KR20160073614A

KR20160073614A - Wavelet-based nonlinear autoregressive moving average model for identification of smart concrete structures under high impact loads

Info

Publication number: KR20160073614A
Application number: KR1020140182117A
Authority: KR
Inventors: 김이석; 남윤영
Original assignee: 순천향대학교 산학협력단
Priority date: 2014-12-17
Filing date: 2014-12-17
Publication date: 2016-06-27

Abstract

The present invention relates to a wavelet-based nonlinear auto-regressive moving average modeling method for identification of a smart concrete structure under a high load. The method includes: a first step of receiving data form a smart concrete structure in which an MR damper is installed as input; a second step of applying wavelet dissembling to the inputted data; a third step of applying a nonlinear auto-regressive moving average (NARMA) model to the data to which the wavelet dissembling is applied; and a fourth step of predicting the moving of the smart concrete structure from the data to which the model is applied.

Description

[0001] Wavelet-based nonlinear autoregressive moving average modeling method for identification of smart concrete structures under high load [

The present invention relates to a wavelet-based nonlinear autoregressive moving average model (WNARMA) modeling method for identifying a smart concrete structure under high load, the method comprising: A second step of applying wavelet decomposition to the input data, a third step of applying a nonlinear autoregressive moving average model (NARMA) model to the data to which the wavelet decomposition is applied, And a fourth step of predicting the behavior of the smart concrete structure from the data to which the model is applied.

In order to perform vibration control, which is one of the main functions of smart concrete structures, it is necessary to identify suitable systems for smart concrete structures.

System identification can be divided into parametric and nonparametric methods, which use finite parameters such as structural mass, stiffness, and damping ratio to represent the physical quantity of the structural system. In the parametric method, a sufficient number of modal parameters must be obtained to identify the correct system model. However, most of the large civil engineering structures have nonlinear behavior, making it difficult to obtain the above parameters and time consuming.

The nonparametric method can predict and evaluate an infinite number of parameters without a complete understanding of the physical system, and the nonparametric method can predict the structural response even if the physical quantity of the identification model can not be expressed. Nonparametric system identification can only be applied to a set or output of input output data.

Output methods based on system identification have become increasingly important since input data is not always available. In other words, it is difficult to obtain input data due to uncertain natural excitation such as traffic congestion, wind, and waves. Therefore, the output-based SI under ambient excitation (OSIA) method is one of the most preferable approaches, and it is possible to recognize dynamic attributes in a real environment by using this method.

The autoregressive moving average (ARMA) time series model is one model that enables the OSIA method and is used by many construction engineers. However, the modeling process of the automatic session moving average model assumes that the dynamic system operates linearly. For example, assume that the predicted output response of the structure is a linear function of the previous value.

Nonlinear models can better predict the nonlinear behavior of structures, and some nonlinear autoregressive moving average (NARMA) models combining automatic station moving average and nonlinear models have been applied by some researchers. However, the nonlinear autoregressive moving average (NARMA) model has a disadvantage in that it requires a long calculation time, and thus has many limitations in implementation for actual structural modeling.

BACKGROUND ART [0002] The background art of the present invention is disclosed in the Korean Intellectual Property Office (KIPO) No. 10-0551986 on Mar. 02, 2006.

In order to solve the above-mentioned problems, the present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to propose a new system technique for predicting nonlinear behavior of a smart concrete structure by combining a wavelet transform method with a non- And a noise-reduced wavelet-based nonlinear automatic session moving average (WNARMA) modeling method.

A method for modeling a nonlinear WBARM (WNARMA) modeling method for identifying a smart concrete structure in a high load according to an embodiment of the present invention, the method comprising the steps of: A second step of applying wavelet decomposition to the input data, a third step of applying a wavelet decomposition applied data to a nonlinear autoregressive moving average (NARMA) model, a third step of applying the wavelet decomposition to the smart concrete structure And a fourth step of predicting the behavior.

The behavior of the smart concrete structure under high load can be predicted by using the system identification method using the Wavelet-based nonlinear automatic station moving average model (WNARMA) according to an embodiment of the present invention.

Figure 1 shows a flow chart of a wavelet based nonlinear autoregressive moving average model (WNARMA).
Figure 2 shows a sensor and environment for obtaining data from a smart concrete structure.
Figure 3 shows the overall structure of the drop tower.
Figure 4 shows the structure of a reinforced concrete beam used in a drop tower.

The details of other embodiments are included in the detailed description and drawings.

BRIEF DESCRIPTION OF THE DRAWINGS The advantages and features of the present invention and the manner of achieving them will become apparent with reference to the embodiments described in detail below with reference to the accompanying drawings. The present invention may, however, be embodied in many different forms and should not be construed as being limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the invention to those skilled in the art. Is provided to fully convey the scope of the invention to those skilled in the art, and the invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout the specification.

Wavelet Transform is a signal transformation technique widely used in recent signal processing fields in addition to FFT (Fast Fourier Transform) and STFT (Short Time Fourier Transform). Among them, FFT is widely used in signal processing and solution of differential equations. However, FFT has a disadvantage in that it can not simultaneously grasp time information and frequency information of a signal because it can only analyze signals in a frequency domain. To overcome this limitation, a STFT with a time-dependent weighting function called a window function was introduced into the Fourier building block. However, since the STFT uses an independent window function with a weighting function applied to the existing Fourier transform, the analysis domain is always constant with respect to time-frequency, and a non-stationary signal whose stochastic characteristic changes with time changes It can not be analyzed efficiently. Since the wavelet transform compensates for these drawbacks and uses the resulting functions as a building block by scaling and moving the mother wavelet by expansion or contraction without using the window function, Frequency analysis.

