Intelligent optimization design method for bridge prestressed steel beam based on multi-factor coupling

Abstract

The invention relates to the technical field of intelligent optimization design, in particular to a bridge prestressed steel beam intelligent optimization design method based on multi-factor coupling, and the method comprises the steps: carrying out the quantitative modeling of a plurality of parameters in the arrangement process of a bridge prestressed steel beam, achieving the precise expression of controllable variables, and improving the design precision of the bridge prestressed steel beam. The overall search capability of the arrangement scheme under multi-target tradeoff is enhanced, the rationality and stability of steel beam distribution are integrally improved, the structure local overload risk is reduced, an energy function sequence is constructed through indexes such as a midspan displacement value, a section stress ratio and the total length, the quality of a solution set is controlled by utilizing an energy difference value, intervention and screening of a residual value are introduced, and the stability of the structure is improved. The robustness of the design under the multivariable condition is enhanced, through logic judgment of displacement and counterforce, unqualified arrangement is eliminated, feasible arrangement output under structural consistency is achieved, structural efficiency and material use economy are enhanced, and design reliability, optimization solution efficiency and structural response control precision are remarkably improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of intelligent optimization design, in particular to a bridge prestressed steel beam intelligent optimization design method based on multi-factor coupling. BACKGROUND

[0002] The core of the intelligent optimization design technical field is to use computer-aided design and intelligent algorithms to model various influencing factors into mathematical models, and to perform global search and iterative correction on design parameters under an optimization solving framework. Intelligent optimization design is not only suitable for performance improvement of structures in the fields of machinery, civil engineering and electronic engineering, but also realizes automatic generation and screening of design schemes through parameterized modeling, multi-objective optimization methods and evolutionary algorithms, thereby reducing experience dependence and improving design efficiency and scientificity of results.

[0003] A bridge prestressed steel beam intelligent optimization design method based on multi-factor coupling aims to comprehensively consider multiple factors such as load distribution, material strength, structural stiffness and construction constraints, to systematically optimize the arrangement scheme of the prestressed steel beam by establishing a multi-objective optimization model, and to realize more reasonable stress of the bridge structure, higher utilization rate of construction materials and lower cost under the premise of ensuring safety, so as to obtain a suitable steel beam arrangement scheme meeting the requirements of mechanical properties, economy and durability.

[0004] The prior art mainly relies on an optimization solving framework with structural parameters as input, and realizes optimization of the structural parameters through a multi-objective function and an evolutionary strategy. In actual engineering, there is a high coupling and nonlinear relationship between the stress boundary conditions, construction constraints and material properties. The prior art simplifies multiple factors in the modeling stage, and often takes a single parameter as an optimization target, which fails to fully reveal the interaction logic between variables, leading to invalidation of the optimization result under real constraints. Ignoring the internal logic between the tensioning sequence and the stress distribution may cause excessive deformation of the cross section and local overlimiting. In the prior art, the screening mechanism for optimization solutions is generally based on static function evaluation, lacks adaptive disturbance and correction processes, and thus the convergence result is prone to fall into local optimum, lacks diversity and stability, increases design verification period and rework cost, and affects engineering progress and safety guarantee. SUMMARY

[0005] The application aims to solve the problems in the prior art and provides a bridge prestressed steel beam intelligent optimization design method based on multi-factor coupling.

[0006] To achieve the above object, the application adopts the following technical scheme: a bridge prestressed steel beam intelligent optimization design method based on multi-factor coupling, comprising the following steps: S1: Obtain the bridge span length, support position, constant load value and live load value through the sensor, calculate the section bending moment value and shear value, superimpose and compare, judge the concrete strength grade and steel beam tensile strength grade threshold, and generate the stress and material constraint set; S2: Based on the stress and material constraint set, encode the steel beam quantity arrangement coordinate tension force integer sequence, and under the overall stiffness value, steel beam total length and peak tension force limit value, combine the simulated annealing algorithm to carry out candidate combination crossover and mutation and screening, obtain a multi-objective candidate arrangement solution set; S3: Based on the multi-objective candidate arrangement solution set, calculate the mid-span displacement value, key section stress ratio, steel beam total length energy and generate neighborhood disturbance solution, compare the energy difference value with the threshold value, select and reserve part of the poor solution according to the set probability, form a new solution sequence, and obtain the energy screening combination; S4: Based on the energy screening combination, construct the stress distribution sequence, tension order sequence difference matrix and compare with the threshold value, correct the gradient projection of the over-limit difference value, generate the modified combination and traverse the neighborhood, compare the residual error network with the energy function residual function, select the combination, and establish the residual error correction solution set; S5: Based on the residual error correction solution set, determine the mid-span displacement value, support reaction force tension order sequence and logical check item by item, determine the combination convergence induction and eliminate unqualified items, and construct a feasible solution set to generate a structure consistency arrangement scheme.

[0007] As a further scheme of the application, the stress and material constraint set includes section bending moment value, section shear value, concrete strength grade threshold and steel beam tensile strength grade threshold, the multi-objective candidate arrangement solution set includes steel beam quantity, arrangement coordinate and tension force integer sequence, the energy screening combination includes mid-span displacement value, key section stress ratio, steel beam total length and energy value, the residual error correction solution set includes stress distribution sequence, tension order sequence, difference matrix and modified combination, and the structure consistency arrangement scheme includes mid-span displacement value, support reaction force, tension order sequence and feasible solution set.

