Inhaul cable parameter identification method based on improved hybrid optimization algorithm

Abstract

The invention discloses an inhaul cable parameter identification method based on an improved hybrid optimization algorithm. The method comprises the following steps: inputting structural geometric parameters and multi-order modal frequency; constructing a correction frequency model fusing tension, bending rigidity and boundary rotation rigidity; designing a weighted relative error objective function; performing global search by adopting an improved genetic algorithm (GA), and introducing logarithm uniform sampling and dynamic crossover / mutation probability; taking the GA optimal solution as an initial population, adopting an improved particle swarm optimization (PSO) to carry out local refinement, and introducing an adaptive inertia weight and random disturbance; and finally verifying the rationality of an identification result through a limit physical model. The method realizes high-precision synchronous identification of the cable force, the bending rigidity and the boundary rotation rigidity, has the advantages of fast convergence, high precision, strong robustness, low calculation cost and the like, and is suitable for health monitoring of structures such as bridge cables, wind power blades, cableways and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of bridge cable health monitoring and parameter identification, and particularly relates to a cable parameter identification method fusing a genetic algorithm (GA) and a particle swarm optimization (PSO) and introducing a physical model correction and a boundary condition, which is suitable for synchronous identification of multiple parameters such as cable force, bending stiffness and boundary rotational stiffness. BACKGROUND

[0002] In engineering structures such as bridges, cables, beam structures and the like, accurate identification of the physical parameters (such as cable force, bending stiffness and boundary conditions) of the structure is of great significance for structural health monitoring, safety evaluation and maintenance decision-making.

[0003] Traditional parameter identification methods often rely on single-mode frequency or static response, and have the following shortcomings:

[0004] 1. Low parameter identification accuracy, easily affected by measurement errors;

[0005] 2. The boundary condition processing does not consider the elastic boundary and ignores the boundary rotational stiffness;

[0006] 4. It is difficult to identify multiple coupled parameters (such as tension, stiffness and boundary stiffness) at the same time;

[0007] Therefore, there is an urgent need for a structure parameter identification method with strong global search ability and capable of processing multi-modal frequency data and complex boundary conditions. SUMMARY

[0008] In view of the above shortcomings of the prior art, the present application aims to provide a cable parameter identification method based on an improved hybrid optimization algorithm, which solves the problems of low identification accuracy, poor robustness and imperfect boundary processing in the prior art, and realizes high-precision synchronous identification of tension, bending stiffness and boundary rotational stiffness.

[0009] The present application adopts the following technical solution: a cable parameter identification method based on an improved hybrid optimization algorithm, comprising:

[0010] Step S1: inputting structural geometric parameters, mass distribution and multi-order modal frequency measurement values;

[0011] Step S2: constructing a frequency calculation model fusing bending stiffness, tension and boundary rotational stiffness;

[0012] Step S3: designing a multi-modal objective function based on frequency relative error weighting;

[0013] Step S4: global search by improved genetic algorithm (GA) to obtain preliminary optimal parameters; logarithmic uniform distribution sampling is used for population initialization to avoid exponential overflow; tournament selection, simulated binary crossover and polynomial mutation are introduced; crossover and mutation probabilities are dynamically adjusted;

[0014] Step S5: local fine-tuning by improved particle swarm optimization (PSO) using GA results as initial population; self-adaptive inertia weight and learning factor are introduced; random disturbance is introduced periodically to enhance the ability to jump out of local optimum;

[0015] Step S6: model validation is performed on the identification results to determine their rationality under extreme physical conditions;

[0016] Step S7: output the final identification parameters and their corresponding frequency fitting errors.

[0017] As a preferred, the frequency calculation model in step S2 is:

[0018]

[0019] where f n is the nth order frequency, n is the modal order, T is the tension, EI is the bending stiffness, k is the boundary rotational stiffness, m is the unit length mass, and L is the structure length.

[0020] As a preferred, the multi-modal objective function in step S3 is:

[0021]

[0022] where f meas,i is the measured value; f calc,i is the calculated value; w i is the modal weight, which decreases with the modal order, n i is the modal order.

