one end of the steel pipe entering the hole is flush with one side steel plate, and the other end is connected to a flange; the inner ring of the other side steel plate is fixed to the outer wall of the steel pipe entering the hole.

[0047] The inner wall, circumferential steel plate, and inlet steel pipe of the pressure box culvert are welded together in sequence; the inner wall, circumferential steel plate, and inlet steel pipe of the pressure box culvert are all welded together with the side steel plates; the inner wall, sealing steel plate, and side steel plate of the pressure box culvert are welded together in sequence; the inner wall, sealing steel plate, and side steel plate of the pressure box culvert are all welded together with the angle brace steel plate; the flange is welded to the inlet steel pipe, and the flange is bolted to the flange cover.

[0048] All components are welded from Q345CZ steel, which has an elastic modulus of 206 GPa, a Poisson's ratio of 0.3, a density of 7850 kg / m³, and a yield strength of 345 MPa. The specific structural parameters of this embodiment are as follows:

[0049] Side steel plate 1: The thickness is the same as that of the circumferential steel plate, which is 12mm, and it is welded perpendicularly to the circumferential steel plate.

[0050] Circumferential steel plate 2: 12mm thick, divided into 8 equal parts in the circumference, used for structural support to achieve effective separation of the flow channel.

[0051] Angle brace steel plate 3: 12mm thick, 6 evenly distributed, welded to the connection between the circumferential steel plate and the side steel plate to enhance structural stability.

[0052] Edge sealing steel plate 4: 12mm thick, welded to the edge of the circumferential steel plate to ensure the sealing of the flow channel.

[0053] Manhole steel pipe 5: 12mm thick, with a 600mm diameter circular manhole located in the middle of a circumferential steel plate for easy maintenance and repair later.

[0054] Flanges: including flange plates and flange covers, used to ensure the sealing and structural strength of the connection.

[0055] Flange 61: 38mm thick, welded to the end of the steel pipe with inlet hole.

[0056] Flange cover 62: 38mm thick, fixed to the flange with M24 bolts.

[0057] Current pressure box culvert 7: thickness 24mm.

[0058] Horizontal spacing: The distance between adjacent side steel plates is 1246mm.

[0059] The separation device is installed inside the pressure culvert to achieve independent separation of the two flow channels. During operation, the water supply can be independently controlled according to the head requirements of the two main lines. The low-head main line does not need to operate at full load along with the high-head main line, reducing energy consumption by more than 30%; the water flow pattern is stable, the head loss is reduced by 15%, and the equipment load is significantly reduced; the structure is subjected to uniform stress, the maximum Mises stress is lower than the material yield strength, and the safety factor reaches 2.49, meeting the engineering safety requirements.

[0060] Example 2

[0061] Figure 3 As shown, a smart design method for a pressure-bearing sealing and partitioning device for a pressure culvert includes the following steps:

[0062] Step S1: Determine the optimization objective and design variables.

[0063] Optimization objectives: maximize safety factor (objective 1), minimize steel consumption (objective 2).

[0064] Design variables: steel plate thickness (A: 12mm, 16mm, 20mm, 24mm), horizontal spacing (B: 1000mm, 1200mm, 1400mm, 1600mm), circumferential fraction (C: 4, 8, 12, 16), flange thickness (D: 12mm, 26mm, 48mm, 76mm). A four-factor, four-level orthogonal experimental design was adopted, with a total of 256 working conditions.

[0065] Based on the experience of those skilled in the art, the steel plate thickness is 12mm~24mm, the circumferential division is 4~16, the horizontal spacing is 1000mm~1600mm, and the flange thickness is 12mm~76mm. Therefore, the operating conditions are selected within this range.

[0066] The steel plate thickness includes the thickness of the side plates, circumferential plates, corner braces, edge sealing plates, and manhole steel pipes. Horizontal spacing is the distance between two side plates. Circumferential division refers to the number of circumferential plates. Flange thickness refers to the individual thickness of the flange and flange cover.

[0067] Step S2: 3D modeling and finite element analysis.

[0068] A 3D model of the separation device under 256 working conditions was established using CAD software to accurately reproduce the dimensions and connection relationships of each component and precisely reflect the inner diameter of the pressure culvert. Finite element analysis was performed using ANSYS APDL: the SHELL181 shell element was selected, and the properties of Q345CZ steel were defined (elastic modulus 206 GPa, Poisson's ratio 0.3, density 7850 kg / m³, yield strength 345 MPa). Local mesh refinement was applied to key areas such as contact interfaces and connection points (e.g., the connection between the circumferential steel plate and the inlet pipe, and the welded joints of the corner braces), with the mesh size controlled below 50 mm. A unilateral hydrostatic pressure of 0.5 MPa (the most unfavorable working condition) was applied, constraining all degrees of freedom at the bottom and the connection surface with the foundation. The maximum Mises stress and steel volume for each working condition were then obtained.

[0069] The safety factor is calculated according to the formula:

[0070]

[0071] Where Fos is the safety factor. The yield strength is 345 MPa. This represents the maximum Mises stress.

[0072] The steel quantity is calculated according to the formula:

[0073] T=V×ρ

[0074] Where T is the amount of steel, V is the total volume of steel, and ρ is 7850 kg / m³.

[0075] Step S3: Kriging Proxy Model Construction and Validation. A proxy model is constructed using Python.

[0076] The general expression for the Kriging proxy model is as follows:

[0077] F(x) = g(x) + K(x)

[0078] In the formula, g(x) is a regression polynomial determined by the sample point information, also known as a global trend model; K(x) is a polynomial with a mean of 0 and a variance of σ. 2 The static stochastic process involves random variables that exhibit a certain correlation (or covariance) at different locations in the design space. The covariance relationship is as follows:

[0079] COV[K(x i ),K(x j )]=σ 2 S[S(x i ,x j )]

[0080] In the formula, S(x)i ,x j The correlation function depends only on the sample point x. i and sample point x j The correlation between the Euclidean distances decreases as the distance increases. For example, the correlation function equals 1 when the distance is zero and equals 0 when the distance is infinite.

[0081] The calculation results of 256 working conditions were divided into a training set and a validation set in a 7:3 ratio, which were used for model training and accuracy verification, respectively. Figures 6a-6d This is a graph used to verify the accuracy of the Kriging surrogate model. Figure 6a For comparison of the safety factor training set, Figure 6b For comparison of steel quantity training set, Figure 6c For the safety factor verification set comparison, Figure 6d For comparison with the steel quantity verification set. Model accuracy verification: The safety factor model has MSE=0.0084, RMSE=0.0915, MAE=0.0656, R²=0.9975; the steel quantity model has MSE=0.0001, RMSE=0.0008, MAE=0.0006, R²=0.9995, both meeting the accuracy requirements.

[0082] Step S4: Multi-objective optimization. The MOPSO program is written in MATLAB, and parameters such as particle population size and maximum number of iterations are set. The particle position represents the combination of design variables. The trained Kriging surrogate model is substituted into the algorithm. By updating the individual optimal solution and the global optimal solution, the Pareto optimal solution set is searched.

[0083] Let the search space of the multi-objective optimization problem be D-dimensional, the number of particles in the particle swarm be M, and the maximum number of iterations for the optimization algorithm be T. In the next iteration, the particles The velocity and position are denoted as follows: and ;particle The optimal position of an individual is denoted as The optimal position of an individual particle swarm is denoted as . That is, the globally optimal solution; Both are D-dimensional vectors, and their expressions are:

[0084] )

[0085]

[0086]

[0087]

[0088] The particle's velocity and position updates are related to the individual optimal solution and the global optimal solution from the previous iteration. By combining individual and group experience, the particle can maintain a balance between global search and local refinement. In the (t+1)th iteration, the particle's velocity and position update method in the dimension is as follows:

[0089] )

[0090]

[0091] In the formula, ω is the inertia weight, representing the particle's velocity inertia; c1 is the individual learning factor, and c2 is the social learning factor, controlling the particle's tendency to move towards the particle's optimal solution and the global optimal solution; r1 and r2 are random numbers between [0, 1], introducing randomness to avoid rigidity in the solution process; i is the particle number, i = 1, 2, ..., M; j is the solution dimension, j t is the iteration number, t To prevent particles from moving too fast and exceeding the search space, particle velocity... The update is limited by a maximum value VMAX. If the particle velocity exceeds VMAX, the velocity direction remains unchanged, and the velocity value is the maximum value VMAX.

