Parameter correction method and device considering multiple response quantities in numerical simulation of water conservancy and hydropower engineering

Abstract

The application provides a parameter correction method and device considering multiple response quantities in a numerical simulation process of a water conservancy and hydropower project, which can practically improve the efficiency and precision of parameter correction of a simulation model of the water conservancy and hydropower project, and makes simulation data closer to the real situation. The method comprises the following steps: step 1, extracting multiple measurement data in actual operation of the engineering structure as real response quantities; step 2, establishing a simulation model and determining model parameters that need to be calibrated; step 3, preliminarily determining an information matrix of a response surface model from the parameters that need to be calibrated and the correction ranges of the parameters; step 4, substituting selected test points into the simulation model, then obtaining calculation values corresponding to actual engineering measurement point positions from the simulation model as simulation response quantities, and then substituting the simulation response quantities into the response surface model considering the interaction of related non-normal parameters; and step 5, constructing a target function considering multiple response quantities, searching for a minimum value, and obtaining corrected parameters.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of numerical simulation of water conservancy and hydropower engineering, and particularly relates to a parameter correction method and device considering multiple response quantities in the numerical simulation process of water conservancy and hydropower engineering. BACKGROUND

[0002] Numerical simulation is also called computer simulation. Relying on an electronic computer, combining the concept of finite elements or finite volumes, through numerical calculation and image display, the purpose of studying engineering problems and physical problems is achieved. Whether the result of numerical simulation is accurate or not depends on the selection of various material constitutive models and the selection of mechanical parameters in the modeling process. In water conservancy and hydropower engineering, a part of mechanical parameters (such as the tangent modulus coefficient of rockfill material and the cohesion of soil) are usually selected by referring to the experience of similar projects without professional instruments, which may cause a large error in the result and produce a wrong judgment of the engineering structure; sometimes even if the mechanical parameters are measured, there are factors such as non-standard operation of instruments and pollution of samples, which make the measured result different from the actual situation. In order to construct a more actual simulation model, it is necessary to correct the material parameters of the model to ensure the accuracy of the result.

[0003] The correction of parameters generally adopts simulation software to model, then sets a series of parameter values, and performs a large number of simulation calculations to construct a mathematical model, converts the parameter correction problem into a minimum value problem of mathematical programming, and obtains the optimal parameter combination. This process often adopts the response surface method to construct a mathematical model, and uses the genetic algorithm to find the minimum value. The current parameter correction method has the following shortcomings:

[0004] First, the response surface function adopted in the current parameter correction is of a specific form, and when the parameters contain correlated non-normal variables, the fitting degree of the response surface model will be limited; however, in water conservancy and hydropower engineering, some parameters such as rock-soil cohesion and internal friction angle, and the percentage of water filling depth and the depth of the tension crack in the rock slope, have obvious statistical negative correlation between the parameters. If the previous specific response surface function is still used, the accuracy of the final response surface model cannot be guaranteed. The response surface method is mostly based on full-factor orthogonal test method, central composite design method, Box-Behnken design method, etc. These experimental design methods have a common problem that when encountering a multi-factor and multi-variable scheme, in order to ensure the accuracy of the response surface model, a large number of cross test schemes need to be arranged, which will consume a lot of manpower and machine time.

[0005] Second, the conventional genetic algorithm uses selection, crossover and mutation operators for searching, and has strong global search ability but weak local search ability, and can only obtain a suboptimal solution of the problem instead of an optimal solution, and the stability of the solution is poor. The stop criterion of the conventional genetic algorithm has two kinds: 1. The evolution of a predetermined number of generations is completed; 2. The difference between the response quantities of the problem before and after is less than a given threshold. The simulation model of a water conservancy and hydropower project is generally complex, there are many parameters to be corrected in a model, the response quantity is usually determined by multiple parameters, the maximum evolution number is difficult to accurately set, and there are often multiple variable combinations that can achieve the same response quantity, which causes the premature phenomenon in the crossover and mutation process and leads to a large error in the result.

