CN102880786A

CN102880786A - Kriging ground settlement time domain monitoring method based on simulated annealing method

Info

Publication number: CN102880786A
Application number: CN2012102851697A
Authority: CN
Inventors: 岳建平; 甄宗杰; 董杰
Original assignee: Hohai University HHU
Current assignee: Hohai University HHU
Priority date: 2012-08-10
Filing date: 2012-08-10
Publication date: 2013-01-16

Abstract

The invention discloses a Kriging land subsidence time-domain monitoring method based on a simulated annealing method. The method includes: extracting monitoring information, performing normalization transformation, determining the dimension scale of equal-dimensional dynamic prediction, and then dividing time into groups, and calculating each For the variogram value corresponding to the time group, select the spherical model as the variogram model, and use the simulated annealing method to fit the variogram model, draw the variogram curve, then calculate the Kriging weight coefficient, calculate the estimated value of the Krining interpolation at the prediction time, and iteratively calculate , interpolate and refine the grid in the time domain, and generate the subsidence monitoring results at the quasi-monitoring time. The method of the present invention can carry out high-precision prediction according to the characteristics of the non-normal distribution of land subsidence data, effectively reflect the abnormal situation at the monitoring point, and can avoid the result of the linearization inversion method strongly depending on the initial model Select the case that causes the solution to fall into a local extremum.

Description

A time-domain monitoring method of Kriging land subsidence based on simulated annealing method

技术领域 technical field

本发明涉及地理信息领域的监控模型，具体地说涉及一种基于模拟退火法的Kringing地面沉降时域监控方法。 The invention relates to a monitoring model in the field of geographic information, in particular to a Kringing land subsidence time domain monitoring method based on a simulated annealing method. the

背景技术 Background technique

地面沉降是一种普遍而又日趋明显的地质现象。它是区域性地面高程下降的一种环境地质变化，大多发生在人口密集、工业发达的城市，其生成缓慢，持续时间长，危害范围广，破坏严重。城市地面沉降监测是保护人民生命财产、减少经济损失的重要手段，对监测资料的分析处理是判断一个城市环境地址优劣的科学依据。因此，在城市发展的同时，必须对其底面沉降状况进行监控，及时掌握第一手资料，并对观测资料进行科学的分析，及时发现可能存在的隐患，从而制定合理的防治措施，以确保城市的现代化进程。目前常用的监控模型有确定性模型、统计模型、人工智能模型等，其中大多数模型限制于自身的要求，仅适用于建立空间监控模型，而其他的时间监控模型又因无法顾及离散数据随机性和结构性特征，致使模型插值预测的结果一直差强人意，在实际工程中难以满足要求。 Land subsidence is a common and increasingly obvious geological phenomenon. It is a kind of environmental geological change caused by regional ground elevation drop, which mostly occurs in densely populated and industrially developed cities. Urban land subsidence monitoring is an important means to protect people's lives and properties and reduce economic losses. The analysis and processing of monitoring data is the scientific basis for judging the quality of an urban environment. Therefore, while the city is developing, it is necessary to monitor the subsidence of its bottom surface, grasp the first-hand information in time, and conduct scientific analysis on the observation data to discover possible hidden dangers in time, so as to formulate reasonable prevention and control measures to ensure that the urban modernization process. Currently commonly used monitoring models include deterministic models, statistical models, artificial intelligence models, etc. Most of these models are limited to their own requirements and are only suitable for establishing spatial monitoring models, while other time monitoring models cannot take into account the randomness of discrete data. And structural features, resulting in model interpolation prediction results have been unsatisfactory, difficult to meet the requirements in actual engineering. the

另一方面，在Kriging模型中，监测数据呈现非正态性会导致扰动信息逐渐积累，使变异函数无法真实地反映沉降数据的分布特征，传统的变异函数拟合模型是基于最小二乘准则的，其在处理数据时，会对测量误差有明显的放大作用。 On the other hand, in the Kriging model, the non-normality of the monitoring data will lead to the gradual accumulation of disturbance information, so that the variation function cannot truly reflect the distribution characteristics of the settlement data. The traditional variation function fitting model is based on the least squares criterion , which will significantly amplify the measurement error when processing data. the