[0008] As a further scheme of the application, the specific steps for generating the stress and material constraint set are: Obtain the bridge span length, support position, constant load value and live load value through the sensor, calculate the bending moment value and shear value of each section point by point, superimpose and compare the bending moment and shear value item by item, and generate the section stress calculation result; Based on the section stress calculation result, judge the concrete strength grade and steel beam tensile strength grade threshold, merge the judgment value and bending moment shear value, convert the merged result into a structure stress and material parameter set, and generate the stress and material constraint set.

[0009] As a further scheme of the present application, the specific step of obtaining the multi-objective candidate arrangement solution set is: Based on the force and material constraint set, the number of steel strands is converted into a numerical sequence and the arrangement coordinates are encoded point by point, while the tensile force numerical section is encoded and combined with the sequence, the encoded items are compared with the constraint threshold value one by one, and the steel strand encoding sequence set is generated; Based on the steel strand encoding sequence set, the bending moment and shear force values of each cross section are calculated, the bending moment and shear force balance error is checked, the matching error result is merged with the steel strand encoding value, and the cross section balance calculation result is generated; Based on the cross section balance calculation result, the candidate combination is divided into several groups and cross-processed within the group, the tensile force and arrangement coordinates are numerically disturbed in the cross-processed sequence, the probability acceptance mechanism of the simulated annealing algorithm is combined to determine the energy difference value and the threshold value, the screened combinations are merged one by one, and the multi-objective candidate arrangement solution set is obtained.

[0010] As a further scheme of the present application, the specific execution process of the simulated annealing algorithm is that, based on the cross section balance calculation result, the candidate combination is divided into several groups and cross-processed within the group, the tensile force and arrangement coordinates are numerically disturbed in the cross-processed sequence, the execution flow is combined, the initial temperature parameter is set and the disturbed solution is generated, the energy difference value between the disturbed solution and the original solution is calculated, the probability function is compared with the threshold value, and part of the poor solutions are accepted under the probability function and reserved in the solution set, the temperature factor is gradually reduced, and the energy difference value is calculated again and the probability is judged, and the multi-objective candidate arrangement solution set is obtained after screening and merging.

[0011] As a further scheme of the present application, the specific checking of the bending moment and shear force balance error is that, based on the steel strand encoding sequence set, the steel strand position and tensile force value are extracted at each key cross section, the internal force under dead load and live load is calculated as bending moment and shear force value, the balance bending moment and shear force required by the cross section are calculated one by one, the bending moment error value and shear force error value are obtained by numerical subtraction, and finally the error value is compared with the set threshold value and the checking result is recorded.

[0012] As a further scheme of the present application, the specific step of obtaining the energy screened combination is: Based on the multi-objective candidate arrangement solution set, the cross mid-span displacement value and the key cross section force ratio are calculated in sections, the total length of the steel strand is combined with the force value to generate an energy value sequence, and the energy value sequence is disturbed point by point to generate an energy disturbed solution set; Based on the energy disturbed solution set, the energy difference value is compared with the threshold value and the comparison result is recorded, the solutions below the threshold value are reserved, the solutions above the threshold value are screened according to the set probability and then combined, and the energy screened combination is obtained.

[0013] As a further scheme of the present application, the specific steps for establishing the residual correction solution set are: Based on the energy screening combination, the stress distribution numerical values are calculated point by point according to the cross-section position, and the parameters are recorded in time sequence to combine the tensioning order, the difference values are obtained by subtracting the two types of sequences item by item, and the difference matrix is generated, and the difference matrix result is constructed; Based on the difference matrix result, the difference values are compared with the threshold value item by item, and the over-limit part is extracted, the over-limit difference values are projected column by column in the gradient direction and corrected, the corrected numerical values are re-merged to form a combined sequence, and the corrected combined sequence is obtained. Based on the corrected combined sequence, the neighborhood combination is traversed point by point and the energy value is calculated, the residual value is superimposed by combining the residual network, and compared with the energy value, the combined sequence is screened according to the conditions, and the sequence is merged to establish the residual correction solution set.

[0014] As a further scheme of the present application, the specific execution process of the residual network is that, based on the corrected combined sequence, the neighborhood combination is traversed point by point and the energy value is calculated, the corrected combined sequence is input as the input layer of the residual network and is transmitted layer by layer, the residual value of each layer output is extracted and is superimposed on the corresponding energy value, a data set containing the comparison of energy and residual is generated, the difference of the superimposed values is screened to select the combined sequence that meets the conditions, and the selected combined sequence is merged to establish the residual correction solution set.

[0015] As a further scheme of the present application, the specific steps for generating the structure consistency arrangement scheme are: Based on the residual correction solution set, the cross-section displacement value is compared with the limit value and the difference value is recorded, then the support reaction force is compared with the bearing threshold value and the over-limit is marked, the tensioning order sequence is checked, the logical judgment is carried out in turn and the qualified and unqualified marks are output, and the checking judgment sequence is generated; Based on the checking judgment sequence, the convergent combination is induced and the unqualified items are removed, the qualified combinations are arranged in turn and then summarized into a set, the arrangement relationship in the set is searched and the structure boundary is fixed, and the structure consistency arrangement scheme is generated.

