Search

CN-122017992-A - Earth-rock dam hidden danger full-wave inversion method based on inter-well observation system

CN122017992ACN 122017992 ACN122017992 ACN 122017992ACN-122017992-A

Abstract

The invention discloses a full waveform inversion method of hidden danger of an earth-rock dam based on an inter-well observation system, which comprises the steps of arranging excitation points and detection points on a dam crest and an upstream dam slope and a downstream dam slope along the cross section of the earth-rock dam, constructing the inter-well observation system to collect seismic data, establishing a hidden danger-free initial model, performing forward modeling as a current model, constructing an objective function based on observation and simulation records, performing wave field reverse pushing through a residual record and calculating a physical parameter gradient of the current model by combining the forward modeling wave field when the objective function value does not reach a preset threshold, then adopting a conjugate gradient method to process to obtain a conjugate gradient, then constructing a preconditioning operator for pressing surface wave interference and accelerating deep updating in the dam body to correct the conjugate gradient, acquiring an optimal inversion step through step search and updating the model, and iterating until the objective function converges to obtain the physical parameter model reflecting the hidden danger of the earth-rock dam. The invention can effectively improve the accuracy and inversion efficiency of hidden danger detection of the earth-rock dam.

Inventors

  • XU HAO
  • JIANG XIAOHUAN
  • MA YUXI

Assignees

  • 武汉轻工大学

Dates

Publication Date
20260512
Application Date
20260309

Claims (10)