1 shows a flow chart of a wavelet-based nonlinear autoregressive moving average model (WNARMA) of the present invention. In a first step 110 of receiving motion data from a smart concrete structure equipped with an MR damper, motion data is collected through five sensors attached to a smart concrete structure, and includes a first sensor for collecting acceleration information, A third sensor for collecting deformation information, a fourth sensor for measuring an impact force, and a fifth sensor for collecting status information of the MR damper.

The second step 120 of applying wavelet decomposition to the motion data collected from the five sensors is composed of a total of N levels, and the scaling function on each level is composed of a convolution of a scale function and a wavelet function ). This can decompose the low frequency region into the high frequency region and the low frequency region again. Frequency bands (eg, noise) that are not so apparent in the original signal would have very small amplitudes and are filtered by the data reduction without loss of any key information through discrete wavelet transform.

Equation 1 represents a scaling function and a wavelet function. The scaling function operates on a low pass filter and the wavelet function operates on a high pass filter.

[Equation 1]

In the third step 130, the nonlinear autoregressive moving average (NARMA) model is applied to the data to which the wavelet decomposition is applied in the second step 120, and the nonlinear autoregressive moving average model can be expressed by the following equation (2).

&Quot; (2) "

Here, A and B denote an auto regression and a moving average model, respectively, and e (n) denotes a noise or prediction error,

Wow

Represents the coefficients of the automatic regression and moving average model.

The nonlinear autoregressive moving average model performs a least squares analysis with a linearly independent vector using Equation (3).

&Quot; (3) "

Here,

ego,

to be. Also, R represents the number of selected linear independent vectors. A criterion function is used to minimize the error e (n) in the least squares analysis, and the reference function is as shown in Equation (4).

&Quot; (4) "

The wavelet-based nonlinear autoregressive moving average model of the present invention combines Equations (1) and (2). The derived wavelet-based nonlinear autoregressive moving average model is shown in Equation (5).

&Quot; (5) "

It is useful to decompose time series data with time and frequency by combining wavelet transform and nonlinear autoregressive moving average model, and it is more efficient than existing nonlinear autoregressive moving average model because it can reduce calculation time and noise.

Figure 2 shows the sensors and environment for obtaining data from a smart concrete structure. In order to obtain information such as acceleration, deflection, strain, impact force and the like from the smart concrete structure, five sensors are used in the smart concrete structure in the present invention. The acceleration information is obtained from two accelerometers 210 and the deflection information is measured from one ACT LVDT displacement transducer 220 located in the middle of the beam. The deformation information is measured from the strain gage 230 of one MM product and the impact force is adjusted using the Central HTC-10K type load cell 240 having a capacity of 4,500 Kg. The MR damper 250 is a reinforced concrete beam 260) in order to mitigate the impact.

In the present invention, the performance of the wavelet-based nonlinear autoregressive moving average model of the present invention was measured through a drop tower test. In the drop tower shown in FIG. 3, an MR-damper is used for vibration control. The MR-damper effectively controls the vibration generated in the building and secures the structural stability. In order to extend the durability of the building, Fluid is utilized. In Fig. 3, the MR-damper is installed on the first floor, but it can be appropriately installed on other layers as necessary. The drop tower has a capacity of 22,500kg and uses a reinforced concrete beam of 10x10x100cm. The reinforcing bars have six axial reinforcing bars, each with a diameter of 0.75cm and a tensile strength of 248MP. The diameter of the wire is 0.25 cm and each is arranged at an interval of 7.5 cm. The compressive strength and the modulus of elasticity of the concrete are 26 MP and 16 GPa, respectively, and the figure is shown in FIG.

It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. It will be easy to understand. If the practice of such improvement, alteration, substitution or addition is within the scope of the following claims, the technical idea is also deemed to belong to the present invention.

Claims

A wavelet-based nonlinear autocore moving average modeling method for identifying a smart concrete structure in a high load,
The method includes: a first step of receiving data from a smart concrete structure provided with an MR damper;
A second step of applying wavelet decomposition to the input data;
A third step of applying the wavelet decomposition applied data to a nonlinear autoregressive moving average (NARMA) model;
And a fourth step of estimating the behavior of the smart concrete structure from the data to which the model is applied.

The method of claim 1, wherein the data of the first step is collected through five sensors attached to the smart concrete structure, and includes a first sensor for collecting acceleration information, a second sensor for collecting deflection information, A third sensor for collecting, a fourth sensor for measuring the impact force, and a fifth sensor for collecting the state information of the MR damper.

The method of claim 1, wherein the wavelet-based nonlinear autoregressive moving average model

silver

The modeling method comprising:

4. The method of claim 3, wherein the wavelet-based nonlinear autoregressive moving average model comprises a linear least-

, And

ego,

The modeling method comprising:

The method of claim 4, wherein the criterion function is used to minimize the error e (n) in the least squares analysis,

The modeling method comprising: