Tunnel lining earthquake accumulated damage evaluation method suitable for vibrating table test
Technical Field
The invention belongs to the field of tunnel design, and particularly relates to a tunnel design safety evaluation method, in particular to a tunnel lining earthquake accumulation damage evaluation method suitable for a vibration table test.
Background
In the process of designing a tunnel anti-seismic structure, when the structure is complex and the existing theoretical calculation cannot solve the problem satisfactorily, researchers often observe the seismic response and the damage form of the structure directly by means of a vibrating table model test so as to evaluate the integral anti-seismic capacity of the structure. The earthquake simulation vibrating table is used as key equipment in the earthquake engineering research and is mainly applied to the structural earthquake resistance test. The device can simulate the earthquake environment, thereby determining the dynamic characteristic of the engineering structure, checking the earthquake resistance of the structure and the damage mechanism thereof under the action of earthquake force, and helping designers to perfect the earthquake resistance design theory and method.
The raw data in the shaking table model test is usually recorded by various sensors (such as dynamic strain gauges, acceleration sensors, displacement sensors, dynamic soil pressure sensors, etc.) arranged on the model structure. Common processing modes of the original data are as follows: (1) Analyzing the frequency spectrum scanning to obtain an amplitude-frequency curve, and knowing the change conditions of the self-vibration frequency, damping and vibration mode of the structure; (2) The acceleration peak value of the measuring point position can be obtained through the acceleration record, and the acceleration amplification coefficient can be obtained by dividing the acceleration peak value input by the table top; (3) Integrating the acceleration data to obtain a displacement time curve of the measuring point, and further obtaining a relative table top displacement peak value and an interlayer displacement peak value; (4) Calculating the time when the crack possibly appears at the position and the deformation condition of the structure by utilizing the strain peak value measured by the dynamic strain gauge; (5) The displacement peak value can also be obtained through a displacement time curve recorded by a displacement sensor; (6) The acceleration sensor arranged along the tangential direction is utilized to calculate the relative table surface torsional deformation peak value and the interlayer torsional deformation peak value; (7) description of test phenomena and crack development.
In the earthquake process, the lining structure mainly generates stretching and compression damage. Local effects exist on two kinds of damage of different characteristic parts of the tunnel lining section, and the damage of a lining structure is related to various factors such as terrain conditions, stratum lithology, burial depth, space position and the like. The damage of the lining structure under the action of the earthquake is not only related to the maximum deformation, but also related to the accumulated damage caused by the low cycle fatigue effect of the lining structure generated by the earthquake reciprocating action. In the low cycle fatigue effect process of the concrete material, the performance of the concrete material is gradually reduced due to accumulation of plasticity and degradation of rigidity until the concrete material is completely destroyed. However, in the existing data processing process, the accumulated damage factor is not taken into consideration, so that the existing data processing process has the omission, and the existing theoretical calculation cannot completely study the inherent mechanism of the accumulated damage.
In order to fully reflect the dynamic damage degree of the lining structure, the limitations of indexes (or corresponding internal force indexes) such as peak acceleration (speed and displacement), peak dynamic strain, peak dynamic soil pressure, frequency spectrum characteristics and the like in the traditional acceleration, dynamic strain and dynamic soil pressure analysis method are broken through. Based on the actual state of damage of the lining structure of the earthquake engineering, the dynamic strain test data analysis of the lining structure is tested according to the vibrating table, the accumulated damage of elastoplastic deformation characteristics is considered, the dynamic damage problem of the lining structure is analyzed more comprehensively, and a more perfect evaluation method is provided, so that the technical problem which is urgently needed to be solved in the field is provided.
Disclosure of Invention
The invention aims to provide a tunnel lining earthquake accumulated damage evaluation method suitable for a vibrating table test, which considers accumulated damage of elastoplastic deformation characteristics of a lining structure, and can better describe mechanical behaviors of the tunnel lining structure under the action of reciprocating load by combining a local weighted regression machine learning means, so that the damage process under the action of cyclic load such as earthquake and the like can be displayed more accurately. The safety state of the lining structure after earthquake can be quantitatively evaluated from the part (or the whole), and a new approach is provided for the fine analysis of the earthquake response process of the lining structure.
Therefore, the invention adopts the following technical scheme:
a tunnel lining earthquake accumulation damage evaluation method suitable for a vibrating table test comprises the following steps:
1) Sensor arrangement: constructing a tunnel primary lining model for a test, designing a layout scheme of dynamic strain sensors according to the structure of the tunnel primary lining model, pasting and layout the dynamic strain sensors on the inner surface and the outer surface of the lining of the tunnel primary lining model, wherein the positions of the dynamic strain sensors on the inner surface and the outer surface correspond to each other one by one, and connecting signals of the dynamic strain sensors to an upper computer after the layout is completed; in order to improve the bonding strength of the dynamic strain sensor, glue can be used for bonding; in addition, in order to shorten the test period and reduce the test cost, a plurality of sensors such as an acceleration sensor, a displacement sensor, a soil pressure sensor or other sensors for monitoring can be arranged in a single test, and the sensors can be monitored simultaneously to obtain required test data;
2) And (3) earthquake testing: placing the tunnel primary lining model in the step 1) on a vibrating table, filling a vibrating table test model according to test requirements, loading test seismic waves for testing, and transmitting data generated in a test process to an upper computer for storage by a dynamic strain sensor;
3) Curve regression: taking test data acquired by a single dynamic strain sensor, and drawing a dynamic strain-time curve graph; then adopting a local weighted regression algorithm, carrying out weighted regression calculation on dynamic strain time curve data by setting the window length of a data segment, and obtaining dynamic strain smooth values corresponding to all time points in the window length; sequentially obtaining dynamic strain smooth values in other windows, and connecting the smooth values to obtain a fitting local regression dynamic strain-time curve corresponding to the dynamic strain sensor test data;
then taking test data obtained by other dynamic strain sensors, and repeating the steps in the step 3) to obtain a fitting local regression dynamic strain-time curve of each other dynamic strain sensor;
the calculation principle of the local weighted regression algorithm is as follows:
one pointxTaking a piece of data with a specified length as a center, adopting a weight function to perform weighted linear regression on the piece of data, and recordingIs the central value of the regression line, wherein +.>Is the corresponding value of the fitted curve. For all ofnData points can then be madenA weighted regression line, wherein the connection line of the central value of each regression line is the data of the segmentLoessA curve.
Wherein: in local weighted regression, the loss function within each data segment is as follows:①
wherein:h θ is a parameterθVector expression of (i) i.e.;/>Is the abscissa of any data point; />Is->Corresponding data true values; />Is->The expression is as follows:②
wherein:xcharacteristic data of the new prediction sample; parameters (parameters)tThe change rate for controlling the weight can be manually given in advance; thenThere are two important properties:
(1) if it isThen->
(2) If it isThen->
Thus, for the off-prediction sample dataxThe weight of the near point is larger and is far from the predicted sample dataxThe far points have small weights.
Thus, finding by iterative solution calculationsJ(θ)Can obtain the target by the minimum value of (2)h θ The output value calculated at this timeI.e. smoothed values obtained by locally weighted regression.
4) Extracting PDS and RS: respectively extracting a local regression dynamic strain peak PDS and a residual dynamic strain value RS corresponding to each dynamic strain sensor through the fitting local regression dynamic strain-time curve graph obtained in the step 3);
5) Calculating PEC: respectively calculating the plastic deformation index of each dynamic strain sensor according to the data obtained in the step 4)PEC:
PEC=RS/PDS③
In the plastic deformation indexPEC (plastic effect coefficient, abbreviated hereinafter as PEC)Is used to indicate the extent of plastic deformation, i.e. the extent of failure deformation, at any point on the tunnel lining.PEC﹤1,PECThe larger the lining is, the greater the unrecoverable deformation degree of the tunnel lining is, and the earthquake accumulated damage effect of the tunnel is shownSCFEThe stronger.
After each seismic wave loading, the data signals of the dynamic strain sensor cannot return to the data zero point by themselves. I.e. the tunnel lining model is considered to produce an unrecoverable interiorDamage deformation, i.e. residual strainRS(Residual Strain, hereinafter abbreviatedRS) Meets the basic theoretical cognition range of elastoplastics of a concrete plastic damage model. Damage index in reference structure inelastic damage performance evaluation methodD I Is based on the basic idea of providing a plastic deformation index of tunnel lining under the action of earthquakePEC(plastic effect coefficient, hereinafter abbreviated as "abbreviation")PEC)。PECFor any point on different characteristic parts of the lining structureRSPeak value of dynamic strainPDS(Peak Dynamic Strain, hereinafter abbreviated as "abbreviation")PDS) Is a ratio of (2). The basic principle is as follows:
as shown in FIG. 2, the concrete sample has a damaged real state in a uniaxial tensile stress state, and the real state contains a plurality of microcracks and micropore holes. If the tensile force applied to the concrete member is assumed to beTA cross-sectional area ofAThen there is cauchy stress。
As shown in FIG. 3, if the concrete member has a cross-sectional area in a non-destructive virtual state under uniaxial tension stress of the concrete memberIs thatAThe "remaining effective area" of microcracks and microperforations is subtracted, the effective stress tensor->The expression is as follows:
④
introduction of damage parameters⑤
Wherein:the cross-sectional area of the concrete member with micro cracks and micro holes is shown;
Aa cross-sectional area representing a damaged condition of the real concrete member;
according to the damage parameter concept of the formula (5),microcrack and micropipe damage, i.e., residual strain as referred to in equation (3)RS,ANamely, the peak value of dynamic strain mentioned in the formula (3)PDSDefining the plastic deformation index of tunnel lining under the action of earthquakePEC。
6) Drawing a PEC variation spectrogram: with the dynamic strain sensor position as the horizontal axis, the seismic intensity as the vertical axis and the plastic deformation indexPECIs a vertical shaft; respectively drawing PEC-seismic intensity-lining characteristic part change spectrograms of the dynamic strain sensor on the inner surface of the tunnel lining, and the PEC-seismic intensity-lining characteristic part change spectrograms of the dynamic strain sensor on the outer surface of the tunnel lining; and evaluating the deformation stage of the tunnel lining earthquake accumulation and destruction process through the PEC variation spectrogram.
The invention breaks through the limitations of indexes (or corresponding internal force indexes) such as peak acceleration (speed and displacement), peak dynamic strain, peak dynamic soil pressure, spectrum characteristics and the like in the traditional acceleration, dynamic strain and dynamic soil pressure analysis method, creatively provides a tunnel lining earthquake accumulation damage (PEC) evaluation method suitable for a vibration table test based on the actual state of damage of a seismic engineering lining structure, according to the analysis of vibration table test lining structure dynamic strain test data, considering the accumulated damage of elastoplastic deformation characteristics and combining with a machine learning means. Based on the basic principle of concrete elastoplastic damage, the plastic deformation index PEC of the tunnel lining with earthquake action is provided, and the quantitative evaluation of the earthquake accumulation damage of the tunnel lining can be realized by accurately extracting the local regression dynamic strain peak value and the residual dynamic strain value.
Drawings
FIG. 1 is a flow chart of a cumulative damage assessment method of the present invention;
FIG. 2 is a true state of damage to a concrete sample in a uniaxial tensile stress state;
FIG. 3 is a non-destructive hypothetical state of a concrete test piece under uniaxial tension stress;
FIG. 4 is a three-dimensional view of a vibrating table model sensor arrangement in example 1;
FIG. 5 is a sectional view of a tunnel lining model dynamic strain sensor arrangement in example 1;
FIG. 6 is a schematic of the dynamic strain versus time graph and dynamic strain smoothing curve of example 1;
FIG. 7 is a PEC-seismic intensity-lining feature variation spectrum of a tunnel lining inner surface dynamic strain sensor of example 1;
FIG. 8 is a PEC-seismic intensity-lining feature variation spectrum of a tunnel lining outer surface dynamic strain sensor of example 1;
FIG. 9 is a three-dimensional view of a vibrating table model sensor arrangement in example 2;
FIG. 10 is a sectional view showing the arrangement of dynamic strain sensors in a tunnel lining model in example 2;
FIG. 11 is a schematic of the dynamic strain-time graph and dynamic strain smoothing curve of example 2;
FIG. 12 is a PEC-seismic intensity-lining feature variation spectrum of a tunnel lining inner surface dynamic strain sensor of example 2;
fig. 13 is a PEC-seismic intensity-lining characteristic change spectrum of the tunnel lining outer surface dynamic strain sensor of example 2.
Detailed Description
The invention is further illustrated by the following examples in conjunction with the accompanying drawings:
example 1
Tunnel crossing main slip plane space power coupling system earthquake accumulation damagePECEvaluation
(1) The dynamic response of a tunnel structure crossing a main sliding surface under the action of an earthquake is researched by adopting a vibration table test method, the layout schemes such as a tunnel lining dynamic strain sensor and the like are designed, the dynamic strain sensor is arranged according to the design schemes and is adhered to the inner surface and the outer surface of a tunnel lining model by using 502 glue, and the design of the vibration table test model and the layout of sensor elements are shown in figures 4 and 5.
(2) And (3) filling a vibrating table test model according to the sensor layout scheme in the step (1), loading a seismic wave test, and obtaining dynamic strain monitoring data of the lining characteristic part of the vibrating table test tunnel.
(3) And (3) drawing a dynamic strain-time curve graph by taking test data acquired by a single dynamic strain sensor, and fitting by adopting a local weighted regression algorithm to obtain a dynamic strain smooth curve through a formula (1). Repeating the steps, and sequentially obtaining fitting local regression dynamic strain-time curves corresponding to the test data of the rest dynamic strain sensors.
(4) Fitting the local regression dynamic strain-time curve graph to accurately extract the local regression dynamic strain peak value (max|of each characteristic part of the tunnel liningPDS|) And a residual dynamic strain value (max|)RS|) As shown in fig. 6.