[0006] Third, in the field observation process of a water conservancy and hydropower project, there are usually multiple measurement data (such as stress, strain, displacement, etc.) and multiple measuring points. The previous parameter correction is based on a single response quantity, and the correction result often leads to the accuracy of the single response quantity, and the remaining response quantities produce greater errors, so the existing method is not suitable for the parameter correction considering multiple response quantities in the numerical simulation process of a water conservancy and hydropower project. SUMMARY

[0007] The present application is carried out to solve the above problems, and aims to provide a parameter correction method and device considering multiple response quantities in the numerical simulation process of a water conservancy and hydropower project, which can practically improve the efficiency and precision of the simulation model parameter correction of a water conservancy and hydropower project, and make the simulation data closer to the actual situation.

[0008] The present application adopts the following scheme to achieve the above purpose.

[0009] <Method>

[0010] The present application provides a parameter correction method considering multiple response quantities in the numerical simulation process of a water conservancy and hydropower project, characterized in that the method comprises the following steps:

[0011] Step 1, extracting multiple measurement data (such as stress, strain, displacement, etc.) in the actual operation of the engineering structure as real response quantities;

[0012] Step 2, establishing a simulation model according to the actual engineering structure size, and using an optimization design technology to determine the parameters (such as structure parameters and material parameters) to be calibrated;

[0013] Step 3, preliminarily determining the information matrix M of the response surface model from the parameters to be calibrated and the correction range χ a (a = 1, 2, …, n), and n is the number of parameters to be calibrated;

[0014] Supposing that an experiment contains N design points x e, e = 1, 2, …, N, the corresponding information matrix M is:

[0015]

[0016]

[0017] In the formula, x e is a design point in the parameter range χ a , each design point contains a set of numerical groups, and each numerical group contains the parameter values of the first to the nth parameters: x e = [Value1 Value2…Value n ]; f1(x e ), f2(x e ), …, f w (x e ) are continuous functions on the parameter range χ a , and the spatial position of the parameter is represented by a polynomial containing these continuous functions f1(x e ), f2(x e ), …, f w (x e );

[0018] The optimization design technique can be Wynn sequence generation algorithm, Fedorov single-point exchange algorithm, offset adjustment algorithm, etc.; the design point that makes the determinant of the information matrix maximum is found, and then the information matrix M is made smaller through the selection of the test combination, and the selected design point is taken as the test point to be substituted into the simulation model for calculation;

[0019] Step 4, substitute the numerical group of each test point selected in the spatial domain into the simulation model for calculation, then obtain the calculation value of the corresponding measurement point of the actual engineering measurement point position from the simulation model as the simulation response, and then substitute the numerical group of each test point and the obtained simulation response into the response surface model considering the interaction of related non-normal parameters, associate each simulation response with all n parameters, and obtain the function relationship between the simulation response and all n parameters;

[0020] Step 5, construct the objective function considering multiple response quantities, and search for the minimum value of the objective function by using the improved genetic algorithm to obtain the size of the modified parameters of the simulation model;

[0021] The improved genetic algorithm is: after the mutation module of the genetic algorithm, a local optimization module is added to speed up the convergence of the genetic algorithm; a variable range detection module is added to prevent the optimization value from exceeding the parameter range after local optimization; an intelligent stop module is added, and the difference between the optimization values before and after is less than the rated value as the stop condition, which ensures the accuracy of the optimization result while reducing the number of algorithm iterations.

[0022] Preferably, the parameter correction method for multiple response quantities in the numerical simulation process of water conservancy and hydropower engineering provided by the present application can further have the following characteristics: in step 2, the simulation model is established according to the actual engineering structure size, and the parameters to be calibrated are determined; when the number of parameters is small (2-7), the next step can be directly performed; if the number of parameters is large (more than 7), factorial analysis can be performed first to preliminarily screen significant effect factors and reduce the subsequent workload. Then, taking the commonly used parameter size as a benchmark, the parameter correction range can be determined according to engineering experience, or the parameter size can be directly multiplied by 1.5 and 0.5 as the upper and lower limits of the correction range.

[0023] Preferably, the parameter correction method for multiple response quantities in the numerical simulation process of water conservancy and hydropower engineering provided by the present application can further have the following characteristics: in step 3, the principle for selecting the test combination is to minimize the number of spatial design points under the premise of being able to solve the response surface model.