  1. 1. The earth-rock dam hidden danger full waveform inversion method based on the inter-well observation system is characterized by comprising the following steps of: S1, arranging excitation points and detection points on the surfaces of a dam crest and upstream and downstream dam slopes along the cross section of an earth-rock dam, constructing an inter-well observation system, and acquiring data to obtain an observation seismic record; s2, establishing an initial physical property parameter model of the dam body without hidden danger, and taking the initial model as a current model; s3, forward modeling is carried out on the current model to obtain a simulated seismic record and a forward wave field; S4, constructing an objective function based on the observation seismic record and the simulated seismic record, and calculating a current objective function value; S5, judging whether the current objective function value is smaller than a preset threshold value, if so, outputting the current model as an inversion result, otherwise, entering a step S6; S6, performing wave field reverse pushing according to residual errors of the observation seismic records and the simulation seismic records to obtain a reverse pushing wave field, and calculating physical property parameter gradients of a current model by combining the forward wave field; S7, processing the physical property parameter gradient of the current model by adopting a conjugate gradient method to obtain a conjugate gradient; S8, constructing a preconditioning operator for suppressing the interference of the face wave and accelerating the update of relevant physical parameters of hidden danger in a deep area in the dam body, and correcting the conjugate gradient by using the preconditioning operator to obtain an update gradient; s9, determining an optimal inversion step length for updating the current model, and updating the current model according to the update gradient and the optimal inversion step length to obtain a new model; and S10, taking the new model as a new current model, and returning to the step S3 for iteration until the current objective function value in the step S5 is smaller than a preset threshold value, so as to obtain a physical property parameter model reflecting hidden danger of the earth-rock dam.
  2. 2. The method according to claim 1, wherein in step S1, the inter-well observation system uses upstream and downstream slopes of the dam body as natural excitation wells and receiving wells, and arranges excitation points and detection points along the surface of the dam body at preset intervals, and the seismic source adopts Ricker wavelets with preset main frequencies for collecting direct longitudinal waves and direct transverse waves penetrating the dam body, reflected waves and converted waves of bedrock interfaces, reflected waves of seepage areas, and multiple reflected waves generated by free boundaries of the slopes and the top of the dam.
  3. 3. The method according to claim 1, wherein in step S2, the geometry of the initial physical parameter model is consistent with the detected earth-rock dam.
  4. 4. The method according to claim 1, wherein in step S3, the forward modeling uses a staggered grid finite difference method for seismic wavefield modeling; in the differential calculation process, positive stress is set at integer grid points, and shear stress and particle vibration velocity components are set at half grid points; the method comprises the steps of setting the top surface of a dam body and upstream and downstream dam slopes as free boundaries, setting the free boundaries at half grid points, setting the areas above the free boundaries as vacuum, and setting the periphery of bedrock and the bottom interface of the dam body as absorption boundaries.
  5. 5. The method according to claim 1, wherein in step S4, the objective function is a normalized L2 norm based on waveform amplitude, which is defined as: Wherein u i,j represents the simulated seismic record of the j-th detector of the i-th gun, d i,j represents the observed seismic record of the j-th detector of the i-th gun, ns is the total number of guns, nr is the total number of detectors, E is the objective function value, and the normalized error between the observed seismic record and the simulated seismic record is represented.
  6. 6. The method according to claim 1, wherein in step S6, calculating the physical property parameter gradient of the current model specifically includes: performing wave field reverse pushing by using residual errors of the observation seismic records and the simulation seismic records as excitation sources to obtain a reverse pushed forward stress wave field and a shear stress wave field; performing cross-correlation calculation on the forward wave field and the inverse push wave field to obtain a gradient of a gradient Mei Canshu: Wherein λ, μ are the lame constant, τ xx 、τ zz 、τ xz is the forward simulated horizontal, vertical and shear wavefields, η xx 、η xx 、η xz is the reverse pushed horizontal, vertical and shear wavefields, g λ is the gradient of the lame constant λ, g μ is the gradient of the lame constant μ, sources represent all shots; Converting the gradient of the pull Mei Canshu into the gradient of longitudinal wave speed, transverse wave speed and density by utilizing a chain rule to obtain the physical property parameter gradient of the current model: wherein V p is longitudinal wave velocity, V s is transverse wave velocity, ρ is density, x 、 z The horizontal displacement field and the vertical displacement field of the reverse thrust are shown, v x 、v z is the horizontal velocity field and the vertical velocity field of the particle vibration, g Vp is the gradient of the longitudinal wave velocity, g Vs is the gradient of the transverse wave velocity, and g ρ is the gradient of the density.
  7. 7. The method according to claim 6, wherein in step S7, the conjugate gradient method is adopted to process the physical property parameter gradient of the current model to obtain the conjugate gradient, and the method specifically comprises: When the number of iterations n=1, the conjugate gradient δg 1 k =g 1 k , where k represents a different physical property parameter, when k=1, g 1 1 represents g Vp of the first iteration, when k=2, g 1 2 represents g Vs of the first iteration, and when k=3, g 1 3 represents g ρ of the first iteration; When n is more than or equal to 2, the calculation formula of the conjugate gradient delta g n k is as follows: wherein g n k is the physical property parameter gradient of the nth iteration, Is the physical property parameter gradient of the nth iteration and the 1 st iteration, delta g n k is the conjugate gradient of the nth iteration, For the conjugate gradient of the n-1 th iteration, beta n is the conjugate correction coefficient, delta is the conjugate identifier, Representing the L2 norm square.
  8. 8. The method according to claim 1, wherein in step S8, the preconditioner is a piecewise function, and the expression is: Wherein p is a preconditioner, Z is depth downward with a dam crest as an origin, Z gradt1 is a first depth threshold, Z gradt2 is a second depth threshold, H is a dam height, Z gradt1 ＜Z gradt2 < H, deltal is a depth range of a linear transition zone, deltal=Z gradt2 -Z gradt1 ,a、a 1 is a depth adjustment factor.
  9. 9. The method according to claim 8, wherein in step S9, determining an optimal inversion step for updating the current model comprises: And performing inversion step length search by adopting a parabolic fitting method to obtain the optimal inversion step length.
  10. 10. The method according to claim 9, wherein in step S9, a calculation formula for updating the current model according to the update gradient and the optimal inversion step size is: Wherein m n is the current model, m n+1 is the new model after updating, alpha n is the optimal inversion step length, p is the preconditioner, and δg n k is the conjugate gradient of the nth iteration.