[0016] Compared with the prior art, the present application has the following advantages and positive effects: 1. In the present application, by quantitatively modeling multiple parameters in the process of arranging the bridge prestressed steel beam, the accuracy of the controllable variables is realized, the global search ability of the arrangement scheme under the multi-objective trade-off is enhanced, the rationality and stability of the steel beam distribution are improved as a whole, and the risk of local overload of the structure is reduced. 2. In the present application, the energy function sequence is constructed by the cross-section displacement value, the cross-section stress ratio and the total length, the energy difference value is used to control the solution set quality, the residual value is introduced for intervention and screening, the robustness of the design under the multi-variable condition is enhanced, and the global adaptability of the local is improved. 3. In the present application, by logical judgment of displacement and counterforce, unqualified arrangements are eliminated, feasible arrangement output under structural consistency is realized, structural efficiency and material use economy are enhanced, and design reliability, optimization solving efficiency and structural response control accuracy are significantly improved. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The present application is a main step schematic diagram. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical scheme and advantages of the present application more clear and explicit, the present application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0019] Example 1 Please refer to Figure 1 The present application provides a technical scheme: a bridge prestressed steel beam intelligent optimization design method based on multi-factor coupling, comprising the following steps: S1: Through the sensor, the bridge span length, support position, constant load value, live load value are obtained, the section bending moment value and shear value are calculated and superimposed and compared, the concrete strength grade and steel beam tensile strength grade threshold are judged, and the stress and material constraint set is generated; S2: Based on the stress and material constraint set, the steel beam number arrangement coordinate tensioning force integer sequence is encoded, the whole stiffness value, steel beam total length and peak tensioning force limit value are combined with the simulated annealing algorithm to carry out candidate combination crossover mutation and screening, and the multi-objective candidate arrangement solution set is obtained; S3: Based on the multi-objective candidate arrangement solution set, the cross displacement value, key section stress ratio, steel beam total length energy are calculated and the neighborhood disturbance solution is generated, the energy difference value and threshold value are compared, and part of the poor solution is selected and reserved according to the set probability, a new solution sequence is formed, and the energy screening combination is obtained; S4: Based on the energy screening combination, the stress distribution sequence, tensioning order sequence difference matrix are constructed and compared with the threshold value, the gradient projection of the over-limit difference value is corrected, the modified combination is generated and traversed in the neighborhood, the residual error network is combined to compare the energy function residual error function, and the combination is selected, the residual error correction solution set is established; S5: Based on the residual error correction solution set, the cross displacement value, support counterforce tensioning order sequence are determined item by item and logically checked, the combination is determined to converge and inducted, and the unqualified items are eliminated to construct the feasible solution set, and the structural consistency arrangement scheme is generated.

[0020] The stress and material constraint set includes cross-section bending moment values, cross-section shear force values, concrete strength grade threshold values, and steel tendon tensile strength grade threshold values, the multi-objective candidate arrangement solution set includes steel tendon quantity, arrangement coordinates, and tensile force integer sequence, the energy screening combination includes mid-span displacement values, key cross-section stress ratios, steel tendon total length, and energy values, the residual correction solution set includes stress distribution sequence, tensile sequence, difference matrix, and correction combination, and the structural consistency arrangement scheme includes mid-span displacement values, support reaction forces, tensile sequence, and feasible solution set.

[0021] The specific steps for generating the stress and material constraint set are as follows: The bridge span length, support position, dead load value, and live load value are obtained through the sensor, the bending moment and shear force values of each cross-section are calculated point by point, the bending moment and shear force values are superimposed and compared item by item, and the cross-section stress calculation result is generated; Based on the cross-section stress calculation result, the concrete strength grade and steel tendon tensile strength grade threshold values are determined, the determined values are matched with the bending moment and shear force values, the matched results are converted into structural stress and material parameter set, and the stress and material constraint set is generated; Based on the bridge span length, support position, dead load value, and live load value collected by the sensor, the bridge structure is discretized into equal-length units, the length of each unit is set to 0.5 meters, the input material elastic modulus parameter is 3.45 times 10 to the power of 10 pascal, the input cross-section moment of inertia parameter is 0.85 cubic meters, the load input includes uniform load value 50 kilonewtons per meter and concentrated load value 35 kilonewtons per meter, the bending moment and shear force values of each unit at five integral points are calculated, the values are superimposed and compared, the bending moment matrix and shear force matrix of each cross-section are recorded point by point, and the cross-section stress calculation result is generated; Based on the cross-section stress calculation result, the concrete strength grade parameter is input as C50 standard corresponding cubic compressive strength 50 megapascal, the steel tendon tensile strength grade threshold value is input as 1860 megapascal, the bending moment and shear force values are compared with the set threshold values point by point, the negative difference values are removed and unified to zero to form a value matrix, the matrix is compared item by item and the matching results are recorded, the matching matrix is combined with the bending moment and shear force values, unified and merged into a parameter set, the matrix is adjusted to a fixed dimension arrangement, and the stress and material constraint set is generated.