[0024] Preferably, the parameter correction method for multiple response quantities in the numerical simulation process of water conservancy and hydropower engineering provided by the present application can further have the following characteristics: in step 4, the response surface model considering the interaction of each related non-normal parameter is:

[0025]

[0026] In the formula, y l is the lth response quantity, l is less than or equal to the number of response quantities; b is the set highest expansion degree of the polynomial; q l is a constant term (independent random variable) to be solved; β1-β w is the 1st to wth coefficient to be solved; f w (x w ) is the wth parameter term, each parameter term is an expression containing at least one parameter, and f1(x1)-f w (x w ) contains all n parameters.

[0027] Preferably, the parameter correction method for multiple response quantities in the numerical simulation process of water conservancy and hydropower engineering provided by the present application can further have the following characteristics: in step 4, for the response surface model f1(x1)-f w (x w) each item is subjected to F test, and invalid items with poor fitting effect are removed; after removing invalid items each time, new coefficients and constant terms are solved, then the response surface model of the response quantity is subjected to F test again, if the test result still does not meet the fitting threshold requirement, the value of b is increased, and then the test is performed again until the test result of the response surface model of the response quantity meets the fitting threshold requirement, and the optimized response surface model reflecting the functional relationship between the response quantity and the parameters is obtained.

[0028] Taking 3 parameters A, B and C and the highest expansion number 2 times as an example, n=3 and b=2 are brought into There are:

[0029] y l = β1A + β2B + β3C + β4AB + β5AC + β6BC + β7A 2 + β8B 2 + β9C 2 + q l ,

[0030] 9 f(x) and mathematical models need to be subjected to F test, and invalid items with P value greater than 5% in the F test result are removed, so as to reduce the misfit effect of the response surface model considering parameter interaction and improve the model fitting precision. If the misfit effect of the model is still relatively significant after removing invalid items, the value of b needs to be increased until the requirement is met. Then the value of each design point A, B and C and the response quantity obtained by simulation calculation are brought into the response surface model, β is solved, and the mathematical model of each response quantity is obtained.

[0031] Preferably, the parameter correction method considering multiple response quantities in the numerical simulation process of the water conservancy and hydropower engineering provided by the application can also have the following characteristics: the fitting threshold requirement is that the misfit effect F test result P value of the response surface model is greater than 5%.

[0032] Preferably, the parameter correction method considering multiple response quantities in the numerical simulation process of the water conservancy and hydropower engineering provided by the application can also have the following characteristics: in step 5, the objective function considering multiple response quantities is:

[0033]

[0034] Wherein, y l ′ is the real response quantity in the actual operation of the engineering structure.

[0035] Preferably, the parameter correction method considering multiple response quantities in the numerical simulation process of water conservancy and hydropower engineering provided by the application can also have the following characteristics: in step 5, the genetic algorithm local optimization operation mainly embeds the nonlinear programming function in the Matlab optimization toolbox into the mutation operation, takes each generation population individual after mutation as the initial value, substitutes the initial value into the programming function, and takes the local optimal value found as a new individual chromosome to continue evolution; when the optimal value appears at the upper and lower boundaries of the parameter range, the local optimization result can exist outside the parameter range, and the zero index is set in the variable detection operation, and the offspring outside the parameter range is not subjected to genetic operation; after multiple generations of genetic operation, the local optimization value gradually tends to be stable, and when the variation range of the optimization value of the n parameters is less than the set error value, the cycle is ended.

[0036] <device>

[0037] Further, the application also provides a parameter correction device based on the above <method>, characterized in that it comprises:

[0038] an extraction unit for extracting multiple measurement data in the actual operation of the engineering structure as real response quantities;

[0039] a parameter determination unit for establishing a simulation model according to the actual engineering structure size and determining parameters to be calibrated;

[0040] an information matrix determination unit for preliminarily determining an information matrix M of a response surface model from the parameters to be calibrated and the correction ranges χ a (a = 1, 2, …, n) of the parameters, n being the number of parameters to be calibrated; assuming that the test contains N design points x e , e = 1, 2, …, N in the spatial domain, the corresponding information matrix M is:

[0041]

[0042]

[0043] In the formula, x e is a design point in the parameter range χ a , each design point contains a group of numerical values, each group of numerical values contains the parameter values of the first to the n-th parameters, x e = [Value1 Value2…Value n ]; f1(x e ), f2(x e ), …, f w (x e ) are continuous functions on the parameter range χ a ;

[0044] The design point that makes the determinant of the information matrix maximum is searched, and the information matrix M is made smaller through the selection of the test combination, and the selected design point is taken as the test point to be substituted into the simulation model for calculation;

[0045] The function relationship determination unit substitutes the numerical group of each test point selected in the spatial domain into the simulation model for calculation, and then obtains the calculation value of the test point corresponding to the actual engineering measurement point position from the simulation model as the simulation response quantity, and substitutes the numerical group of each test point and the obtained simulation response quantity into the response surface model considering the interaction of the related non-normal parameters, associates each simulation response quantity with all n parameters, and obtains the function relationship between the simulation response quantity and all n parameters;

[0046] The modified parameter obtaining unit constructs a target function considering multiple response quantities, searches for the minimum value of the target function by using the improved genetic algorithm, and obtains the size of the parameter after the simulation model is modified; the improved genetic algorithm is as follows: a local optimization module is added after the mutation module of the genetic algorithm to accelerate the convergence of the genetic algorithm; a variable range detection module is added to prevent the optimized value from exceeding the parameter range after local optimization; an intelligent stop module is added, and the difference between the optimized values before and after the optimization is less than a rated value as a stop condition, so that the accuracy of the optimized result is ensured and the iteration number of the algorithm is reduced; and

[0047] The control unit is in communication connection with the extraction unit, the parameter determination unit, the information matrix determination unit, the function relationship determination unit and the modified parameter obtaining unit, and controls the operation of them.

[0048] Preferably, the parameter modification device for considering multiple response quantities in the numerical simulation process of the water conservancy and hydropower engineering provided by the application can further comprise an input display unit in communication connection with the control unit, so that an operator can input operation instructions, and the corresponding information can be displayed according to the control instructions.

[0049] Preferably, the parameter modification device for considering multiple response quantities in the numerical simulation process of the water conservancy and hydropower engineering provided by the application can further have the following features: the input display unit can display the input, output data and processing process of the extraction unit, the parameter determination unit, the information matrix determination unit, the function relationship determination unit and the modified parameter obtaining unit in the form of a data table or an image in a static or dynamic manner according to the control instructions, and can display the response quantity value obtained based on the function relationship between the response quantity and the parameter on the model response position on the simulation model of the water conservancy and hydropower engineering, and display the structural shape, displacement or stress change corresponding to the response quantity value.

[0050] Effects of the application

[0051] The parameter correction method and device for considering multiple response quantities in the numerical simulation process of water conservancy and hydropower engineering can greatly reduce the number of tests while fully considering the interaction between related non-normal variables, add a local optimization module, a variable range detection module and an intelligent stop module under a genetic algorithm mutation module, and improve the overall stability of the genetic algorithm; a multiple response quantity target function is constructed; the efficiency and precision of the parameter correction of the water conservancy and hydropower engineering simulation model are improved through the foregoing processing, so that the water conservancy and hydropower engineering simulation can fully consider the influence of related factors, obtain simulation data closer to the actual situation, provide a scientific basis for accurate and reliable simulation of complex and changeable water conservancy and hydropower engineering in actual conditions, and provide reliable technical support for safe and stable operation of water conservancy and hydropower engineering. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 The flowchart of the parameter correction method for considering multiple response quantities in the numerical simulation process of water conservancy and hydropower engineering related to the embodiments of the present application is shown.

[0053] Figure 2 The steel pipe model diagram related to the embodiments of the present application is shown, wherein (a) is a field test layout diagram, and (b) is a model section diagram.

[0054] Figure 3 The steel pipe observation point layout schematic diagram related to the embodiments of the present application is shown, wherein 1, 2, 3, etc. pure numbers represent strain measuring points, and B1, B2, B3, etc. "B+ numbers" represent displacement measuring points.

[0055] Figure 4 The multi-linear follow-up reinforcement three-fold line constitutive schematic diagram related to the embodiments of the present application is shown.

[0056] Figure 5 The effect comparison diagram of the improved genetic algorithm and the conventional genetic algorithm in the embodiments of the present application is shown.

[0057] Figure 6 The comparison diagram of the measured values of measuring points and the finite element calculation values before and after parameter correction related to the embodiments of the present application is shown, wherein (a) is a B1 measuring point water pressure-strain diagram, and (b) is a B1 measuring point water pressure-displacement diagram. DETAILED DESCRIPTION

[0058] The parameter correction method and device for considering multiple response quantities in the numerical simulation process of water conservancy and hydropower engineering related to the embodiments of the present application are described in detail in combination with the drawings.