发明内容 Contents of the invention

发明目的：本发明的目的是提供一种减少扰动信息积累、降低测量误差的一种基于模拟退火法的Kriging地面沉降时域监控方法。 Purpose of the invention: The purpose of the invention is to provide a Kriging land subsidence time-domain monitoring method based on the simulated annealing method that reduces disturbance information accumulation and measurement errors. the

技术方案：为了实现上述技术方案，本发明的基于模拟退火法的Kriging地面沉降时域监控方法，包括如下步骤： Technical scheme: In order to realize above-mentioned technical scheme, the Kriging land subsidence time-domain monitoring method based on simulated annealing method of the present invention, comprises the steps:

（1）提取监测时刻t_i，及沉降数据Z_i，进行规范化处理，其中，i＝1,2,…,N； (1) Extract the monitoring time t _i and the settlement data Z _i for normalization, where i=1,2,…,N;

（2）将沉降数据Z_i进行正态化转换； (2) Normalize the settlement data Z _i ;

（3）确定等维动态预测的维数尺度； (3) Determine the dimensional scale of the equal-dimensional dynamic prediction;

（4）划分时间分组，用{t′_m}表示： (4) Divide the time into groups, expressed by {t′ _m }:

${{{t t}_{m m}^{' '}}} = = m m \times \times \frac{Δ Δ {T T}^{' '}}{{N N}_{T T}},, m m = = 1,2 1,2,, . . . . . .,, {N N}_{T T}$

当各期采样的时间间隔比较均匀时， When the time interval of sampling in each period is relatively uniform,

{t′_m}＝m×T′，m＝1,2,...,N_T {t′ _m }=m×T′, m=1,2,...,N _T

式中，ΔT′为以确定维数尺度的时间间隔，T′为相邻两时间点的最短时间间隔，N_T表示时间分组的个数，N_T≥4； In the formula, ΔT' is the time interval with a certain dimension scale, T' is the shortest time interval between two adjacent time points, _NT represents the number of time groups, _NT ≥ 4;

（5）计算各时间组所对应的对数变异函数值： (5) Calculate the logarithmic variation function value corresponding to each time group:

$r r * * (({t t}_{m m}^{' '})) = = \frac{11}{22 N N (({t t}_{m m}^{' '}))} {Σ Σ}_{i i = = 11}^{N N (({t t}_{m m}^{' '}))} {[[In In (({Z Z}_{i i} (({x x}_{i i},, {y the y}_{i i})))) - - In In (({Z Z}_{i i} (({x x}_{i i} + + t t,, {y the y}_{i i} + + t t))))]]}^{22}$

式中，N(t′_m)表示时间间隔为t的所有监测时间间隔的个数，Z_i(x_i,y_i)为第i点(x_i,y_i)在某时刻的沉降监测值，Z_i(x_i+t,y_i+t)为第i点在经历t时间后的沉降监测值； In the formula, N(t′ _m ) represents the number of all monitoring time intervals with a time interval of t, Z _i ( _xi , y _i ) is the settlement monitoring value of point i ( _xi , y _i ) at a certain moment , Z _i ( _xi +t, y _i +t) is the settlement monitoring value of point i after t time;

（6）选择球状模型作为变异函数模型，并采用模拟退火法拟合变异函数模型，并绘制变异函数曲线； (6) Select the spherical model as the variogram model, and use the simulated annealing method to fit the variogram model, and draw the variogram curve;

（7）计算Kriging权系数： (7) Calculate the Kriging weight coefficient:

λ＝K^-1M， λ=K ^- 1M,

其中， $K = [\begin{matrix} γ_{11} & γ_{12} & \cdot \cdot \cdot & γ_{1 n} & 1 \\ γ_{21} & γ_{22} & \cdot \cdot \cdot & γ_{21} & 1 \\ \cdot \cdot \cdot & \cdot \cdot \cdot & \cdot \cdot \cdot & \cdot \cdot \cdot & \cdot \cdot \cdot \\ γ_{n 1} & γ_{n 1} & \cdot \cdot \cdot & γ_{nn} & 1 \\ 1 & 1 & 1 & 1 & 0 \end{matrix}],$ $M = [\begin{matrix} {\overset{&OverBar;}{γ}}_{1 p} \\ {\overset{&OverBar;}{γ}}_{2 p} \\ \cdot \cdot \cdot \\ {\overset{&OverBar;}{γ}}_{np} \\ 1 \end{matrix}];$ in, $K = [\begin{matrix} γ_{11} & γ_{12} & &Center Dot; &Center Dot; &Center Dot; & γ_{1 no} & 1 \\ γ_{twenty one} & γ_{twenty two} & &Center Dot; \cdot &Center Dot; & γ_{twenty one} & 1 \\ &Center Dot; &Center Dot; &Center Dot; & &Center Dot; &Center Dot; \cdot & \cdot \cdot \cdot & \cdot \cdot &Center Dot; & \cdot &Center Dot; &Center Dot; \\ γ_{no 1} & γ_{no 1} & &Center Dot; &Center Dot; \cdot & γ_{n} & 1 \\ 1 & 1 & 1 & 1 & 0 \end{matrix}],$ $m = [\begin{matrix} {\overset{&OverBar;}{γ}}_{1 p} \\ {\overset{&OverBar;}{γ}}_{2 p} \\ &Center Dot; &Center Dot; &Center Dot; \\ {\overset{&OverBar;}{γ}}_{np} \\ 1 \end{matrix}];$

（8）计算预测时刻Krining插值的估计值： (8) Calculate the estimated value of Krining interpolation at the time of prediction:

$Z Z * * (({X x}_{00})) = = {&Sum; &Sum;}_{i i = = 11}^{n no} {λ λ}_{i i} Z Z (({X x}_{i i}))$

（9）重复步骤（4）至步骤（8），在时间域上内插和加密网格，生成准监测时刻的沉降监测结果。 (9) Repeat steps (4) to (8), interpolate and refine the grid in the time domain, and generate settlement monitoring results at the quasi-monitoring time. the

在Kriging时域监控模型中，监测数据呈现出明显的非正态性，本发明先使用多项式拟合参考时刻的实测值作为其真值的近似值，并将其与实测值作差，然后对差值作BOX-COX转换，最终形成符合正态分布的数据序列。所述步骤（2）中正态化的具体步骤为： In the Kriging time-domain monitoring model, the monitoring data presents obvious non-normality. The present invention first uses the measured value of the polynomial fitting reference moment as the approximate value of its true value, and makes a difference between it and the measured value, and then compares the difference The values are converted by BOX-COX, and finally form a data sequence conforming to the normal distribution. The specific steps of normalization in the step (2) are:

（201）使用多项式拟合参考时刻的实测值作为其真值的近似值； (201) Use the polynomial to fit the measured value at the reference moment as an approximation of its true value;

（202）将拟合值与实测值作差； (202) Make a difference between the fitted value and the measured value;

（203）对差值作BOX-COX转换，最终形成符合正态分布的数据序列。 (203) Perform BOX-COX conversion on the difference, and finally form a data sequence conforming to the normal distribution. the

其中，BOX-COX转换的具体方法为： Among them, the specific method of BOX-COX conversion is:

${y the y}^{r r} = = \{\begin{matrix} \frac{{y the y}^{r r - - 11}}{λ λ},, r r &NotEqual; &NotEqual; 00 \\ Iny Iny,, r r = = 00 \end{matrix}$

式中，r是可变参数，由原始数据估计得出；y为一个变换族。 In the formula, r is a variable parameter estimated from the original data; y is a transformation family. the

本发明采用等维动态预测的模式，保证时间相关性，减少扰动信息累积。即不断地更新监测数据，删除旧有的数据，同时也要保证数据的维数尺度不变。维数尺度的确定根据实际情况而定，所述步骤（3）中维数尺度为8~15。 The invention adopts an equal-dimensional dynamic prediction mode to ensure time correlation and reduce disturbance information accumulation. That is to continuously update the monitoring data, delete the old data, and at the same time ensure that the dimension scale of the data remains unchanged. The dimension scale is determined according to the actual situation, and the dimension scale in the step (3) is 8-15. the

本发明选取地面沉降量作为区域化变量，其具有生长缓慢、持续时间长、影响范围广、成因机制复杂等特点，同时也易受局部建筑施工、临时地下水抽采、突发地质破坏等不确定因素的影响，使得监测结果容易出现异常，为减少上述因素的影响，提出对区域化变量取自然对数的措施。因此，作为本发明的进一步优化，所述步骤（5）中计算各时间组所对应的变异函数值取自然对数： The present invention selects land subsidence as the regionalized variable, which has the characteristics of slow growth, long duration, wide range of influence, and complex genetic mechanism, and is also susceptible to uncertainties such as local construction, temporary groundwater extraction, and sudden geological damage. The influence of these factors makes the monitoring results prone to abnormalities. In order to reduce the influence of the above factors, a measure of taking the natural logarithm of the regionalized variables is proposed. Therefore, as a further optimization of the present invention, in the step (5), the value of the variation function corresponding to each time group is calculated to take the natural logarithm:

本发明取代传统的最小二乘准则，利用模拟退火算法拟合球状函数模型。所述步骤（6）中球状模型变异函数为： The invention replaces the traditional least square criterion and uses a simulated annealing algorithm to fit a spherical function model. In the step (6), the spherical model variation function is:

$r r ((t t)) = = \{\begin{matrix} 00 & t t = = 00 \\ {C C}_{00} + + C C ((\frac{33}{22} \cdot \cdot \frac{t t}{a a} - - \frac{11}{22} \cdot \cdot \frac{{t t}^{33}}{{a a}^{33}})) & 00 < < t t \leq \leq a a \\ {C C}_{00} + + C C & t t > > a a \end{matrix},,$

其中，C₀为块金值，C为偏基台值，a表示变程。 Among them, C ₀ is the nugget value, C is the partial sill value, and a represents the range.

其模拟退火法包括如下步骤： Its simulated annealing method comprises the following steps:

（601）初始化变异函数模型参数C₀、C、a，确定参数变化范围，计算目标函数值E(C₀,C,a)； (601) Initialize the variation function model parameters C ₀ , C, and a, determine the variation range of the parameters, and calculate the objective function value E(C ₀ ,C,a);

（602）给当前模型C₀、C、a进行扰动产生一个新模型参数C₀′、C′、a′，计算目标函数值E′(C₀′,C′,a′)； (602) Perturb the current model C ₀ , C, and a to generate a new model parameter C ₀ ′, C′, a’, and calculate the objective function value E′(C ₀ ′,C′,a′);

模型扰动采用Ingber(1989)提出的依赖于温度的似Cauchy分布产生新模型，具体形式如下： The model perturbation uses the temperature-dependent Cauchy-like distribution proposed by Ingber (1989) to generate a new model, the specific form is as follows:

m_i′＝m_i+y_i(B_i-A_i)， m _i '=m _i +y _i (B _i -A _i ),

yi＝Tsgn(μ-0.5)[(1+1/T)^|2μ-1|-1]， yi=Tsgn(μ-0.5)[(1+1/T) ^|2μ-1| -1],

式中，m_i为当前模型中的第i个变量；μ为[0,1]均匀分布的随机数；[A_i,B_i]为m_i的取值范围，且要求扰动后的m_i∈[A_i,B_i]；sgn为符号函数； In the formula, m _i is the i-th variable in the current model; μ is a random number uniformly distributed in [0,1]; [A _i , B _i ] is the value range of m _i , and it is required that the disturbed m _i ∈[A _i , B _i ]; sgn is a sign function;

（603）判断ΔE＝E′(C₀′,C′,a′)-E(C₀,C,a)是否小于0，若小于0，则新模型参数C₀′、C′、a′被接收，否则按概率P进行接收； (603) Judging whether ΔE=E′(C ₀ ′,C′,a′)-E(C ₀ ,C,a) is less than 0, if it is less than 0, the new model parameters C ₀ ′, C′, a′ is received, otherwise it is received with probability P;

$P P = = {[[11 - - \frac{((11 - - h h)) ΔE ΔE}{T T}]]}^{\frac{11}{11 - - h h}}$

其中，T表示温度，h为某一不为1的常数。 Among them, T represents the temperature, and h is a constant not equal to 1. the

（604）当模型参数被接收时，置C₀=C₀′、C=C′、a=a′； (604) When the model parameters are received, set C ₀ =C ₀ ′, C=C′, a=a′;

（605）缓慢降低温度T，根据退火机制进行迭代，输出变异模型参数最优值；其中，退火机制为： (605) Slowly lower the temperature T, iterate according to the annealing mechanism, and output the optimal value of the variation model parameters; where the annealing mechanism is:

$T T ((k k)) = = {T T}_{00} {a a}^{{k k}^{11 / / N N}}$

式中，T₀为初始温度；k为迭代次数；N为给定常数，一般为1或2；a通常选择0.7≤a≤1。 In the formula, T ₀ is the initial temperature; k is the number of iterations; N is a given constant, usually 1 or 2; a is usually chosen to be 0.7≤a≤1.