[0022] The specific steps for obtaining the multi-objective candidate arrangement solution set are as follows: Based on the stress and material constraint set, the steel tendon quantity is converted into a value sequence and the arrangement coordinates are encoded point by point, the tensile force value is segmented and combined with the sequence, the encoded values are compared with the constraint threshold values item by item, and the steel tendon encoding sequence set is generated; Based on the steel tendon encoding sequence set, the cross-section bending moment and shear force values are calculated, the bending moment and shear force balance error is checked, the error results are merged and matched with the steel tendon encoding values, and the cross-section balance calculation result is generated; Based on the cross-section balance calculation results, the candidate combinations are divided into several groups and cross within the group. In the sequence after the cross, the numerical perturbation of the tension force and the arrangement coordinates is carried out. The probability acceptance mechanism of the simulated annealing algorithm is combined to judge the energy difference value and the threshold value. The screened combinations are merged item by item to obtain the multi-objective candidate arrangement solution set. Based on the stress and material constraint set, the number of steel strands is converted into a numerical sequence. The specific operation is to divide the number of steel strands into several intervals according to the span direction, and the length of each interval is set to 0.5 meters. When encoding, the starting number is specified as 1, and the maximum number does not exceed 100. When encoding the arrangement coordinates, the center line of the support is taken as the zero point, the left coordinate is taken as negative, the right coordinate is taken as positive, and the step size is 0.1 meters. The tension force range is set to 100 to 500 kilonewtons, with a segmented interval of 50 kilonewtons. Each segment of the tension force value is combined with the steel strand sequence to generate a combination pair. When checking the code, the constraint threshold is specified. The bending moment threshold is set to 120 kilonewton-meters, the shear threshold is set to 80 kilonewtons, and the tension force peak threshold is set to 500 kilonewtons. The threshold values of the combinations are compared item by item, and the combinations that meet the requirements are retained to generate the steel strand encoding sequence set. Based on the steel strand encoding sequence set, the bending moment values and shear force values of each section of each steel strand encoding are calculated. During the calculation process, the constant load input value of 50 kilonewtons per meter and the live load input value of 35 kilonewtons per meter are applied to the corresponding sections item by item. The bending moment calculation takes the section span of 1 meter as the interval, and the shear force calculation takes the unit length of 0.5 meters as the interval. The bending moment and shear force balance error is checked. The difference between the required bending moment and the calculated bending moment of each section and the difference between the required shear force and the calculated shear force of each section are calculated. The error value is kept to two decimal places. The difference value exceeding 5 kilonewton-meters is determined as an out-of-limit value. When merging the error results, the bending moment error value and the shear force error value are merged point by point to form a two-dimensional matrix. The error matrix is arranged corresponding to the steel strand encoding sequence to generate the cross-section balance calculation results. Based on the cross-section balance calculation results, the simulated annealing algorithm is used to divide the candidate combinations into several groups, with each group size set to 20 combinations. In the in-group cross operation, the tension force values and arrangement coordinates of each group of combinations are swapped to generate a new combination sequence. Numerical perturbation is carried out in the sequence after the cross. The perturbation amplitude is the coordinate value ±0.05 meters and the tension force value ±10 kilonewtons. The initial temperature parameter is set to 1000, the temperature decay factor is 0.95, and the maximum number of iterations is 500. In each iteration, the energy difference value between the perturbed solution and the original solution is calculated. If the energy difference value is less than or equal to 10, the solution is accepted. If the energy difference value is greater than 10, part of the solution is accepted according to the set probability function. The probability function adopts an exponential distribution form. The input energy difference value and temperature parameter are used to calculate the probability value step by step. The screened combinations are merged item by item into a new combination set to obtain the multi-objective candidate arrangement solution set.

[0023] The specific execution process of the simulated annealing algorithm is as follows: Based on the cross-sectional equilibrium calculation results, the candidate combinations are divided into several groups and cross-groups are performed. In the cross-group sequence, the tension force and arrangement coordinates are numerically perturbed. Combined with the execution process, the initial temperature parameters are set and the perturbed solution is generated. The energy difference between the perturbed solution and the original solution is calculated. The energy difference is compared with the threshold to set a probability function. Under the judgment of the probability function, some poor solutions are accepted and retained in the solution set. The temperature factor is gradually reduced and the energy difference is repeatedly calculated and the probability judgment is performed again. After merging and screening, a multi-objective candidate arrangement solution set is obtained. Simulated annealing algorithm, according to the formula:

[0024] in: This represents the overall probability value of the perturbation solution being accepted. Indicates the corresponding energy difference The weighting coefficients, This represents the change in mid-span displacement. The weighting coefficients, This represents the change in the corresponding support reaction force. The weighting coefficients, This indicates the difference in the corresponding structural consistency score. The weighting coefficients, This represents the difference in unit structural energy between the perturbed solution and the original solution. This represents the change in the mid-span displacement value before and after the disturbance. This represents the change in the value of the support reaction force before and after the disturbance. This represents the difference in structural layout consistency score before and after the disturbance. Indicates the first The weighting coefficients corresponding to the item differences, The weights satisfy , This represents the current temperature parameter, controlling the rate at which the acceptance probability decreases during iteration. The initial value is 1000, and the decay factor is 0.95. The exponential function representing the natural logarithm; Execution process: First, determine the difference in unit structural energy before and after the disturbance. Then, calculate the result by weighted combination of mid-span displacement, total length of steel strands, and force ratio under constraint sets. Points were taken at 0.5-meter intervals along the bridge span, and the difference between the mean mid-span displacements before and after the disturbance was used as the basis for the determination. Change in support reaction force The structural consistency score difference is derived from the difference in the mean reaction forces at all support nodes before and after the disturbance. The score is based on two logical components: layout coordination and the monotonicity of the tensioning sequence. The maximum score is 100 points, and the difference before and after the disturbance is the result. The four difference terms are respectively assigned a weight coefficient wherein , , , The weights are determined by the mean square deviation response of each term in the disturbance sensitivity test, the sum of the weights is normalized, each difference value is multiplied by the respective weight and then linearly superimposed, and the current temperature parameter forms a ratio, and after being substituted into an exponential function, the comprehensive acceptance probability of the disturbance solution is obtained .

[0025] The specific error of the checking bending moment and shear force balance is that, based on the steel tendon coding sequence set, the steel tendon position and tension force value is extracted at each key section, the internal force under the action of dead load and live load is calculated as the bending moment and shear force value, the difference value of the balanced bending moment and shear force required by the section is calculated item by item, the bending moment error value and the shear force error value are obtained by numerical subtraction, and finally the error value is compared with the set threshold value and the checking result is recorded.

[0026] The specific steps of obtaining the energy screening combination are as follows: Based on the multi-objective candidate arrangement solution set, the mid-span displacement value and the stress ratio of the key section are calculated in sections, the total length of the steel tendon is combined with the stress value to generate an energy value sequence, and a disturbance is applied to each point in the energy value sequence to generate an energy disturbance solution set; Based on the energy disturbance solution set, the energy difference value is compared with the threshold value and the comparison result is recorded, the solutions below the threshold value are retained, the solutions above the threshold value are screened according to the set probability and then merged, and the energy screening combination is obtained; Based on the multi-objective candidate arrangement solution set, the mid-span displacement is calculated in sections, the sectional interval is set to 0.5 meters, the mid-span displacement value is recorded point by point, the key section position is selected as the quarter point of the span, the midpoint and the support position, the stress ratio value is calculated, the total length of the steel tendon is combined with the stress value of each section, and an energy value sequence is generated. The energy value is recorded in the form of superposition of bending moment multiplied by displacement and steel tendon length multiplied by stress, all values are kept to three decimal places, a disturbance is applied to each point in the energy value sequence, the disturbance method is to increase the mid-span displacement value by 0.02 millimeters, increase the total length of the steel tendon by 0.1 meters, and increase the stress ratio by 0.01. The disturbed values are recombined to form multiple arrangements to generate an energy disturbance solution set. Based on the energy disturbance solution set, the energy difference value is compared with the threshold value, the threshold value is set to 50 kilo Newton meters, if the energy difference value is less than or equal to the threshold value, it is directly reserved, and whether to reserve is determined by the energy difference value greater than the threshold value, the probability screening adopts the exponential distribution form, the input energy difference value and the set temperature parameter are input, the initial temperature is set to 100, the temperature attenuation factor is set to 0.9 in the iteration, the iteration number is set to 200 times, the probability value is recalculated in each iteration, the probability value is taken to four decimal places and compared with the random number between 0 and 1 point by point, if the random number is less than the probability value, it is reserved, if the random number is greater than the probability value, it is rejected, the reserved solutions are merged and combined, and the energy screening combination is obtained.

[0027] The specific steps for establishing the residual error correction solution set are: Based on the energy screening combination, the stress distribution value is calculated point by point and arranged according to the cross-section position, the parameters are recorded in time sequence and combined in turn, the difference value is obtained by subtracting the two types of sequences in turn and generating the difference matrix result; Based on the difference matrix result, the difference value is compared with the threshold value item by item and the over-limit part is extracted, the over-limit difference value is projected column by column in the gradient direction and corrected, the corrected value is re-merged to form a combination sequence, and the corrected combination sequence is obtained; Based on the corrected combination sequence, the neighborhood combination is traversed point by point and the energy value is calculated, the residual error value is superimposed combined with the residual error network and compared with the energy value, the combination screening is carried out on the items meeting the conditions and the sequence is merged, and the residual error correction solution set is established; Based on the energy screening combination, the cross-section position is equally divided to take points, the span is numbered 1 to 9 according to the equally divided nine positions, and the fixed point coordinate interval ratio is 0.125 span, the stress distribution value is calculated and recorded to three decimal places in units of mega Pascal, the tensioning sequence is recorded in time sequence according to the construction time and numbered from the first hour to the 20th hour with an hour as a time step, a tensioning sequence sequence with a length of 20 is established, and the tensioning section number range 1 to 5 is represented by an integer, the two types of sequences are arranged correspondingly in the time step and cross-section point double index mode, the stress distribution reference difference and the tensioning sequence reference difference are obtained by subtraction, the two types of difference values are respectively stored in the form of nine rows and twenty columns, the row index corresponds to the cross-section position number, and the column index corresponds to the time step, the difference value range corresponds to the stress difference in the interval of negative 50 to positive 50 mega Pascal and the tensioning sequence difference in the interval of negative 4 to positive 4, after two-dimensional splicing, a nine-row and twenty-column difference arrangement is formed, and a difference matrix result is generated; Based on the difference matrix result, the difference values of each column are compared with the threshold value one by one, and the exceeding part is extracted, the stress difference threshold value is set to 1.5 MPa, the tension order difference threshold value is set to 1 step, the exceeding mark is represented by 1, the non-exceeding mark is represented by 0, and the gradient direction projection is performed according to the column, the direction rule is that the positive exceeding is corrected along the negative direction and the negative exceeding is corrected along the positive direction, the step length coefficient is set to 0.2, and the correction increment is given by multiplying the step length coefficient by the exceeding amplitude, the maximum correction amplitude of a single point is limited to that the stress difference does not exceed 5 MPa and the tension order difference does not exceed 2 steps, the upper limit of the iteration round is 10 rounds, and the difference from the threshold value is compared again after each iteration, the subsequent correction is stopped when the amplitude limit is reached, and after completing a single column, the corrected difference value and the original encoding index are re-merged according to the column index order, arranged in time sequence as twenty time steps and corresponding to nine cross-section positions, forming a tension order and stress distribution integrated sequence, and obtaining a modified combined sequence; Based on the modified combined sequence, a residual network method is used, and a combination is generated according to the coordinate and tension force double-dimensional neighborhood of each point, and the energy value is calculated, the coordinate perturbation radius is set to positive and negative 0.05 meters with a step length of 0.01 meters, the tension force perturbation radius is set to positive and negative 10 kilonewtons with a step length of 2 kilonewtons, no more than 50 neighborhood combinations are generated for the reference point, the energy value is recorded according to the cross-section displacement and stress ratio and the steel beam length and stress superposition method, and is unified to the interval negative one thousand to positive one thousand, the residual network input vector length is set to 16, including nine cross-section stresses, cross-section displacements, steel beam total lengths, tension forces, integer encodings and three constraint indicators, the network structure includes three groups of residual units, each group contains two fully connected layers, the number of nodes in each layer is 64, the activation function uses a rectified linear unit, and the residual connection uses a constant mapping, in the training hyperparameters, the learning rate is set to 0.001, the batch size is 32, the training rounds are 100, the weight initialization uses a uniform distribution range of negative 0.05 to positive 0.05, the dropout rate is 0.1, the output residual value is limited to negative 100 to positive 100 and rounded to one decimal place according to the absolute value, the residual value and the corresponding energy value are compared item by item according to the neighborhood points, and are screened according to the pre-set condition threshold value 10, the items that meet the conditions are added to the sequence and are merged according to the time sequence and the cross-section number, and the residual correction solution set is established.

[0028] The specific execution process of the residual network is that, based on the modified combined sequence, the neighborhood combination is traversed point by point and the energy value is calculated, the modified combined sequence is input as the input layer of the residual network and is transmitted layer by layer, the residual values of each layer output are extracted and are superimposed on the corresponding energy values, a data set containing energy and residual comparison is generated, the difference between the superimposed values is compared, the combination sequence that meets the condition is screened out, the screened combination sequence is merged, and the residual correction solution set is established; The residual network is according to the formula:

[0029] Among them: represents the final fused residual score value, represents the input vector sequence, length 16, containing 9 cross-section stress values, 1 mid-span displacement value, 1 total length of steel beam value, 1 integer number of tension force value, 3 constraint indicators and 1 layout structure feature score, represents the first branch residual transformation function, represents the second branch regularization residual function, represents the structure compatibility scoring item, represents the residual error density item, represents the fusion coefficient of the first residual branch output, represents the fusion coefficient of the second residual branch output, represents the fusion coefficient of the structure compatibility scoring item, represents the fusion coefficient of the error density item; Execution process: first construct the input vector sequence , containing nine cross-section stress values, one mid-span displacement value, one total length of steel beam value, one integer number of tension force value, three structure constraint indicators and one layout structure score value, a total of 16 parameters, the input vector is input into the first residual branch perform double-layer fully connected mapping and output residual values, then input into the second branch perform normalization and Dropout regularization processing to form a residual response that suppresses overfitting, while externally connecting the structure compatibility scoring item and the error density item , all output items are multiplied by weight coefficients weighted sum and add back the original input get the fusion score .

[0030] The specific steps of generating a structure consistent layout scheme are as follows: Based on the residual correction solution set, compare the mid-span displacement value with the limit value and record the difference, then compare the support reaction with the bearing threshold and mark the over-limit, check the tension sequence, perform logical judgment in turn and output qualified and unqualified marks, generate a check judgment sequence; Based on the check judgment sequence, induce convergent combinations and eliminate unqualified items, sort the qualified combinations in turn and then aggregate into a set, search for the layout relationship in the set and fix the structure boundary, generate a structure consistent layout scheme; Based on the residual correction solution, the cross mid-span displacement value is compared with the limit value point by point, the cross mid-span displacement limit value is preset to 30mm, the displacement value of each section is accurate to 0.01mm and is subtracted from the limit value one by one, the difference value is generated and recorded as the displacement difference sequence, the support reaction and the bearing threshold are compared, the bearing threshold is preset to 3000kN, the reaction value is compared item by item and the over-limit result is marked, the over-limit mark uses 1, the non-over-limit mark uses 0, the tension sequence is checked, the sequence range is 1 to 10 and is arranged step by step according to the time step, the checking logic is the monotone increasing rule, if there is reverse order, it is judged as unqualified and recorded as 0, the qualified record is 1, the judgment output is arranged in order to form a Boolean matrix, the matrix dimension is the number of sections multiplied by the number of checking items, and the checking judgment sequence is generated; Based on the checking judgment sequence, the combinations marked as qualified in the sequence are summarized, the clustering parameters are preset to cluster radius 0.2 and cluster capacity upper limit 50, the unqualified items are removed in the convergence operation and only the combinations with mark value 1 are reserved, the qualified combinations are arranged in order, the cross span position and the tension sequence are re-arranged according to the double index, a two-dimensional array is formed, the array row index corresponds to the section number, and the column index corresponds to the time sequence, the qualified combinations are summarized in order according to the column priority, a set matrix is obtained, then the arrangement relationship in the set is searched, the search method is to compare the arrangement coordinates and the tension force value point by point, if the coordinate difference between two points is less than 0.05m and the tension force difference is less than 5kN, they are classified as the same arrangement unit, the boundary in the set is fixed, the boundary value includes the cross span starting point 0m and the terminal full span length and the support position coordinates, after the arrangement boundary is fixed, the structure consistency arrangement scheme is generated.

[0031] The above is only the preferred embodiment of the present application, not other forms of the present application, any skilled person in the art can use the above disclosed technical content to change or modify the equivalent embodiment applied to other fields, but any simple modification, equivalent change and modification of the above embodiment without departing from the technical solution content of the present application, according to the technical essence of the present application, still belongs to the protection scope of the technical solution of the present application.

Claims

1. A bridge prestressed steel cable intelligent optimization design method based on multi-factor coupling, characterized in that, The method comprises the following steps: S1: obtaining the bridge span length, support position, dead load value and live load value through a sensor, calculating the section bending moment value and shear value, superimposing and comparing, judging the concrete strength grade and steel beam tensile strength grade threshold, and generating the stress and material constraint set; S2: based on the stress and material constraint set, encoding the steel beam quantity arrangement coordinate tension force integer sequence, combining the simulated annealing algorithm under the overall stiffness value, steel beam total length and peak tension force limit value to carry out candidate combination crossover and mutation and screening, and obtaining a multi-objective candidate arrangement solution set; S3: based on the multi-objective candidate arrangement solution set, calculating the mid-span displacement value, key section stress ratio, steel beam total length energy and generating the neighborhood disturbance solution, comparing the energy difference value with the threshold value, selecting and retaining part of the poor solutions according to the set probability, forming a new solution sequence, and obtaining the energy screened combination; S4: based on the energy screened combination, constructing the stress distribution sequence, tension sequence difference matrix and threshold value comparison, correcting the gradient projection of the over-limit difference value, simultaneously generating the modified combination and traversing the neighborhood, selecting the combination after comparing the residual error network with the energy function residual function, and establishing the residual error correction solution set; S5: based on the residual error correction solution set, determining the mid-span displacement value, support reaction force tension sequence and logical check item by item, judging the combination convergence induction and eliminating unqualified items to construct the feasible solution set, and generating the structure consistency arrangement scheme.

2. The method for intelligent optimization design of bridge prestressed steel tendon based on multi-factor coupling according to claim 1, characterized in that, The stress and material constraint set comprises the section bending moment value, section shear value, concrete strength grade threshold and steel beam tensile strength grade threshold, the multi-objective candidate arrangement solution set comprises the steel beam quantity, arrangement coordinate and tension force integer sequence, the energy screened combination comprises the mid-span displacement value, key section stress ratio, steel beam total length and energy value, the residual error correction solution set comprises the stress distribution sequence, tension sequence, difference matrix and modified combination, and the structure consistency arrangement scheme comprises the mid-span displacement value, support reaction force, tension sequence and feasible solution set.

3. The method for intelligent optimization design of bridge prestressed steel tendon based on multi-factor coupling according to claim 1, characterized in that, The specific steps for generating the stress and material constraint set are: obtaining the bridge span length, support position, dead load value and live load value through a sensor, calculating the bending moment value and shear value of each section point by point, superimposing and comparing the bending moment and shear value item by item, and generating the section stress calculation result; based on the section stress calculation result, judging the concrete strength grade and steel beam tensile strength grade threshold, merging and matching the judgment value and bending moment shear value, converting the merged result into the structure stress and material parameter set, and generating the stress and material constraint set.