[0059] <EMBODIMENT>

[0060] In this embodiment, the numerical simulation of the hydraulic test of the penstock of a water conservancy and hydropower project is taken as an example to illustrate the parameter correction method considering multiple response quantities in the numerical simulation process of the water conservancy and hydropower project of the present invention.

[0061] like Figure 1 As shown, the parameter correction method considering multiple response quantities in the numerical simulation process of water conservancy and hydropower projects provided in this embodiment includes the following contents:

[0062] Step 1. Penstock hydraulic model structure Figure 2 , a prototype test was conducted in an open air environment, with multiple strain gauges and displacement meters set up along the pipe. Figure 3 Then, the strains ε1, ε3, ε5, ε7 of the 1, 3, 5, and 7 measuring points of the AA section and the displacements d1, d2, and d3 of the B1, B2, and B3 measuring points during the pressurization process are extracted as the true response y l ′.

[0063] Step 2. Create a simulation model in the finite element software Abaqus with the actual engineering structure size. The steel material adopts the multi-linear follow-up strengthening three-fold line constitutive model. Figure 4 Based on the type of response, the Poisson's ratio μ and the three-fold constitutive parameters k1, k2, k3, σ1, and σ2 are temporarily listed as parameters that need to be calibrated. The values ​​of k1, k2, k3, σ1, and σ2 are roughly determined based on the actual stress-strain curve. μ is set to 0.3. Then, the parameter values ​​are directly multiplied by 1.5 and 0.5 as the upper and lower limits of the correction range. k1>k2>k3 and σ1<σ2 are added as parameter constraints.

[0064] Step 3. Determine the information matrix M of the response surface model based on the number of parameters n=6 and the parameter range. Use the offset adjustment method to set the search step to 0.05 and the maximum step to 0.8 to find the design point that maximizes the determinant of the information matrix. Then, through the selection of experimental combinations, it is determined that the number of experiments required is 38. The conventional central composite design response surface rule requires 86 experiments, while the 6-factor 3-level full factor orthogonal analysis that wants to achieve the same effect requires 729 experiments. It can be seen that the introduction of the optimization design concept can greatly reduce the number of experiments.

[0065] Step 4. The 38 groups of μ, k1, k2, k3, σ1, σ2 values contained in the 38 selected design points in the spatial domain are substituted into the finite element simulation model for calculation, and the strain and displacement calculation values corresponding to the actual engineering measurement point positions are extracted as the simulation response quantities. The numerical groups of each design point and the simulation response quantities are substituted into the response surface model considering the interaction between each related non-normal parameters, and the highest expansion order of the model is 2. F test is performed on each parameter term and interaction term, and invalid terms with P value greater than 5% are removed, thereby reducing the misfit effect of the response surface model considering parameter interaction and improving the model fitting accuracy. After removing the invalid terms, the misfit P values of the strain model and the displacement model are 0.1436 and 0.1224 respectively, which are greater than 0.1, indicating that the misfit effect is not significant and the fitting effect is very good. The misfit P value of the model without considering parameter interaction is 0.042, which is less than 0.05, indicating that the misfit effect is very significant. It can be seen that the necessity and importance of considering parameter interaction in the method. Then, the to-be-estimated coefficients of the 37 response quantity mathematical models of ε1, ε3, ε5, ε7 and d1, d2 are determined.

[0066] Step 5. After the genetic algorithm mutation module, add a local optimization module, a variable range detection module and an intelligent stop module. The local optimization operation mainly embeds the nonlinear programming function in the Matlab optimization toolbox into the mutation operation, substitutes each generation of population individuals after mutation into the programming function as initial values, and takes the local optimal value found as a new individual chromosome for evolution. When the optimal value appears at the upper and lower boundaries of the parameter range, the result of local optimization may exceed the parameter range. In the variable detection operation, set the zero index and do not perform genetic operation on the offspring exceeding the parameter range. After multiple generations of genetic operation, the local optimization value gradually tends to be stable, and when the optimization value of the six parameters changes by less than the set error value 0.001, the cycle is ended. See Figure 5 The improved genetic algorithm automatically terminates at the 32nd generation when the stop condition is met, the average value and the best value of the 32nd generation coincide, the minimum function value found is 13.424, while the conventional genetic algorithm converges slowly, the average value and the best value of the 150th generation do not coincide and still have evolution space, and the minimum function value found under the set number of generations is 13.552. It can be seen that the improvement of the genetic algorithm can not only ensure the accuracy of the optimization result, but also reduce the number of iterations of the algorithm.