[0023] As a preferred, the improved genetic algorithm in step S4 includes logarithmic uniform distribution sampling for population initialization, tournament selection, simulated binary crossover, polynomial mutation, and dynamic adjustment of crossover / mutation probabilities.

[0024] As a preferred, the improved particle swarm optimization in step S5 includes self-adaptive inertia weight and learning factor, speed / position boundary control, periodic random disturbance, and initial population generated partially from genetic algorithm optimal solutions; Gaussian random disturbance is introduced every 50 generations to enhance the ability to jump out of local optimum; speed limit is 20% of the search space.

[0025] As preferred, the limit physical condition verification in the step S6 comprises: simulating ideal string vibration (EI≈0, k≈0); simulating pure bending beam (T≈0, k≈0); and comparing whether the theoretical frequency is consistent with the model calculation frequency.

[0026] As preferred, the frequency is a measured frequency obtained by an acceleration sensor.

[0027] Advantages:

[0028] (1) The present application can identify three key structural parameters at one time by constructing a modified frequency model fusing tension, bending stiffness and boundary rotational stiffness, combining with multi-modal frequency data, breaking through the limitation of traditional method which can only identify a single parameter or needs to identify step by step, and significantly improving the identification efficiency and engineering applicability.

[0029] (2) The improved genetic algorithm (GA) is used for global search, and then the improved particle swarm optimization (PSO) is introduced for local fine-tuning with the GA optimal solution as the initial population, which takes into account the global exploration and local optimization ability, and the identification accuracy is better than that of traditional single algorithm, the convergence speed is fast, and the identification accuracy is high.

[0030] (3) The limit physical model verification mechanism is introduced, and typical structural working conditions such as ideal string and pure bending beam are simulated respectively to verify the rationality of the identification result under the theoretical limit condition, to ensure that the identification parameters not only have optimal values, but also have physical meaning and engineering interpretability, and to improve the scientificity and reliability of the method.

[0031] (4) The target function adopts the form of relative error, which reduces the amplification effect of high frequency mode on error contribution; the weighting strategy makes the weight of high-order mode decrease, which effectively suppresses the interference of measurement noise on the identification result; the algorithm is not sensitive to initial value, and has strong fault tolerance and engineering adaptability. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 is a flow chart of the method of the present application;

[0033] Figure 2 is a GA-PSO hybrid structure diagram of the present application;

[0034] Figure 3 is a frequency fitting comparison diagram (measured vs. calculated) of the present application;

[0035] Figure 4 is a parameter convergence trajectory diagram (T, EI, k) of the present application;

[0036] Figure 5 is a model test diagram of the present application. DETAILED DESCRIPTION

[0037] The application will be further described below in conjunction with the accompanying drawings and specific embodiments of the application.

[0038] As Figures 1-5 shown, a cable parameter identification method based on an improved hybrid optimization algorithm, comprising:

[0039] 1. Using high-sensitivity vibration sensors to measure the vibration of the steel strand and record the frequency value and structural parameters;

[0040] The high-sensitivity vibration sensor is attached to the cable and is located 5-10 cm away from the top or bottom. The steel strand is subjected to different tension forces: 50kN, 100kN, 150kN, and 200kN. The cable end is knocked, and the steel strand vibration signal after each knock is recorded to ensure that enough sample signals are recorded to cover different vibration modes. At the same time, the structural parameters of the steel strand are recorded, including the length L1 of the steel strand, the linear density p of the steel strand, and the corresponding cable force;

[0041] 2. Based on the Euler-Bernoulli beam theory and considering the effects of tension and boundary rotational stiffness, the calculation model is derived by energy method:

[0042]

[0043] According to the energy method, the modal shape is assumed to be a sine function:

[0044]

[0045] Bending potential energy:

[0046] Tension potential energy:

[0047] Boundary potential energy:

[0048] Calculate kinetic energy:

[0049] According to the energy conservation (maximum potential energy equals maximum kinetic energy), an approximate frequency equation is obtained:

[0050]