Description

Earth-rock dam hidden danger full-wave inversion method based on inter-well observation system Technical Field The invention relates to the field of full waveform inversion of seismic waves, in particular to a full waveform inversion method of hidden danger of an earth-rock dam based on an inter-well observation system. Background The earth-rock dam is the most widely applied dam type in China, and accounts for more than 90% of the total number of dams in China, and the safety of the earth-rock dam is an important guarantee for normal operation of various hydraulic engineering. Compared with a concrete dam, the earth-rock dam filling material has the defects of easy erosion and weak anti-damage capability, and the early earth-rock dam design standard is low, construction technology and equipment are relatively backward, and effective management and maintenance measures are lacking in the long-term operation process, so that a large number of dam bodies have hidden troubles such as leakage, low-speed loose layers and the like after being operated for decades. If the detection and the treatment are not timely, the stability of the earth and rockfill dam structure can be obviously reduced, the ageing and the damage of the dam body are aggravated, the flood control and the irrigation of hydraulic engineering can be seriously influenced, and potential threats are formed to the life and property safety and the ecological environment of downstream people. The seismic wave method is used as a high-efficiency, nondestructive and low-cost geophysical detection method and is widely applied to detection of hidden danger of a dam. The method adopts a seismic wave method to detect hidden danger of the dam and evaluate safety of the dam body, and is characterized in that a longitudinal wave or transverse wave velocity profile in the dam body is obtained through analysis and processing of seismic data. Common processing methods for obtaining the dam speed profile include reflected wave speed analysis, surface wave inversion, full waveform inversion and the like. The full waveform inversion is more suitable for constructing the earth-rock dam speed model than other methods by virtue of comprehensive utilization of complete wave field information and high spatial resolution advantage thereof. For detecting hidden danger of earth and rockfill dams, a conventional seismic data acquisition mode is to arrange detectors along the longitudinal section direction of the dam body, as shown in fig. 1. Because the range of the hidden danger area inside the dam body is small, the physical property difference between the hidden danger area and the dam body is small, the energy of the reflected wave field of the hidden danger area is weak, the wave field information related to the hidden danger of the dam body in the seismic data acquired by adopting the traditional observation system is less, full waveform inversion is carried out by utilizing the seismic data, and the accuracy of identifying the hidden danger area inside the dam body is low. Compared with the traditional seismic exploration developed on the ground, the coverage area detected by the well observation system is wider, the acquired seismic data is richer, full waveform inversion is carried out by utilizing the data, and the inversion speed section is more accurate. However, drilling holes on two sides of a detection area is needed for developing the well-to-well seismic exploration, so that the detection cost is high, and the dam structure can be damaged. The nonlinear precondition conjugate gradient method becomes one of the inversion algorithms widely adopted at present due to high inversion precision and good stability, and one of the most important contents of the method in the full waveform inversion process of developing earth-rock dam hidden danger is to calculate the model update gradient. The gradient calculation is mainly divided into two steps, wherein the first step is to normally calculate the gradient value of the whole model area, and the second step is to construct a precondition operator to correct the gradient and highlight the gradient value of the key area so as to improve the inversion efficiency and accuracy. The preconditioning operator needs to be set according to the actual inversion situation and by combining the propagation characteristics of seismic waves in the dam body, and no preconditioning operator calculation scheme aiming at full waveform inversion of hidden danger of the earth and rockfill dam exists at present. In summary, the method for detecting hidden danger of earth and rockfill dam by adopting the seismic wave method has the defects of the traditional data acquisition mode and inversion method, and is not beneficial to the development of the high-precision and high-efficiency technique for detecting hidden danger of earth and rockfill dam. Disclosure of Invention The invention aims to provide a full waveform inversion method of hidd