[0067] Step 6. Construct a target function considering multiple response quantities, substitute the 37 response quantity mathematical models of ε1, e3, ε5, ε7 and d1, d2 and the actual engineering corresponding true response quantity values into the target function formula:

[0068] MinY = |e1-e1'| + |e3-e3'| + |e5-e5'| + |e7-e7'| + |d1-d1'| + |d2-d2'| + |d3-d3'|, wherein, y i ′ is the real response quantity extracted in the actual project, the improved genetic algorithm is used for searching minimum value of the target function, and the size of the parameter after the simulation model is corrected is obtained.

[0069] As shown in Figure 6 FIG. 1, taking strain measuring point 1 and displacement measuring point B1 of A-A section as an example, a comparison chart of the measured value of the pressure pipeline hydrostatic test measuring point and the finite element calculation value before and after the parameter correction is drawn; through comparison, it is illustrated that the calculation result after the parameter correction and the measured value are consistent in the law, and compared with the calculation result before the parameter correction, the calculation result after the parameter correction is more close to the measured value in the value, and the rationality and effectiveness of the parameter correction method are demonstrated.

[0070] Further, the embodiment also provides a device capable of automatically implementing the above method, and the device comprises an extraction part, a parameter determination part, an information matrix determination part, a function relationship determination part, a corrected parameter obtaining part, an input display part and a control part.

[0071] The extraction part performs the content described in step 1, and extracts various measurement data in the actual operation of the engineering structure as the real response quantity.

[0072] The parameter determination part performs the content described in step 2, establishes a simulation model according to the actual engineering structure size, and determines the parameters to be calibrated.

[0073] The information matrix determination part performs the content described in step 3, and preliminarily determines the information matrix M of the response surface model from the parameters to be calibrated and the correction range χ a (a = 1, 2, ···, n) of each parameter; a design point that makes the determinant of the information matrix maximum is found, and then the selection of the test combination makes the information matrix M smaller and smaller, and the selected design point is taken as a test point to be substituted into the simulation model for calculation.

[0074] The function relationship determination part performs the content described in step 4, substitutes the numerical group of each test point selected in the space domain into the simulation model for calculation, then obtains the calculation value of the measuring point corresponding to the actual engineering measuring point position from the simulation model as the simulation response quantity, and then substitutes the numerical group of each test point and the obtained simulation response quantity into the response surface model considering the interaction of the related non-normal parameters, associates each simulation response quantity with all n parameters, and obtains the function relationship between the simulation response quantity and all n parameters.

[0075] The correction parameter obtaining part performs the content described in step 5 above, constructs a target function considering multiple response quantities, and searches for a minimum value of the target function by using an improved genetic algorithm to obtain the size of the parameter of the simulation model after correction; the improved genetic algorithm is: after the mutation module of the genetic algorithm, a local optimization module is added to accelerate the convergence of the genetic algorithm; a variable range detection module is added to prevent the optimized value from exceeding the parameter range after local optimization; an intelligent stop module is added, and the difference between the optimized values before and after is less than the rated value as the stop condition, which ensures the accuracy of the optimized result while reducing the number of algorithm iterations.

[0076] The input display part is used for inputting operation instructions by the operator and displaying corresponding information according to the control instructions. For example, the input display part can statically or dynamically display the input, output data and processing process of the extraction part, the parameter determination part, the information matrix determination part, the function relationship determination part and the correction parameter obtaining part in the form of a data table or an image according to the control instructions, and can display the response quantity value obtained based on the function relationship between the response quantity and the parameter on the response position of the simulation model of the water conservancy and hydropower engineering, and display the structure shape, displacement or stress change corresponding to the response quantity value.

[0077] The control part is in communication with the extraction part, the parameter determination part, the information matrix determination part, the function relationship determination part, the correction parameter obtaining part and the input display part, and controls the operation of them.

