Monitoring strategy optimization method based on observation information value analysis

Abstract

The invention discloses a monitoring strategy optimization method based on observation information value analysis, and belongs to the field of monitoring strategy optimization, and the method comprises the steps: employing the observation information value to quantify the importance of observation data, taking the observation information quality as the guidance to optimize the monitoring position, the monitoring time length and the monitoring frequency with the highest value, and carrying out the optimization of the monitoring position, the monitoring time length and the monitoring frequency; the optimization of the monitoring network in two-dimensional and three-dimensional spaces can be realized, and the operation efficiency of the monitoring system is scientifically and effectively improved.

Description

Technical Field

[0001] The present invention belongs to the field of monitoring strategy optimization, and particularly relates to a method for optimizing a monitoring strategy based on the analysis of the value of observation information. Background Art

[0002] Monitoring is essentially a means of information collection and analysis, and the quantity and quality of information are the key indicators for measuring the effectiveness of monitoring behavior. The importance of information quantity and quality is determined by the core objective of the monitoring behavior. When the monitoring objective is the observation information itself, the weight of the quantity of observation information is higher. For example, the water level, water quality, and air pollutant content of groundwater, etc. In view of this, traditional monitoring optimization methods all focus on issues such as the quantity of the obtained observation information, the repeatability and redundancy of the observation information, and they take obtaining the most information as the core idea of optimization.

[0003] At present, there is still a lack of a scientific guiding method for the design of monitoring duration and monitoring frequency, resulting in problems such as high monitoring costs, low monitoring efficiency, and easy distortion of monitoring results to varying degrees in some monitoring networks. Therefore, there is an urgent need for a scientific and systematic method to guide the selection of monitoring points, monitoring duration, and monitoring frequency, while reasonably controlling costs, collecting high-quality observation information as accurately as possible, and effectively improving the overall efficiency of monitoring work. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides a method for optimizing a monitoring strategy based on the analysis of the value of observation information, including:

[0005] Constructing a numerical model describing the system behavior based on the research objective, and setting alternative monitoring points in the research area based on the numerical model;

[0006] Constructing the parameters to be inverted in the research, and setting the initial value and confidence interval of the parameters to be inverted;

[0007] Inputting the parameters to be inverted into the numerical model, obtaining the simulated observation information in the alternative monitoring points, and generating a calibration data set based on the simulated observation information;

[0008] Performing parameter inversion on the calibration data set to obtain a posterior covariance matrix of the parameters;

[0009] Calculating the value of observation information of all the alternative monitoring points based on the posterior covariance matrix of the parameters, and optimizing the monitoring network according to the ranking of the value of observation information;

[0010] Optimizing the monitoring duration and monitoring frequency based on the optimized monitoring network to obtain the optimal monitoring duration and the optimal monitoring frequency.

[0011] Preferably, the process of constructing a numerical model to describe the system behavior based on the research objective includes:

[0012] Collect data required for the numerical model according to the monitoring objective and the nature of the observation information;

[0013] Based on the data, construct a physical model describing the environment or medium state where the observation information is located and a mathematical model reflecting the relationship between the observation information and the target parameters.

[0014] Preferably, the process of constructing the parameters to be inverted in the research and clarifying the initial values and confidence intervals of the inversion parameters includes: formulating the parameters to be inverted, setting the initial values of the parameters to be inverted according to the reliable data, expert experience and literature results in the historical research of the target parameters, and giving the prior interval of the parameters to be inverted.

[0015] Preferably, the process of generating a calibration data set based on the simulated observation information includes:

[0016] Based on historical data, formulate a set of approximate values of the parameters to be inverted as the standard values;

[0017] Input the standard values into the numerical model to obtain the simulated observation information of each alternative monitoring point through numerical simulation;

[0018] Process the simulated observation information of each alternative monitoring point to obtain the calibration data set.

[0019] Preferably, the process of performing parameter inversion on the calibration data set to obtain the prior and posterior probability distributions of the parameters includes:

[0020] Input the initial values and confidence intervals of the parameters to be inverted into the numerical model to obtain the parameter covariance matrix before any calibration;

[0021] Take the observation information of all alternative monitoring points as the data set, perform iterative calculations on each alternative monitoring point among all alternative monitoring points, and output until the preset conditions are met to obtain the posterior covariance matrix of the parameters.