有益效果：本发明的基于模拟退火法的Kriging地面沉降时域监控方法，可针对地面沉降数据非正态分布的特点进行高精度的预测，有效针对监测点出现的异常情况进行真实的反映，采用等维动态预测的模式，保证时间相关性，减少扰动信息累积；防止误差放大。模拟退火法不用求目标函数的偏导数及解大型矩阵方程组，即能找到一个全局最优解，而且易于加入约束条件，并可避免线性化反演方法结果强烈依赖于初始模型的选取导致解落入局部极值的情况。 Beneficial effects: the Kriging land subsidence time-domain monitoring method based on the simulated annealing method of the present invention can make high-precision predictions for the characteristics of non-normal distribution of land subsidence data, and effectively reflect the abnormal conditions at the monitoring points. The equal-dimensional dynamic prediction mode ensures time correlation, reduces the accumulation of disturbance information, and prevents error amplification. The simulated annealing method can find a global optimal solution without seeking the partial derivatives of the objective function and solving large matrix equations, and it is easy to add constraints, and it can avoid the linearization inversion method. falls into a local extremum. the

附图说明 Description of drawings

图1是本发明的总流程图。 Fig. 1 is the general flowchart of the present invention. the

具体实施方式Detailed ways

下面结合附图对本发明作更进一步的说明。 The present invention will be further described below in conjunction with the accompanying drawings. the

请参考图1，本发明的基于模拟退火法的Kriging地面沉降时域监控方法，包括如下步骤： Please refer to Fig. 1, the Kriging land subsidence time-domain monitoring method based on simulated annealing method of the present invention, comprises the steps:

（1）提取监测时刻t_i，及沉降数据Z_i，进行规范化处理，其中，i=1,2,…,N； (1) Extract monitoring time t _i and settlement data Z _i for normalization, where i=1,2,…,N;

（2）将沉降数据Z_i进行正态化转换，具体步骤为： (2) Normalize the settlement data Z _i , the specific steps are as follows:

（3）确定等维动态预测的维数尺度，维数尺度为8~15； (3) Determine the dimension scale of equal-dimensional dynamic prediction, and the dimension scale is 8~15;

{t′_m}＝m×T′，m＝1,2,...,N_T {t′ _m }=m×T′, m=1,2,...,N _T

（5）计算各时间组所对应的变异函数值: (5) Calculate the variation function value corresponding to each time group:

$r r * * (({t t}_{m m}^{' '})) = = \frac{11}{22 N N (({t t}_{m m}^{' '}))} {Σ Σ}_{i i = = 11}^{N N (({t t}_{m m}^{' '}))} {[[{Z Z}_{i i} (({x x}_{i i},, {y the y}_{i i})) - - {Z Z}_{i i} (({x x}_{i i} + + t t,, {y the y}_{i i} + + t t))]]}^{22}$

作为优选方案，通过取自然对数可减少监测结果异常带来的误差放大： As an optimal solution, the error amplification caused by abnormal monitoring results can be reduced by taking the natural logarithm:

$r r * * (({t t}_{m m}^{' '})) = = \frac{11}{22 N N (({t t}_{m m}^{' '}))} {Σ Σ}_{i i = = 11}^{N N (({t t}_{m m}^{' '}))} {[[In In (({Z Z}_{i i} (({x x}_{i i},, {y the y}_{i i})))) - - In In (({Z Z}_{i i} (({x x}_{i i} + + t t,, {y the y}_{i i} + + t t))))]]}^{22};;$

（6）本发明取代传统的最小二乘准则，利用模拟退火算法拟合球状函数模型，并绘制变异函数曲线； (6) The present invention replaces the traditional least square criterion, uses the simulated annealing algorithm to fit the spherical function model, and draws the variation function curve;

球状模型变异函数为： The spherical model variation function is:

$r r ((t t)) = = \{\begin{matrix} 00 & t t = = 00 \\ {C C}_{00} + + C C ((\frac{33}{22} \cdot &Center Dot; \frac{t t}{a a} - - \frac{11}{22} \cdot \cdot \frac{{t t}^{33}}{{a a}^{33}})) & 00 < < t t \leq \leq a a \\ {C C}_{00} + + C C & t t > > a a \end{matrix},,$