[0092] Figure 7 This is a Pareto front plot obtained from the MOPSO algorithm optimization. The first objective is the reciprocal of the safety factor, and the second objective is the amount of steel. The Pareto front is the set of all Pareto optimal solutions in a multi-objective optimization problem, intuitively showing the "trade-off relationship" between objectives. The inflection point of its curve is the "critical decision boundary" in Pareto objective optimization.

[0093] Step S5: Optimization result verification and scheme determination.

[0094] The previous step yielded the Pareto optimal solution set. The optimal solution was selected from the Pareto front points: steel plate thickness 12mm, horizontal spacing 1246mm, circumferential division into 8 equal parts, and flange thickness 38mm.

[0095] A 3D model of the scheme was created and imported into ANSYS APDL for finite element analysis to ensure a safety factor ≥ 2.0 and that the amount of steel used meets the economic requirements of the project.

[0096] The actual calculated amount of steel was 5.814t, with a safety factor of 2.49. The errors between this and the predicted value from the surrogate model (5.962t of steel, with a safety factor of 3.15) were 2.5% and 20%, respectively. The safety factor meets the design requirement of ≥2.0, and the amount of steel used is optimal in terms of economy.

[0097] The optimal design scheme was finally determined, achieving a synergistic optimization of safety and economy.

[0098] The intelligent design method of this invention significantly improves design efficiency by combining proxy models and intelligent algorithms. Compared with traditional design methods, it shortens the optimization cycle and achieves synergistic optimization of safety and economy, which can provide a reference for the structural design of similar water conservancy projects.

[0099] Figure 4 This is a flowchart of the Kriging agent model construction and training process.

[0100] like Figure 4 As shown, step S3 includes the following steps:

[0101] Step S3-1: Use Python code to build a proxy model and divide the 256 sets of data obtained in step S2 into a training set (179 sets) and a validation set (77 sets) in a 7:3 ratio.

[0102] Step S3-2: Check the model calculation results to see if the accuracy meets the standard. If the accuracy meets the standard, import the model into step S4 for optimization. If the accuracy does not meet the standard, return to step S3-1 to retrain until the accuracy meets the standard.

[0103] Figure 5 This is the flowchart of the MOPSO multi-objective optimization algorithm.

[0104] like Figure 5 As shown, step S4 includes the following steps:

[0105] Step S4-1: Algorithm initialization settings: Particle population size 100, maximum iterations 100, inertia weight ω=0.8, individual learning factor. =1, social learning factor =2.

[0106] Step S4-2: Calculate the objective function value of each particle in the population in each dimension, use the optimized objective function as the fitness function, and calculate the fitness of each particle according to the specific expression of the optimized objective function.

[0107] Step S4-3: Update the individual best position and the global best position.

[0108] Step S4-4: Repeat steps S4-2 to S4-3 until the maximum number of iterations reaches the preset number of iterations.

[0109] Step S4-5: Output the Pareto optimal solution.

[0110] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A pressure-bearing sealing and separating device for a pressure box culvert, characterized in that: It includes two side steel plates, multiple circumferential steel plates, multiple corner bracing steel plates, two edge sealing steel plates, an inlet steel pipe, and a flange. The multiple circumferential steel plates are evenly distributed along the circumferential direction, fixing the inlet steel pipe in the middle of the sealing section of the pressure box culvert. The two annular side steel plates are fixed parallel to each other on both sides of the circumferential steel plates, with the inner ring connected to the inlet steel pipe and the outer ring connected to the inner wall of the pressure box culvert. An annular edge sealing steel plate is obliquely fixed between the inner wall of the pressure box culvert and the outer side of one side steel plate, and the three are reinforced by a circumferentially evenly distributed corner bracing steel plate. The flange is set at one end of the inlet steel pipe.