[0022] Preferably, the process of optimizing the monitoring network according to the ranking of the value of the observation information includes:

[0023] Calculate the change in the variance of the target parameters when the posterior covariance matrix of the parameters is incorporated into different calibration data sets;

[0024] Based on the change in the variance of the target parameters for different calibration data sets, calculate the cumulative change in variance of all parameters;

[0025] Use the cumulative change in variance of all parameters to optimize the monitoring network according to the ranking of the value of the observation information.

[0026] Preferably, the process of obtaining the optimal monitoring duration includes:

[0027] Obtain the observation information of all preferred monitoring points in the optimized monitoring network, perform discretization processing on the observation information for the monitoring duration, and obtain a monitoring duration data set;

[0028] Calculate the posterior probability distribution of parameters corresponding to the monitoring duration data set with different durations, and quantify the value of time data based on the posterior probability distribution of parameters;

[0029] Integrate the value of the time data with the monitoring time cost to obtain the optimal monitoring duration.

[0030] Preferably, the process of obtaining the optimal monitoring frequency includes:

[0031] Obtain the observation information of all preferred monitoring points in the optimized monitoring network, adjust the monitoring frequency, perform discretization grouping on the observation information, and obtain a monitoring frequency data set;

[0032] Calculate the posterior probability distribution of parameters corresponding to the monitoring frequency data set at different frequencies, and quantify the value of frequency data based on the posterior probability distribution of parameters;

[0033] Obtain the optimal monitoring frequency based on the changing trend of the quantified frequency data value with the monitoring frequency.

[0034] On the other hand, the present invention also provides an electronic device, including a memory, a processor, and a computing program stored in the memory and executable on the processor. When the processor executes the computing program, the method is implemented.

[0035] On the other hand, the present invention also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method is implemented.

[0036] Compared with the prior art, the present invention has the following advantages and technical effects:

[0037] The theory of the value of observation information proposed by the present invention can quantify the contribution of observation information to the monitoring target, make up for the potential defect of traditional monitoring network optimization methods that emphasize the quantity of information while ignoring the quality of information, can identify the monitoring positions that take into account both the monitoring cost and the achievement of the monitoring target, and significantly reduce the monitoring cost while improving the monitoring efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The drawings constituting a part of this application are used to provide a further understanding of this application. The schematic embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation to this application. In the drawings:

[0039] Figure 1 Schematic diagram of the method flow of the embodiment of the present invention;

[0040] Figure 2 Schematic diagram of the value and ranking of the observation information of the alternative monitoring stations in the embodiment of the present invention;

[0041] Figure 3 Schematic diagram of the layout of the alternative monitoring stations in the heterogeneous geological model in the embodiment of the present invention in the study area and the optimization results of the alternative monitoring stations within the vertical depth range of 40 - 50 m;

[0042] Figure 4 Schematic diagram of the change of the observation information value of each alternative monitoring station with the monitoring frequency in the embodiment of the present invention. Detailed implementation manners

[0043] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The present application will be described in detail below with reference to the drawings and in combination with the embodiments.

[0044] It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0045] Embodiment 1

[0046] As Figure 1 shown, in this embodiment, a monitoring strategy optimization method based on the analysis of the value of observation information is provided, including:

[0047] S1: Based on the research objective, combined with the mathematical and physical backgrounds of the monitoring object, collect relevant data and establish a numerical model describing the behavior of the system;

[0048] According to the monitoring objective and the nature of the observation information, collect the data required for the numerical model, establish a physical model that can describe the environment or medium state where the observation information is located, and a mathematical model that can reflect the relationship between the observation information and the target parameters.

[0049] For example, when inferring oil and gas reserves and distributions using seismic data collected in oil and gas exploration, geological data (such as well data, stratification data, etc.), seismic acquisition data (such as the amplitude and phase of seismic waves, etc.) need to be collected, and a geological reservoir model, a rock physics model, and a seismic wave propagation model are established. During the inversion process, the geological reservoir model is used as a constraint condition to describe the geometric shape, spatial distribution, and physical properties of the oil and gas reservoir, ensuring that the inversion results conform to the actual geological and geophysical situations. The rock physics model is used to describe the relationship between the elastic parameters of rocks (such as velocity, density) and their physical properties (such as porosity, saturation, permeability). The seismic wave propagation model is used to describe the propagation characteristics of seismic waves in rocks. When inversely identifying the location of potential pollution sources using the concentration data of pollutants in groundwater, geological exploration data (such as borehole data, geophysical exploration data, etc.), hydrogeological data (such as permeability coefficient, etc.) need to be collected, and a heterogeneous geological model, a groundwater flow model, and a solute transport model are established. The heterogeneous geological model is used to describe the locations of aquifers, aquitards, pervious layers, etc., and the characteristics of the heterogeneous distribution of each stratum. The groundwater flow model is usually based on Darcy's law and the continuity equation and is used to describe the heterogeneity, anisotropy of the aquifer, as well as information such as the distribution, flow velocity, and flow direction of the groundwater level. The solute transport model is usually based on the convection-dispersion equation, combined with the results of the groundwater flow model (such as flow velocity, flow direction, etc.), and parameters such as the initial concentration and boundary conditions of the pollutant to describe information such as the concentration distribution and migration path of the pollutant in groundwater.

[0050] S2: Based on the numerical model described in S1, set alternative monitoring points;

[0051] The positions of the alternative monitoring points can be flexibly selected according to the optimization objective and are not restricted by the spatial scale. If the optimization objective is to streamline and optimize the existing monitoring network, the alternative monitoring points are all the existing monitoring points in the monitoring network; if the optimization objective is to plan the positions of new monitoring points, the number and positions of the alternative monitoring points can be freely set according to requirements. The optimized monitoring positions will be selected from the alternative monitoring points, and the number of alternative monitoring points will directly affect the calculation cost of subsequent optimization.

[0052] S3: Formulate the parameters to be inverted, and clarify the initial values and confidence intervals of the inversion parameters;

[0053] S3.1: Determine the parameters to be inverted in the study;

[0054] The parameters to be inverted refer to the unknown parameters that can be solved using mathematical methods through the relationship between the observed information and the constructed model in different monitoring systems.

[0055] The selection of parameters should be based on the characteristics of specific fields and application scenarios, comprehensively considering factors such as the physical meaning of parameters, data sensitivity, stability, ease of observation and measurement, representativeness, practical application value, and ease of inversion and verification. It is recommended to select the core parameters of the experiment or research or the parameters that have a greater impact on the results as the target parameters for inversion. The correct selection of inversion parameters is helpful for achieving the monitoring target.

[0056] S3.2: Preset the initial values and prior intervals of the inversion parameters;

[0057] Set the initial values of the parameters according to the reliable data, expert experience, or literature results in the historical research of the target parameters, and give the prior intervals.

[0058] The setting of the initial values and prior intervals has an important impact on the accuracy and efficiency of the inversion results. The initial value is the starting point of the iterative calculation of parameter inversion, and it also directly affects the accuracy of parameter inversion. Setting a reasonable initial value is beneficial to shortening the iterative process and making the results converge to the optimal solution faster. The prior interval provides constraint conditions for parameter inversion to ensure that the parameters fluctuate within a reasonable range. Setting a reasonable prior interval can narrow the calculation range and reduce the amount of calculation. Especially when dealing with complex problems or a large amount of observational information, setting reasonable initial values and prior intervals is particularly important.

[0059] S4: Determine the "standard value" of the parameter, and conduct simulations based on the numerical model in S1 to obtain the calibration data set;

[0060] S4.1: According to materials such as measured information, reference documents, and experimental data, draw up a set of approximate values for the parameters to be inverted in S3.1 and assume them as the "standard values" without errors.

[0061] The so-called "standard value" is an artificially determined approximate value of the parameter, not the true value of the parameter. It reflects the researcher's estimation of the parameter to be inverted and will be used to check the accuracy of the subsequent parameter inversion results. For different monitoring systems, there are different methods to determine the "standard value", which can be obtained through means such as experiments, measurements, and data retrieval. For example, in materials science, the standard values of the mechanical parameters of materials can be obtained through tensile experiments, compression experiments, etc.; in groundwater science research, the hydrogeological conditions can be obtained through groundwater dynamic observation data or field exploration materials, so as to determine the measured values of hydrogeological parameters. If there are no experimental or observational conditions, it can be set by referring to previous research or relevant standards.

[0062] S4.2: Incorporate the "standard values" of the parameters in S4.1 into the numerical model described in S1. Obtain the observation information of each alternative monitoring point through numerical simulation and regard it as the calibration dataset. To ensure the accuracy of subsequent monitoring network optimization, the duration of numerical simulation and the setting of monitoring frequency can be set relatively large to ensure sufficient observation information is obtained.

[0063] S5: Perform parameter inversion based on the calibration dataset in S4 to obtain the prior and posterior probability distributions of the parameters;

[0064] Assume that there is a linear correlation between the parameters and the output of the numerical model. Then the model output p is expressed as:

[0065] p = Xs + ε

[0066] In the formula, X represents the sensitivity matrix, s is an N×1 vector containing all the parameters inverted by the numerical model, and ε represents the measurement noise.

[0067] Based on the Bayesian formula for linear prediction uncertainty, define the posterior covariance matrix U′(s) of the parameters:

[0068] U′(s) = U(s) - U(s)X t [XU(s)X t + U(ε)] -1 XU(s)

[0069] In the formula, U′(s) represents the posterior covariance matrix of the parameters, U(s) represents the prior covariance matrix of the parameters, and the second term on the right represents the reduction of parameter uncertainty after introducing the calibration dataset.

[0070] S5.1: Introduce the information such as the parameters to be inverted, initial values, and prior intervals selected in S3.2 into the numerical model described in S1, and calculate the parameter covariance matrix U(s) before any calibration.

[0071] S5.2: Use the observation information of a single alternative monitoring point as the calibration dataset, perform iterative calculations on this alternative monitoring point. The iteration termination condition is to meet the convergence accuracy or reach the maximum number of iterations. The convergence accuracy and the number of iterations can be freely determined according to research needs. After the iterative calculation is completed, obtain U′(s) of the target inversion parameters, that is, the parameter covariance matrix after incorporating the calibration dataset.

[0072] S5.3: Repeat step S5.2 for all alternative monitoring points, and calculate U′(s) corresponding to each alternative monitoring point.

[0073] S6: Quantify the value of the observation information of each alternative monitoring point and optimize the monitoring network;

[0074] S6.1: Calculate the change in the variance of the target parameter when different calibration datasets are incorporated.

[0075] When using the observation information of the \(i\)-th alternative monitoring point as the calibration data set, the relative reduction amount \(\gamma\) of the variance of the \(n\)-th parameter n-i is:

[0076]

[0077] In the formula, and respectively represent the prior variance and the posterior variance of the \(n\)-th parameter when using the observation information of the \(i\)-th alternative monitoring point as the calibration data set, corresponding to the \(n\)-th elements on the diagonals of \(U(s)\) and \(U^{\prime}(s)\) respectively. The value range of \(\gamma\) n-i is from 0 to 1. The larger the value, the stronger the weakening ability of the observation information at the \(i\)-th alternative monitoring point to the uncertainty of the \(n\)-th parameter.

[0078] S6.2: Calculate the cumulative change amount of the variances of all parameters corresponding to each alternative monitoring point.

[0079] The cumulative change amount \(R\) of the variances of the parameters of the \(i\)-th alternative monitoring point i is defined as follows:

[0080]

[0081] In the formula, \(N\) represents the number of parameters to be inverted. \(R\) i can be used as a quantitative index of the value of the observation information at the \(i\)-th alternative monitoring point.

[0082] S6.3: Sort the alternative monitoring points according to the value of the observation information from high to low, and preferentially select the alternative monitoring points with high value of the observation information. The number of optimized monitoring points can be freely determined. The preferentially selected monitoring points will jointly form the final optimized monitoring network.

[0083] S7: Based on the monitoring network described in S6, optimize the optimal monitoring duration;

[0084] S7.1: Collect the observation information of all preferentially selected monitoring points in the optimized monitoring network described in S6, and perform discretization processing on the monitoring duration of the observation information.

[0085] Suppose the total duration (monitoring duration) simulated when obtaining the calibration data set in S4.2 is \(T\). Discretize \(T\) into \(k\) sub-durations \(t\), that is, \(T = kt\). The values of the parameters \(k\) and \(t\) can be freely selected according to requirements. The amount of observation information is proportional to the monitoring duration. The 0 - \(kt\) time period corresponds to all the observation information, and each \(t\) corresponds to a part of the observation information. The discretization of the observation information is realized by the discretization of the monitoring duration.

[0086] S7.2: Use the observation data of the optimized monitoring network as the calibration data set to quantify the change of the value of observation information with the monitoring duration.

[0087] Design the monitoring durations to be (0 - t), (0 - 2t), (0 - 3t)...(0 - kt) respectively. Reorganize the observation information obtained from the optimized monitoring network based on the above monitoring durations, with a total of k groups. Sequentially introduce each group of observation information as the calibration data set into the numerical model described in S1 to carry out parameter inversion. Information such as the parameters to be inverted, initial values, and prior intervals refers to S3.2. Calculate the posterior probability distribution of the parameters corresponding to the calibration data sets of different durations and quantify their data values, where R 1t 、R 2t 、R 3t …R kt respectively represent the value of observation information in the time periods of (0 - t), (0 - 2t), (0 - 3t)…(0 - kt).

[0088] S7.3: Weigh the value of observation information against the cost of monitoring time to optimize the monitoring duration.

[0089] Theoretically, the value of observation information will increase with the increase of the monitoring duration, growing rapidly in the early stage and then gradually stabilizing. The optimized monitoring duration t can be selected based on the following formula p :

[0090] R tp ≥Q t R kt

[0091] In the formula, R tp represents the value of observation information corresponding to the monitoring duration of 0 - t p , t p represents the shortest time that meets the above formula, R kt represents the value of observation information corresponding to the monitoring duration of 0 - kt, and Q t represents the time trade-off coefficient. If the cost associated with extending the monitoring time is high, Q t should take a lower value. Conversely, Q t can be appropriately increased. The recommended value range of Q t is 0.75 - 0.95.

[0092] S8: Optimize the optimal monitoring frequency based on the monitoring network described in S6;

[0093] The monitoring frequency represents the number of sampling times in a unit time step. The optimization of the monitoring frequency is independent of the optimization of the monitoring duration. The optimization of the monitoring frequency can be carried out separately after the optimization of the monitoring network (S6), or it can be carried out after the optimization of both the monitoring network and the monitoring duration (S6 and S7).

[0094] S8.1: Collect the observation information of all the preferred monitoring points in the optimized monitoring network described in S6, adjust the monitoring frequency, and discretize and group the observation information.

[0095] Let the total duration (monitoring duration) simulated when obtaining the calibration data set in S4.2 be T, the initial number of monitoring information be n0, and the initial monitoring frequency be m0. Then, n0 = m0T is satisfied. The monitoring frequency is reduced and grouped using the bisection method. Let the monitoring frequencies m be respectively The parameter y represents the number of groups and can be freely set according to requirements. After the above operations, the observation information is re-divided into y groups, and the corresponding numbers of monitoring information are respectively

[0096] S8.2: Use the discretely grouped observation data as the calibration data set to quantify the change in the value of the observation information with the monitoring frequency.

[0097] Successively introduce the y groups of observation information as the calibration data set into the numerical model described in S1 to carry out parameter inversion. Information such as the parameters to be inverted, initial values, and prior intervals refers to S3.2. Calculate the posterior probability distribution of the parameters corresponding to calibration data sets of different durations, and quantify their data value, R 1m 、R 2m 、R 3m …R ym respectively represent the value of the observation information corresponding to the monitoring frequencies of when.

[0098] S8.3: Weigh the value of the observation information and the potential impact of the monitoring frequency, and optimize the monitoring frequency.

[0099] Theoretically, the value of the observation information will increase with the increase of the monitoring frequency, growing rapidly in the early stage and then gradually stabilizing. The optimized monitoring frequency m can be selected based on the following formula p :

[0100] R mp ≥Q m R m0

[0101] In the formula, R mp represents the value of the observation information corresponding to the monitoring frequency of m p , m p represents the lowest frequency that meets the above formula, R m0 represents the value of the observation information corresponding to the monitoring frequency of m0, and Q m represents the frequency trade-off coefficient. If the impact associated with increasing the monitoring frequency is large, Q m should take a low value. Conversely, Q m can be appropriately increased. Q mThe recommended value range is 0.70 to 0.95.

[0102] Example Two

[0103] In this embodiment, a monitoring strategy optimization method based on the analysis of the value of observation information is provided, including:

[0104] S1: Collect geological exploration data and hydrogeological information of the research area, construct a geological model describing the stratum distribution of the research area and a groundwater numerical model reflecting the heterogeneous distribution characteristics of the aquifer;

[0105] S1.1: Conduct field surveys on the research area, collect relevant information on the lithologic distribution characteristics of the sedimentary structure part of the research area, simplify the lithologic distribution of the sedimentary structure into three types, namely fine sand (44.0%), silt (22.4%) and clay loam (33.6%), and draw a geological cross-section of the target area in combination with the hydrogeological conditions collected on site.

[0106] Utilize abundant borehole data, physical exploration data and detailed field outcrop observation data in the research area to obtain the geological spatial structure and heterogeneous characteristics of the research area, and construct a geological model reflecting the heterogeneity of the sedimentary structure of the research area through geostatistics and numerical simulation methods.

[0107] The size of the heterogeneous geological model constructed based on the sedimentary structure of the research area is 20km×10km×60m, which is divided into six layers vertically, with each layer having a thickness of 10m. The complex sedimentary continuous structure is discretized into regular unit grids with a size of 1km×0.5km×10m.

[0108] S1.2: In order to reduce the construction difficulty of the numerical model and be more conducive to identifying the inversion accuracy of the heterogeneous characteristic parameters, the parameters of different lithologic media are homogenized respectively. According to the results of grid division, the K-means corresponding to the lithology of each grid is assigned, aiming to eliminate the local heterogeneity within the same lithology.

[0109] S2: Based on the numerical model described in S1, set the positions of alternative groundwater monitoring stations;

[0110] In this embodiment, the research purpose is to plan new monitoring stations for the research area. Therefore, the positions of alternative monitoring stations can be freely set. Horizontally, the alternative monitoring stations are arranged in the heterogeneous geological model in the form of a 3×5 matrix, and vertically, they are arranged at depths of 5m, 15m, 25m, 35m, 45m and 55m, with a total of 90.

[0111] S3: Determine the parameters to be inverted, and clarify the initial values and confidence intervals of the inversion parameters;

[0112] S3.1: Determine the parameters to be inverted in the research;

[0113] Select the permeability coefficients (silty sand \(K_s\), fine sand \(K\) f , silty clay \(K\) c ) that affect the convective movement velocity of groundwater pollutants and the dispersivities (longitudinal \(\alpha\) 11 , transverse \(\alpha\) 22 , vertical \(\alpha\) 33 ) that reflect the pollutant dispersion ability of the medium as the target inversion parameters for the analysis of the value of observation information.

[0114] S3.2: Preset the initial values and prior intervals of the inversion parameters;

[0115] Set the initial parameter values according to the historical research data of the study area, set the upper and lower limits of \(K\) based on the mean value of the Darcy experiment results and the typical range of lithology, and set the upper and lower limits of \(\alpha\) based on the relatively small dispersivity at the laboratory scale and empirical values;

[0116] For the detailed information of the inversion parameters, see Table 1 in the appendix.

[0117] Table 1

[0118]

[0119]

[0120] S4: Based on the data collected in the study area, formulate the "standard values" of the parameters, and conduct simulations based on the numerical model in S1 to obtain the observation information at each alternative monitoring station as the calibration dataset;

[0121] S4.1: Based on the historical data of \(K\) and \(\alpha\) collected in the study area, take the mean value as the "standard value".

[0122] S4.2: Incorporate the parameter "standard values" set in S4.1 into the groundwater numerical model established in S1, use numerical simulation software to conduct flow and solute transport simulations, obtain a set of standardized breakthrough curves (BTC) at each alternative monitoring station, and a total of 90 sets of data (corresponding to 90 alternative monitoring stations) are obtained. Most BTCs will tend to the final stable value within 10 days. To ensure sufficient observation information is obtained, the initial experimental monitoring duration is set to 10 days.

[0123] S5: Conduct parameter inversion based on the calibration dataset in S4 to obtain the prior and posterior probability distributions of the parameters;

[0124] Take the BTC obtained at each alternative monitoring station in S4.2 as the calibration data set corresponding to that alternative monitoring station. Calculate the standard error and covariance matrix of the parameters through the PREDUNC and PREDVAR programs in the parameter estimation and uncertainty analysis software PEST suite, and output the parameter estimation values and the parameter covariance matrix. As the number of introduced alternative monitoring stations increases and the amount of observed data increases, the confidence interval of the parameter estimation values output by the program will gradually converge to the "standard value" set in S4.1 until the iterative calculation ends, obtaining the posterior model and outputting the final parameter inversion result, that is, the posterior information.

[0125] S5.1: Complete the parameter input in PEST. When no observed data is introduced, output the prior covariance matrix U(s) of all parameters at each alternative monitoring station, and extract the elements on the diagonal from U(s) to obtain the prior variances of the parameters.

[0126] S5.2: Introduce the BTC data of each alternative monitoring station one by one as the calibration data set for that alternative monitoring station, carry out parameter inversion through the PEST software, obtain the posterior model, output the posterior covariance matrix U′(s) of all parameters calculated based on the posterior model at each alternative monitoring station, and extract the elements on the diagonal from U′(s) to obtain the posterior variances of the parameters.

[0127] S6: Quantify the value of the observed information of each alternative monitoring station, and select the alternative monitoring stations with high observed information value as the optimized monitoring stations;

[0128] S6.1: Calculate the relative reduction γ of the variances of the 6 parameters respectively n-i (i = 1, 2, 3…6);

[0129]

[0130] S6.2: Calculate the cumulative variance change amount R of all parameters at each alternative monitoring station i (i = 1, 2, 3…6), that is, the value of the observed information of each alternative monitoring station;

[0131]

[0132] Statistically analyze the value of the observed information of each alternative monitoring station and sort them (see the appendix Figure 2 ), and there are 38 alternative monitoring stations with the value of the observed information, accounting for about 42% of the total number of alternative monitoring stations.

[0133] S6.3: Optimize the monitoring network according to the value of the observed information;

[0134] In this embodiment, all alternative monitoring stations with the value of the observed information are retained as the optimized monitoring stations, where the appendixFigure 3 The layout of alternative monitoring stations in the study area and the optimization results of monitoring stations within the vertical depth range of 40 to 50 m are shown.

[0135] S7: Optimizing the monitoring station network based on S6 and optimizing the optimal monitoring time;

[0136] In this embodiment, the observation information is discretized into k = 10 groups, the unit time step is t = 1 day, the simulation duration is T = 10 days, and the observation information is reorganized;

[0137] S7.1: Based on the optimized monitoring station network obtained in S6, select the BTC data of the i (i=1, 2, 3…10) unit time step periods before the research period of each optimized monitoring station one by one, and the monitoring periods are (0-1 day), (0-2 days), (0-3 days)…(0-10 days) respectively. In turn, introduce the 10 sets of observation information as calibration data sets into the groundwater numerical model established in S1, set the parameter-related information according to Table 1, carry out parameter inversion, and output the parameter posterior probability distribution of the calibration data sets corresponding to the 10 sets of different monitoring periods, and calculate the observation information value R1, R2, R3…R 10 .

[0138] S7.2: Weigh the value of observation information against the cost of monitoring time and optimize monitoring duration;

[0139] As time goes on, the value of observation information increases rapidly in the first 4 days, and then gradually stabilizes. Weighing the value of observation information and monitoring costs, we set Q t =0.95, R tp ≥0.95×R 10 =5.127, so it is more appropriate to set the monitoring time to 5 days.

[0140] S8: Optimizing the optimal monitoring frequency based on the optimized monitoring station network described in S7;

[0141] In this embodiment, after the optimization of the monitoring station network arrangement and the optimization of the monitoring duration are completed (S6 and S7), the monitoring frequency is further optimized.

[0142] S8.1: Based on the optimization results of S6 and S7, the observation information of each alternative monitoring station (0-5 days) is extracted (i.e., T′=5 days), the initial sampling frequency is set to m0=1440 times / day (monitoring stations are monitored once every 1 min), the initial number of monitoring information is n0=1440×5=7200, and the monitoring frequency is adjusted by dichotomy (720, 360, ...45, 22...1, 0.5, 0.2 times / day, a total of 13 groups), and the observation information is discretized and reorganized.

[0143] S8.2: Introduce the 13 groups of observation information into the groundwater numerical model established in S1 as calibration datasets in sequence. Set the relevant parameter information according to Table 1, conduct parameter inversion, respectively output the posterior probability distributions of the parameters corresponding to the calibration datasets introduced with 13 different monitoring frequencies, and calculate the observation information values R′1, R′2, R′3…R′ 13 。

[0144] S8.3: Weigh the potential impacts of the observation information value and the monitoring frequency, and optimize the monitoring frequency.

[0145] In this embodiment, the variation of the observation information value of each alternative monitoring station with the monitoring frequency can be seen in the appendix Figure 4 (To more intuitively see the variation of the observation information value, the monitoring frequency in the abscissa is plotted using a logarithmic coordinate axis). Weigh the observation information value against the monitoring cost that climbs with the increase in the monitoring frequency. Let Q m = 0.95, R mp ≥ 0.95 × R 13 = 4.888. Therefore, it is more appropriate to set the monitoring frequency to m p = 22 times per day.

[0146] The above is only the preferred specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the technical field of this application within the technical scope disclosed in this application should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. A method for optimizing a monitoring strategy based on the analysis of the value of observation information, characterized in that Including: Construct a numerical model to describe the system behavior based on the research objective, and set alternative monitoring points in the research area based on the numerical model; Construct the parameters to be inverted in the research, and set the initial values and confidence intervals of the parameters to be inverted; Input the parameters to be inverted into the numerical model, obtain the simulated observation information in the alternative monitoring points, and generate a calibration data set based on the simulated observation information; Perform parameter inversion on the calibration data set to obtain the parameter posterior covariance matrix; Calculate the value of the observation information for all the alternative monitoring points based on the parameter posterior covariance matrix, and optimize the monitoring network according to the ranking of the observation information value; Optimize the monitoring duration and monitoring frequency based on the optimized monitoring network to obtain the optimal monitoring duration and the optimal monitoring frequency.

2. The method according to claim 1, wherein The process of constructing a numerical model to describe the system behavior based on the research objective includes: Collect the data required for the numerical model according to the monitoring objective and the nature of the observation information; Construct a physical model to describe the environment or medium state of the observation information and a mathematical model to reflect the relationship between the observation information and the target parameters based on the data; 3. The method according to claim 1, characterized in that, The process of constructing the parameters to be inverted in the research and clarifying the initial values and confidence intervals of the inversion parameters includes: formulating the parameters to be inverted, setting the initial values of the parameters to be inverted according to the reliable data, expert experience and literature results in the historical research of the target parameters, and giving the prior interval of the parameters to be inverted.

4. The method according to claim 1, wherein The process of generating a calibration data set based on the simulated observation information includes: Formulate a set of approximate values of the parameters to be inverted as the standard values based on the historical data; Input the standard values into the numerical model to obtain the simulated observation information of each alternative monitoring point through numerical simulation; Process the simulated observation information of each alternative monitoring point to obtain the calibration data set.

5. The method according to claim 1, wherein The process of performing parameter inversion on the calibration data set to obtain the prior and posterior probability distributions of the parameters includes: Input the initial values and confidence intervals of the parameters to be inverted into the numerical model to obtain the parameter covariance matrix before any calibration; Take the observation information of all alternative monitoring points as the data set, perform iterative calculations on each alternative monitoring point among all alternative monitoring points until the preset conditions are met and then output to obtain the parameter posterior covariance matrix.

6. The method according to claim 1, wherein The process of optimizing the monitoring network according to the ranking of the observation information value includes: Calculate the change amount of the target parameter variance when the parameter posterior covariance matrix incorporates different calibration data sets; Calculate the cumulative change amount of the variances of all parameters based on the change amount of the target parameter variance for different calibration data sets; Use the cumulative change amount of the variances of all parameters as the observation information value to optimize the monitoring network.

7. The method according to claim 1, wherein The process of obtaining the optimal monitoring duration includes: Obtain the observation information of all the preferred monitoring points in the optimized monitoring network, perform discretization processing on the monitoring duration of the observation information to obtain a monitoring duration data set; Calculate the parameter posterior probability distribution corresponding to the monitoring duration data set of different durations, and quantify the value of the time data based on the parameter posterior probability distribution; Integrate the value of the time data and the monitoring time cost to obtain the optimal monitoring duration.

8. The method according to claim 1, characterized in that The process of obtaining the optimal monitoring frequency includes: Obtaining the observation information of all the preferred monitoring points in the optimized monitoring network, adjusting the monitoring frequency, discretizing and grouping the observation information, and obtaining a monitoring frequency data set; Calculating the posterior probability distribution of the parameters corresponding to the monitoring frequency data set at different frequencies, and quantifying the value of the frequency data based on the posterior probability distribution of the parameters; Obtaining the optimal monitoring frequency based on the change trend of the quantified frequency data value with the monitoring frequency.

9. An electronic device, comprising a memory, a processor, and a computing program stored in the memory and executable on the processor, characterized in that, When the processor executes the calculation program, the method described in any one of claims 1-8 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, the method described in any one of claims 1-8 is implemented.

Images