[0078] The above examples are only illustrative of the technical solutions of the present application. The parameter correction method and device considering multiple response quantities in the numerical simulation process of the water conservancy and hydropower engineering involved in the present application are not limited to the content described in the above examples, but are subject to the scope defined in the claims. Any modification, supplement or equivalent replacement made by the person skilled in the art on the basis of the examples is within the scope claimed by the claims of the present application.

Claims

1. A parameter correction method considering multiple response quantities in the numerical simulation process of water conservancy and hydropower projects, characterized by: The steps include: Step 1: extract various measurement data from the actual operation of the engineering structure as the true response quantity; Step 2: Establish a simulation model based on the actual engineering structure dimensions and determine the parameters that need to be calibrated; Step 3: The parameters to be calibrated and the correction range of each parameter a (a=1,2,…,n) Preliminarily determine the information matrix M of the response surface model, where n is the number of parameters that need to be calibrated; Assume that the experiment contains N design points x in the spatial domain e , e=1,2,…,N, then the corresponding information matrix M is: Where x e is the parameter range χ a Each design point contains a set of value groups, and each value group contains the parameter values ​​of the 1st to nth parameters: x e =[Value1 Value2…Value n ];f1(x e ),f2(x e ),…,f w (x e ) is the parameter range χ a Continuous functions on the e ),f2(x e ),…,f w (x e ) to represent the spatial position of the parameter; Find the design point that maximizes the determinant of the information matrix, and then make the information matrix M as small as possible by selecting the experimental combination. The selected design point will be used as the experimental point to be substituted into the simulation model for calculation; Step 4: Substitute the numerical value group of each test point selected in the spatial domain into the simulation model for calculation, then obtain the calculated numerical value of the measurement point corresponding to the actual engineering measurement point position from the simulation model as the simulation response, and then substitute the numerical value group of each test point and the obtained simulation response into the response surface model considering the interaction of related non-normal parameters. For each simulation response, it is associated with all n parameters, and its functional relationship with all n parameters is obtained; Step 5: Construct an objective function that considers multiple response quantities, use an improved genetic algorithm to search for the minimum value of the objective function, and obtain the corrected parameters; The improved genetic algorithm is as follows: a local optimization module is added after the genetic algorithm mutation module to accelerate the convergence of the genetic algorithm; a variable range detection module is added to prevent the optimization value from exceeding the parameter range after local optimization; an intelligent stop module is added, and the difference between the optimization values ​​before and after is less than the rated value as the stopping condition, thereby ensuring the accuracy of the optimization results while reducing the number of algorithm iterations.

2. The parameter correction method considering multiple response quantities in the numerical simulation process of water conservancy and hydropower projects according to claim 1 is characterized by: in, In step 3, the principle for selecting the experimental combination is to minimize the number of spatial design points required while solving the response surface model.

3. The parameter correction method considering multiple response quantities in the numerical simulation process of water conservancy and hydropower projects according to claim 1 is characterized by: in, In step 4, the response surface model considering the interaction of relevant non-normal parameters is: Where y l is the lth response, b is the maximum expansion number of the set polynomial, q l is the constant term to be solved, β1~β w are the 1st to wth coefficients to be solved, f w (x w ) is the wth parameter term, each parameter term is an expression containing at least one parameter, f1(x1)~f w (x w ) contains all n parameters.

4. The parameter correction method considering multiple response quantities in the numerical simulation process of water conservancy and hydropower projects according to claim 1 is characterized by: in, In step 4, the response surface model f1(x1)~f w (x w ) Each item is subjected to F test to eliminate invalid items with poor fitting effect; each time the invalid items are eliminated, it is necessary to re-solve and calculate new coefficients and constant terms, and then perform F test on the response surface model of the response quantity. If the test result still does not meet the fitting threshold requirement, the value of b is increased, and the solution is re-solved and then tested until the test result of the response surface model of the response quantity meets the fitting threshold requirement, and an optimized response surface model reflecting the functional relationship between the response quantity and the parameters is obtained.

5. The parameter correction method considering multiple response quantities in the numerical simulation process of water conservancy and hydropower projects according to claim 4 is characterized by: in, The fitting threshold requirement is that the P value of the lack-of-fit F test result of the response surface model is greater than 5%.