(5) Accurately extracting a local regression dynamic strain peak value and a residual dynamic strain value, and calculating the plastic deformation index of the tunnel lining under the earthquake action through a formula (3)PEC. And respectively drawing PEC-seismic intensity-lining characteristic part change spectrograms (shown in fig. 7) of the dynamic strain sensors on the inner surface of the tunnel lining, and drawing PEC-seismic intensity-lining characteristic part change spectrograms (shown in fig. 8) of the dynamic strain sensors on the outer surface of the tunnel lining.
As shown in fig. 7 and 8, at the stage of 0.05 to 0.15 and g,PECthe slow increase indicates that the plastic deformation of the tunnel lining is small and is in an elastic deformation stage. In the stage of 0.15-0.3 g,PECthe increase of a certain degree indicates that the plastic deformation degree of the tunnel lining is gradually increased, and the tunnel lining starts to enter an elastoplastic stage.PECThe plastic deformation of the tunnel lining is greatly increased in the stage of 0.3-0.6 g, which indicates that the plastic deformation of the tunnel lining is sharply increased, and the tunnel lining starts to enter the plastic deformation stage.PECThe deformation stage of the tunnel lining structure is fully considered, so that the dynamic damage of the tunnel lining under the earthquake excitation has obvious cumulative effect. On the basis of fully considering the plastic deformation characteristics of the tunnel lining, the accumulated earthquake effect of the tunnel lining mainly comprises the following three stages: progressive destructive effect (slow deformation) stage<0.15 g),An initial effect (elastoplastic deformation) stage (0.15-0.3 g), and a plastic effect (plastic deformation) stage (0.3-0.6 g).
Example 2
Earthquake accumulation damage of tunnel crossing traction section sliding surface space power coupling systemPECEvaluation
(1) The dynamic response of a tunnel structure crossing a sliding surface of a traction section under the action of an earthquake is researched by adopting a vibration table test method, layout schemes such as a tunnel lining dynamic strain sensor and the like are designed, the dynamic strain sensor is arranged according to the design schemes and is adhered to the inner surface and the outer surface of a tunnel lining model by using 502 glue, and the design of the vibration table test model and the layout of sensor elements are shown in fig. 9 and 10.
(2) And (3) filling a vibrating table test model according to the sensor layout scheme in the step (1), loading a seismic wave test, and obtaining dynamic strain monitoring data of the lining characteristic part of the vibrating table test tunnel.
(3) And (3) drawing a dynamic strain-time curve graph by taking test data acquired by a single dynamic strain sensor, and fitting by adopting a local weighted regression algorithm to obtain a dynamic strain smooth curve through a formula (1). Repeating the steps, and sequentially obtaining fitting local regression dynamic strain-time curves corresponding to the test data of the rest dynamic strain sensors.
(4) Fitting the local regression dynamic strain-time curve graph to accurately extract the local regression dynamic strain peak value (max|of each characteristic part of the tunnel liningPDS|) And a residual dynamic strain value (max|)RS|) As shown in fig. 11.
(5) Accurately extracting a local regression dynamic strain peak value and a residual dynamic strain value, and calculating the plastic deformation index of the tunnel lining under the earthquake action through a formula (3)PEC. And respectively drawing PEC-seismic intensity-lining characteristic part change spectrograms (shown in fig. 12) of the dynamic strain sensor on the inner surface of the tunnel lining, and drawing PEC-seismic intensity-lining characteristic part change spectrograms (shown in fig. 13) of the dynamic strain sensor on the outer surface of the tunnel lining.
As shown in fig. 12 and 13, at the stage of 0.05 to 0.15 and g,PECapproximately linearly increasing, indicating that the plastic deformation of the tunnel lining is very small, basicallyIn the elastic deformation phase. In the stage of 0.15-0.3 g,PECthe increase of a certain degree indicates that the plastic deformation degree of the tunnel lining is gradually increased, and the tunnel lining starts to enter an elastoplastic deformation stage.PECThe plastic deformation of the tunnel lining is greatly increased in the stage of 0.3-0.4 g, which shows that the plastic deformation of the tunnel lining is sharply increased, and the tunnel lining starts to enter the plastic deformation stage.PECThe deformation stage of the tunnel lining structure is fully considered, so that the dynamic damage of the tunnel lining under the earthquake excitation has obvious cumulative effect. On the basis of fully considering the plastic deformation characteristics of the tunnel lining, the accumulated earthquake effect of the tunnel lining mainly comprises the following three stages of an elastic deformation stage #<0.15 g), an elastoplastic deformation stage (0.15-0.3 g), and a plastic deformation stage (0.3-0.4 g).