模拟退火法包括如下步骤： The simulated annealing method includes the following steps:

m_i′＝m_i+y_i(B_i-A_i)， m _i '=m _i +y _i (B _i -A _i ),

yi＝Tsgn(μ-0.5)[(1+1/T)^|2μ-1|-1]， yi=Tsgn(μ-0.5)[(1+1/T) ^|2μ-1| -1],

（603）判断ΔE＝E′(C₀′,C′,a′)-E(C₀,C,a)是否小于0，若小于0，则新模型参数C₀′、C′、 a′被接收，否则按概率P进行接收； (603) Judging whether ΔE=E′(C ₀ ′,C′,a′)-E(C ₀ ,C,a) is less than 0, if it is less than 0, the new model parameters C ₀ ′, C′, a′ is received, otherwise it is received with probability P;

（605）缓慢降低温度T，根据退火机制进行迭代，输出变异模型参数最优值； (605) Slowly lower the temperature T, iterate according to the annealing mechanism, and output the optimal value of the variation model parameters;

其中，退火机制为： Among them, the annealing mechanism is:

$T T ((k k)) = = {T T}_{00} {a a}^{{k k}^{11 / / N N}}$

（7）计算Kriging权系数： (7) Calculate the Kriging weight coefficient:

λ＝K^-1M， λ=K ^- 1M,

其中， $K = [\begin{matrix} γ_{11} & γ_{12} & \cdot \cdot \cdot & γ_{1 n} & 1 \\ γ_{21} & γ_{22} & \cdot \cdot \cdot & γ_{21} & 1 \\ \cdot \cdot \cdot & \cdot \cdot \cdot & \cdot \cdot \cdot & \cdot \cdot \cdot & \cdot \cdot \cdot \\ γ_{n 1} & γ_{n 1} & \cdot \cdot \cdot & γ_{nn} & 1 \\ 1 & 1 & 1 & 1 & 0 \end{matrix}],$ $M = [\begin{matrix} {\overset{&OverBar;}{γ}}_{1 p} \\ {\overset{&OverBar;}{γ}}_{2 p} \\ \cdot \cdot \cdot \\ {\overset{&OverBar;}{γ}}_{np} \\ 1 \end{matrix}];$ in, $K = [\begin{matrix} γ_{11} & γ_{12} & \cdot &Center Dot; &Center Dot; & γ_{1 no} & 1 \\ γ_{twenty one} & γ_{twenty two} & &Center Dot; &Center Dot; \cdot & γ_{twenty one} & 1 \\ &Center Dot; &Center Dot; \cdot & &Center Dot; \cdot \cdot & \cdot \cdot \cdot & \cdot \cdot &Center Dot; & \cdot &Center Dot; &Center Dot; \\ γ_{no 1} & γ_{no 1} & &Center Dot; \cdot &Center Dot; & γ_{n} & 1 \\ 1 & 1 & 1 & 1 & 0 \end{matrix}],$ $m = [\begin{matrix} {\overset{&OverBar;}{γ}}_{1 p} \\ {\overset{&OverBar;}{γ}}_{2 p} \\ \cdot \cdot \cdot \\ {\overset{&OverBar;}{γ}}_{np} \\ 1 \end{matrix}];$

（8）计算预测时刻Kriging插值的估计值： (8) Calculate the estimated value of Kriging interpolation at the time of prediction:

（9）重复步骤（1）至步骤（8），在时间域上内插和加密网格，生成准监测时刻的沉降监测结果。 (9) Repeat steps (1) to (8), interpolate and refine the grid in the time domain, and generate settlement monitoring results at the quasi-monitoring time. the

以上所述仅是本发明的优选实施方式，应当指出：对于本技术领域的普通技术人员来说，在不脱离本发明原理的前提下，还可以做出若干改进和润饰，这些改进和润饰也应视为本发明的保护范围。 The above is only a preferred embodiment of the present invention, it should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, some improvements and modifications can also be made. It should be regarded as the protection scope of the present invention. the

Claims

1. a Kriging land subsidence time-domain monitoring method based on simulated annealing method, is characterized in that: the method may further comprise the steps:

(1) Extract the monitoring time t _i and settlement monitoring data Z _i for normalization, where i=1,2,…,N;

(2) Normalize the settlement data Z _i ;

(3) Determine the dimensional scale of the equal-dimensional dynamic prediction;

(4) Divide the time into groups, expressed by {t′ _m }:

When the time interval of sampling in each period is relatively uniform,

{t′ _m }=m×T′, m=1,2,...,N _T ,

In the formula, ΔT' is the time interval with a certain dimension scale, T' is the shortest time interval between two adjacent time points, _NT represents the number of time groups, _NT ≥ 4;

(5) Calculate the variation function value corresponding to each time group:

In the formula, N(t′ _m ) represents the number of all monitoring time intervals with a time interval of t, Z _i ( _xi , y _i ) is the settlement monitoring value of point i ( _xi , y _i ) at a certain moment , Z _i ( _xi +t, y _i +t) is the settlement monitoring value of point i after t time;

(6) Select the spherical model as the variogram model, and use the simulated annealing method to fit the variogram model, and draw the variogram curve;

(7) Calculate the Kriging weight coefficient:

λ=K ^- 1M,

in,

(8) Calculate the estimated value of Krining interpolation at the prediction time:

(9) Repeat steps (4) to (8), interpolate and refine the grid in the time domain, and generate settlement monitoring results at the quasi-monitoring time. the

2. a kind of Kriging land subsidence time-domain monitoring method based on simulated annealing method according to claim 1, is characterized in that:

The specific steps of normalization in the step (2) are:

(201) Use the polynomial to fit the measured value at the reference moment as an approximation of its true value;

(202) Make a difference between the fitted value and the measured value;

(203) Perform BOX-COX conversion on the difference, and finally form a data sequence conforming to the normal distribution.

3. A Kriging land subsidence time-domain monitoring method based on simulated annealing method according to claim 1, characterized in that: the dimension scale in the step (3) is 8-15.

4. A kind of Kriging land subsidence time-domain monitoring method based on simulated annealing method according to claim 1, is characterized in that: in the described step (5), calculate the variation function value corresponding to each time group to take natural logarithm:

In the formula, N(t′ _m ) represents the number of all monitoring time intervals with a time interval of t, Z _i ( _xi , y _i ) is the settlement monitoring value of point i ( _xi , y _i ) at a certain moment , Z _i ( _xi +t, y _i +t) is the settlement monitoring value of point i after t time.

5. A kind of Kriging land subsidence time-domain monitoring method based on simulated annealing method according to claim 1, is characterized in that: in the described step (6), the spherical model variation function is:

Among them, C ₀ is the nugget value, C is the partial sill value, and a represents the range.

6. A kind of Kriging land subsidence time domain monitoring method based on simulated annealing method according to claim 1 or 5, it is characterized in that: in described step (6), simulated annealing method comprises the following steps:

(601) Initialize the variation function model parameters C ₀ , C, and a, determine the variation range of the parameters, and calculate the objective function value E(C ₀ ,C,a);

(602) Disturb the current model C ₀ , C, a to generate a new model parameter C ₀ ′, C’, a’, and calculate the objective function value E′(C ₀ ′, C′, a′), the disturbance The model is:

m _i '=m _i +y _i (B _i -A _i ),

yi=Tsgn(μ-0.5)[(1+1/T) ^|2μ-1| -1],

In the formula, m _i is the i-th variable in the current model; μ is a random number uniformly distributed in [0,1]; [A _i , B _i ] is the value range of m _i , and it is required that the disturbed m _i ∈[A _i , B _i ]; sgn is a sign function;

(603) Judging whether ΔE=E′(C ₀ ′,C′,a′)-E(C ₀ ,C,a) is less than 0, if it is less than 0, the new model parameters C ₀ ′, C′, a′ is received, otherwise it is received with probability P;

Among them, T represents the temperature, h is a constant not equal to 1;

(604) When the model parameters are received, set C ₀ =C ₀ ′, C=C′, a=a′;

(605) Slowly lower the temperature T, iterate according to the annealing mechanism, and output the optimal value of the variation model parameters;

Among them, the annealing mechanism is:

In the formula, T ₀ is the initial temperature; k is the number of iterations; N is a given constant; a is usually chosen to be 0.7≤a≤1.