6. The parameter correction method considering multiple response quantities in the numerical simulation process of water conservancy and hydropower projects according to claim 1, characterized in that: in, In step 5, the objective function considering multiple responses is: Among them, y l ′ is the actual response quantity of the engineering structure in actual operation.

7. The parameter correction method considering multiple response quantities in the numerical simulation process of water conservancy and hydropower projects according to claim 1 is characterized by: in, In step 5, the local optimization operation of the genetic algorithm mainly embeds the nonlinear programming function in the Matlab optimization toolbox into the mutation operation, substitutes the individuals of each generation of population after mutation into the planning function as the initial value, and uses the local optimal value found as the new individual chromosome to continue evolving; when the optimal value appears at the upper and lower boundaries of the parameter range, the result of the local optimization may exceed the parameter range. A zeroing indicator is set in the variable detection operation, and the offspring that exceeds the parameter range are not inherited; after multiple generations of inheritance, the local optimization value gradually stabilizes, and the cycle ends when the change range of the optimization value of n parameters is less than the set error value.

8. A parameter correction device considering multiple response quantities in the numerical simulation process of water conservancy and hydropower projects, characterized in that: include: The extraction part extracts various measurement data from the actual operation of the engineering structure as the real response quantity; The parameter determination department establishes a simulation model based on the actual engineering structure dimensions and determines the parameters that need to be calibrated; The information matrix determination part is composed of the parameters to be calibrated and the correction range of each parameter χ a (a=1,2,…,n) Preliminarily determine the information matrix M of the response surface model, n is the number of parameters that need to be calibrated; suppose the experiment contains N design points x in the spatial domain e , e=1,2,…,N, then the corresponding information matrix M is: Where x e is the parameter range χ a Each design point contains a set of value groups, and each value group contains the parameter values ​​of the 1st to nth parameters: x e =[Value1 Value2…Value n ];f1(x e ),f2(x e ),…,f w (x e ) is the parameter range χ a Continuous functions on the e ),f2(x e ),…,f w (x e ) to represent the spatial position of the parameter; Find the design point that maximizes the determinant of the information matrix, and then make the information matrix M as small as possible by selecting the experimental combination. The selected design point will be used as the experimental point to be substituted into the simulation model for calculation; The functional relationship determination unit substitutes the numerical value group of each test point selected in the spatial domain into the simulation model for calculation, and then obtains the calculated numerical value of the measuring point corresponding to the actual engineering measuring point position from the simulation model as the simulation response quantity. The numerical value group of each test point and the obtained simulation response quantity are then substituted into the response surface model that considers the interaction of relevant non-normal parameters. For each simulation response quantity, it is associated with all n parameters and its functional relationship with all n parameters is obtained; The modified parameter acquisition section constructs an objective function that considers multiple response quantities and uses an improved genetic algorithm to search for the minimum value of the objective function to obtain the modified parameters. The improved genetic algorithm includes: adding a local optimization module after the genetic algorithm mutation module to accelerate the convergence of the genetic algorithm; adding a variable range detection module to prevent the optimal value from exceeding the parameter range after local optimization; adding an intelligent stopping module, using the difference between the optimal value before and after being less than the rated value as the stopping condition, to ensure the accuracy of the optimization result while reducing the number of algorithm iterations; and The control unit is connected to the extraction unit, the parameter determination unit, the information matrix determination unit, the functional relationship determination unit, and the correction parameter acquisition unit to control their operations.

9. The parameter correction device considering multiple response quantities in the numerical simulation process of water conservancy and hydropower engineering according to claim 8, characterized in that: Also includes: The input display unit is connected to the control unit for communication, allowing the operator to input operation instructions and displaying corresponding information according to the control instructions.

10. The parameter correction device considering multiple response quantities in the numerical simulation process of water conservancy and hydropower engineering according to claim 9, characterized in that: in, The input display unit can statically or dynamically display the input and output data and processing procedures of the extraction unit, the parameter determination unit, the information matrix determination unit, the functional relationship determination unit, and the correction parameter acquisition unit in the form of a data table or image according to the control instructions, and can display the response quantity value obtained based on the functional relationship between the response quantity and the parameter at the model response position on the water conservancy and hydropower engineering simulation model, and display the structural shape, displacement or force change corresponding to the response quantity value.

Images