[0051] Where f n is the nth order frequency, T is the tension, EI is the bending stiffness, k is the boundary rotational stiffness, m is the unit length mass, L is the structure length, ω is the transverse vibration circular frequency, A is the amplitude, and n is the modal order;

[0052] 3. Construct a weighted objective function and perform hybrid algorithm optimization, introducing adaptive inertia weight and learning factor

[0053]

[0054] w i is the modal weight, which decreases with the modal order;

[0055] f meas,i is the measured value; f calc,i is the calculated value;

[0056] w is the inertia weight, usually set to 0.9 to 0.4 and gradually reduced during the iteration process;

[0057] c is the learning factor, usually set to 2 to 1.5 and gradually reduced during the iteration process;

[0058] 4. Model verification is performed to ensure that it can correctly degenerate to the known theoretical solution in two extreme cases

[0059] 5. The identification result is output, and the measured value is compared with the calculated value, as shown in Table 1.

[0060] Table 11 Identification result of m steel strand

[0061]

[0062]

[0063] The embodiments of the present application are described in detail in combination with the drawings, but the present application is not limited to the described embodiments. Various changes, modifications, variations or replacements made within the essential scope of the present application are also within the protection scope of the present application for ordinary skilled in the art.

Claims

1. A cable parameter identification method based on an improved hybrid optimization algorithm, characterized in that: The method comprises the following steps: Step S1: inputting structural geometric parameters, mass distribution and multi-modal frequency measurement values; Step S2: constructing a frequency calculation model integrating bending stiffness, tension and boundary rotational stiffness; Step S3: designing a multi-modal objective function based on frequency relative error weighting; Step S4: performing global search by using an improved genetic algorithm to obtain preliminary optimal parameters; logarithmic uniform distribution sampling is used for population initialization to avoid exponential overflow; tournament selection, simulated binary crossover and polynomial mutation are introduced; and the crossover and mutation probabilities are dynamically adjusted; Step S5: performing local fine-tuning by using an improved particle swarm algorithm with the GA result as the initial population; adaptive inertia weight and learning factor are introduced; and random disturbance is introduced periodically to enhance the ability to jump out of local optimum; Step S6: verifying the model based on the identification result to determine its rationality under extreme physical conditions; Step S7: outputting the final identification parameters and the corresponding frequency fitting error.

2. The cable parameter identification method based on the improved hybrid optimization algorithm according to claim 1, characterized in that: The frequency calculation model in step S2 is as follows: where f n is the nth order frequency, n is the mode order; T is the tension, EI is the bending stiffness, k is the boundary rotational stiffness, m is the unit length mass, and L is the structure length.

3. The cable parameter identification method based on the improved hybrid optimization algorithm according to claim 2, characterized in that: The multi-modal objective function in step S3 is as follows: where f meas,i is the measured value; f calc,i is the calculated value; w i is the modal weight, which decreases with modal order; n i is the modal order.

4. The cable parameter identification method based on the improved hybrid optimization algorithm according to claim 3, characterized in that: The improved genetic algorithm in step S4 comprises logarithmic uniform distribution sampling for population initialization, tournament selection, simulated binary crossover, polynomial mutation and dynamic adjustment of the crossover / mutation probabilities.

5. The cable parameter identification method based on the improved hybrid optimization algorithm according to claim 4, characterized in that: The improved particle swarm algorithm in step S5 comprises adaptive inertia weight and learning factor, speed / position boundary control, periodic random disturbance and initial population generated partially from the optimal solution of the genetic algorithm; Gaussian random disturbance is introduced every 50 generations to enhance the ability to jump out of local optimum; and the speed limit is 20% of the search space.

6. The cable parameter identification method based on the improved hybrid optimization algorithm according to claim 5, characterized in that: The extreme physical condition verification in step S6 comprises simulating ideal string vibration, EI≈0 and k≈0; simulating pure bending beam, T≈0 and k≈0; and comparing whether the theoretical frequency and the model calculation frequency are consistent.

7. The cable parameter identification method based on the improved hybrid optimization algorithm according to claim 6, characterized in that: The frequency is measured by an acceleration sensor.

Images