2. The pressure-bearing sealing and separating device for a pressure box culvert according to claim 1, characterized in that: The pressure box culvert has a circular sealing section, an annular side steel plate, a rectangular circumferential steel plate, a right-angled triangular corner bracing steel plate, an annular sealing steel plate, and a circular inlet pipe. The flange includes an annular flange plate and a circular flange cover. The flange plate is fixed to one end of the inlet pipe, and the flange cover is detachably connected to the flange plate.

3. A pressure-bearing sealing and separating device for a pressure box culvert according to claim 2, characterized in that: The length of the steel pipe entering the hole is greater than the horizontal distance between the two side steel plates; one end of the steel pipe entering the hole is flush with one side steel plate, and the other end is connected to a flange; the inner ring of the other side steel plate is fixed to the outer wall of the steel pipe entering the hole.

4. A pressure-bearing sealing and separating device for a pressure box culvert according to claim 2, characterized in that: The inner wall, circumferential steel plate, and inlet steel pipe of the pressure box culvert are welded together in sequence; the inner wall, circumferential steel plate, and inlet steel pipe of the pressure box culvert are all welded together with the side steel plates; the inner wall, sealing steel plate, and side steel plate of the pressure box culvert are welded together in sequence; the inner wall, sealing steel plate, and side steel plate of the pressure box culvert are all welded together with the angle brace steel plate; the flange is welded to the inlet steel pipe, and the flange is bolted to the flange cover.

5. A smart design method for a pressure-bearing sealing and separating device for a pressure box culvert according to any one of claims 1 to 4, characterized in that, Includes the following steps: S1: Determine the optimization objectives and design variables, with the dual objectives of maximizing the safety factor of the separation device and minimizing the amount of steel used. S2: Construct a three-dimensional model and perform finite element analysis to obtain sample data; S3: Construct a Kriging agent model and train and validate it; S4: Use the MOPSO algorithm for multi-objective optimization to obtain the Pareto optimal solution set; S5: Select a scheme from the Pareto optimal solution set and perform finite element verification to determine the final design scheme.

6. The intelligent design method for a pressure-bearing sealing and separating device for a pressure box culvert according to claim 5, characterized in that, In step S1, the design variables are steel plate thickness, horizontal spacing, circumferential fraction, and flange thickness; where steel plate thickness includes the thickness of side steel plates, circumferential steel plates, corner brace steel plates, edge sealing steel plates, and inlet hole steel pipes; circumferential fraction refers to the number of circumferential steel plates; flange thickness refers to the individual thickness of the flange and flange cover.

7. The intelligent design method for a pressure-bearing sealing and separating device for a pressure box culvert according to claim 5, characterized in that, In step S2, four parameters are selected from the empirical range for the four design variables. 256 working conditions are generated using orthogonal experimental design, and finite element analysis is performed by applying unilateral hydrostatic pressure.

8. The intelligent design method for a pressure-bearing sealing and separating device for a pressure box culvert according to claim 5, characterized in that, In step S3, a Kriging proxy model is constructed using a Python program; the sample data obtained in step S2 is divided into a training set and a validation set according to the proportion, the Kriging proxy model is trained and its accuracy is verified.

9. A smart design method for a pressure-bearing sealing and separating device for a pressure box culvert according to claim 5, characterized in that, In step S4, the MOPSO program is written using MATLAB, the MOPSO algorithm parameters are set, the objective function value is calculated using the Kriging model, the optimized objective function is used as the fitness function, and the Pareto optimal solution set is searched by updating the individual optimal solution and the global optimal solution.

10. A smart design method for a pressure-bearing sealing and separating device for a pressure box culvert according to claim 5, characterized in that, In step S5, candidate solutions are selected from the Pareto optimal solution set and finite element verification is performed to ensure that the safety factor is ≥2.0.

Images