4. The method for intelligent optimization design of bridge prestressed steel tendon based on multi-factor coupling according to claim 1, characterized in that, The specific steps for obtaining the multi-objective candidate arrangement solution set are: based on the stress and material constraint set, converting the steel beam quantity into a numerical sequence and encoding the arrangement coordinate point by point, and encoding the tension force value segment and combining with the sequence, checking the constraint threshold after encoding, and generating the steel beam encoding sequence set; based on the steel beam encoding sequence set, calculating the bending moment value and shear value of each section, checking the bending moment and shear balance error, merging and matching the error result and steel beam encoding value, and generating the section balance calculation result; Based on the cross-section balance calculation result, the candidate combination is divided into several groups and cross-grouped, numerical perturbation is performed on the tension force and arrangement coordinates in the crossed sequence, the energy difference value is determined in combination with the probability acceptance mechanism of the simulated annealing algorithm, and the threshold value is screened, and the screened combination is merged item by item to obtain a multi-objective candidate arrangement solution set.

5. The method for intelligent optimization design of bridge prestressed steel tendon based on multi-factor coupling according to claim 1, characterized in that, The specific execution process of the simulated annealing algorithm is that, based on the cross-section balance calculation result, the candidate combination is divided into several groups and cross-grouped, numerical perturbation is performed on the tension force and arrangement coordinates in the crossed sequence, in combination with the execution process, the initial temperature parameter is set and the perturbed solution is generated, the energy difference value between the perturbed solution and the original solution is calculated, the probability function is set by comparing the energy difference value with the threshold value, and part of the poor solutions are accepted under the probability function and reserved in the solution set, the temperature factor is gradually reduced, and the energy difference value is calculated again and the probability is determined, and the multi-objective candidate arrangement solution set is obtained by merging and screening.

6. The method for intelligent optimization design of bridge prestressed steel tendon based on multi-factor coupling according to claim 1, characterized in that, The specific execution process of the simulated annealing algorithm is that, based on the cross-section balance calculation result, the candidate combination is divided into several groups and cross-grouped, numerical perturbation is performed on the tension force and arrangement coordinates in the crossed sequence, in combination with the execution process, the initial temperature parameter is set and the perturbed solution is generated, the energy difference value between the perturbed solution and the original solution is calculated, the probability function is set by comparing the energy difference value with the threshold value, and part of the poor solutions are accepted under the probability function and reserved in the solution set, the temperature factor is gradually reduced, and the energy difference value is calculated again and the probability is determined, and the multi-objective candidate arrangement solution set is obtained by merging and screening.

7. The method for intelligent optimization design of bridge prestressed steel tendon based on multi-factor coupling according to claim 1, characterized in that, The specific steps for obtaining the energy screened combination are: Based on the multi-objective candidate arrangement solution set, the cross-section displacement value and the stress ratio of the key section are calculated, the total length of the steel tendon is combined with the stress value to generate an energy value sequence, and perturbation is applied to each point in the energy value sequence to generate an energy perturbed solution set; Based on the energy perturbed solution set, the energy difference value is compared with the threshold value and the comparison result is recorded, the solutions below the threshold value are reserved, the solutions above the threshold value are screened according to the set probability, and then the combination is merged to obtain the energy screened combination.

8. The method for intelligent optimization design of bridge prestressed steel tendon based on multi-factor coupling according to claim 1, characterized in that, The specific steps for establishing the residual correction solution set are: Based on the energy screened combination, the stress distribution value is calculated point by point and arranged according to the cross-section position, the tension sequence is recorded in time sequence, and the difference value is obtained by subtracting the two types of sequences and generating a difference matrix, and the difference matrix result is constructed; Based on the difference matrix result, the difference value is compared with the threshold value item by item and the out-of-limit part is extracted, the out-of-limit difference value is projected column by column in the gradient direction and corrected, the corrected value is re-merged to form a combination sequence, and the corrected combination sequence is obtained; Based on the corrected combination sequence, the neighborhood combination is traversed point by point and the energy value is calculated, the residual value is superimposed on the energy value by combining the residual network, and the energy value is compared, the combination is screened according to the conditions and the sequence is merged, and the residual correction solution set is established.

9. The method for intelligent optimization design of bridge prestressed steel tendon based on multi-factor coupling according to claim 1, characterized in that, The specific execution process of the residual network is that, based on the corrected combination sequence, the neighborhood combination is traversed point by point and the energy value is calculated, the corrected combination sequence is input as the input layer of the residual network and is transmitted layer by layer, the residual value of each layer output is extracted and is superimposed on the corresponding energy value, a data set containing the energy and residual comparison is generated, the difference between the superimposed values is compared and screened for the combination sequence that meets the conditions, the screened combination sequence is merged, and the residual correction solution set is established.

10. The method for intelligent optimization design of bridge prestressed steel tendon based on multi-factor coupling according to claim 1, characterized in that, The specific steps for generating the structural consistency arrangement scheme are as follows: Based on the residual correction solution set, the cross median displacement value is compared with the limit value and the difference is recorded, then the support reaction is compared with the bearing threshold value and the overrun is marked, the tension sequence is checked, the logical judgment is carried out in turn and the qualified and unqualified marks are output, and the checking judgment sequence is generated; Based on the checking judgment sequence, the convergence combination is induced and the unqualified item is removed, the qualified combination is arranged in turn and then is collected into a set, the arrangement relationship in the set is searched and the structural boundary is fixed, and the structural consistency arrangement scheme is generated.

Images