Patent title translation: Digital-twin-based patrol data analysis method, device and equipment

Abstract

The application discloses a digital-twin-based patrol data analysis method, device and equipment, relates to the technical field of patrol data analysis, and comprises the following steps: constructing a digital twin environment of a preset area; acquiring initial atmospheric patrol data of a mobile inspection device in the preset area; performing probability inversion of atmospheric diffusion to generate a plurality of candidate pollution trajectories and initial probability distribution thereof; performing multi-round iteration analysis of a patrol path based on maximum confidence of causal reasoning; performing multi-round supplementary patrol data analysis until at least one pollution trajectory meets a preset convergence condition; and outputting a target pollution trajectory. The application solves the technical problems of low atmospheric pollution trajectory positioning accuracy, lack of intelligence in patrol path planning, and insufficient data analysis efficiency and accuracy in the prior art, achieves accurate atmospheric pollution trajectory positioning and intelligent iteration optimization of a patrol path, and improves the accuracy and efficiency of atmospheric pollution patrol data analysis.

Description

Technical Field

[0001] This invention relates to the field of survey data analysis technology, specifically to survey data analysis methods, devices, and equipment based on digital twins. Background Technology

[0002] In the field of air pollution monitoring and source tracing, traditional air pollution survey data analysis methods rely heavily on manual planning of survey routes, comparison of single monitoring data, and empirical judgment of pollution trajectories. They lack accurate simulation and quantitative analysis of pollution diffusion patterns and fail to build an integrated analysis system that combines multi-dimensional data such as geography, meteorology, and pollution sources. This not only makes it difficult to accurately locate pollution trajectories but also results in problems such as poor rationality of survey route planning and untimely iterative optimization. This easily leads to low utilization rate of survey data and low analysis efficiency. Furthermore, the lack of quantitative assessment of the causal relationship of pollution trajectories makes it difficult to guarantee the accuracy and reliability of pollution source tracing results, thus failing to meet the current needs for refined and intelligent air pollution prevention and source tracing.

[0003] Existing technologies suffer from low accuracy in locating atmospheric pollution trajectories, a lack of intelligence in survey path planning, and insufficient efficiency and accuracy in data analysis. Summary of the Invention

[0004] This application provides a method, apparatus, and equipment for analyzing survey data based on digital twins, which are used to address the technical problems of low accuracy in locating atmospheric pollution trajectories, lack of intelligence in survey path planning, and insufficient efficiency and accuracy in data analysis in existing technologies.

[0005] In view of the above problems, this application provides a method, apparatus and equipment for survey data analysis based on digital twins.

[0006] The first aspect of this application provides a method for analyzing survey data based on digital twins, the method comprising: A digital twin environment for a predefined area is constructed, including 3D geographic information, meteorological field data, fixed monitoring station data, and a pollution source list. Initial atmospheric survey data of a mobile inspection device in the predefined area is acquired. When the initial atmospheric survey data shows anomalies, atmospheric diffusion probability inversion is performed based on the initial atmospheric survey data and the digital twin environment to generate multiple candidate pollution trajectories and their initial probability distributions. For the multiple candidate pollution trajectories, combined with meteorological time-series data, geographical blocking factors, and pollution source emission characteristics, a multi-round iterative analysis of the survey path based on maximizing the confidence of causal inference is performed. The mobile inspection device is controlled to perform multiple rounds of supplementary survey data analysis based on the iterative analysis results until at least one pollution trajectory meets the preset convergence condition, and the target pollution trajectory is output.

[0007] A second aspect of this application provides a digital twin-based survey data analysis device, the device comprising: The system includes a digital twin environment construction module for building a digital twin environment for a preset area, including 3D geographic information, meteorological field data, fixed monitoring station data, and a pollution source list; an atmospheric survey data acquisition module for acquiring initial atmospheric survey data from mobile inspection equipment in the preset area; a probability inversion module for performing probability inversion of atmospheric diffusion based on the initial atmospheric survey data and the digital twin environment when the initial atmospheric survey data shows anomalies, generating multiple candidate pollution trajectories and their initial probability distributions; and a target pollution trajectory output module for performing multi-round iterative analysis of the multiple candidate pollution trajectories based on maximizing the confidence of causal inference, combining meteorological time-series data, geographical blocking factors, and pollution source emission characteristics, and controlling the mobile inspection equipment to perform multiple rounds of supplementary survey data analysis based on the iterative analysis results until at least one pollution trajectory meets the preset convergence condition, and outputting the target pollution trajectory.

[0008] A third aspect of this application provides an electronic device comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the digital twin-based survey data analysis method provided in this application.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages: A digital twin environment of a preset area is constructed; initial atmospheric survey data of the preset area is acquired by a mobile inspection device; when the initial atmospheric survey data shows anomalies, atmospheric diffusion probability inversion is performed to generate multiple candidate pollution trajectories and their initial probability distributions; for the multiple candidate pollution trajectories, combined with meteorological time-series data, geographical blocking factors, and pollution source emission characteristics, a multi-round iterative analysis of the survey path based on maximizing the confidence of causal inference is performed, and the mobile inspection device is controlled to perform multiple rounds of supplementary survey data analysis based on the iterative analysis results until at least one pollution trajectory meets the preset convergence condition, and the target pollution trajectory is output. This achieves the technical effect of accurate positioning of atmospheric pollution trajectories and intelligent iterative optimization of survey paths, improving the accuracy and efficiency of atmospheric pollution survey data analysis. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A schematic diagram of the process for analyzing survey data based on digital twins provided in this application embodiment.

[0012] Figure 2 A schematic diagram of the structure of the survey data analysis device based on digital twin provided in the embodiments of this application.

[0013] Figure 3 This is a schematic diagram of the structure of an electronic device provided in this application.

[0014] Figure labeling: Digital twin environment construction module 10, atmospheric survey data acquisition module 20, probability inversion module 30, target pollution trajectory output module 40, processor 21, memory 22, input device 23, output device 24. Detailed Implementation

[0015] This application provides a digital twin-based survey data analysis method, apparatus, and equipment to address the technical problems of low accuracy in locating atmospheric pollution trajectories, lack of intelligence in survey path planning, and insufficient efficiency and accuracy in data analysis in existing technologies.

[0016] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0017] Example 1, as Figure 1 As shown, this application provides a method for analyzing survey data based on digital twins, the method comprising: Step S100: Construct a digital twin environment for the preset area, including three-dimensional geographic information, meteorological field data, fixed monitoring station data, and a pollution source list.

[0018] Specifically, a digital twin environment for a predefined area is constructed. This environment integrates and builds high-precision three-dimensional geographic information covering the entire predefined area, including geospatial features such as terrain, landforms, and buildings. It also synchronously accesses meteorological field data for the area, including the spatiotemporal distribution and variation data of meteorological elements such as wind speed, wind direction, temperature, and humidity. It integrates real-time and historical monitoring data from various fixed monitoring stations within the area to achieve digital mapping of station monitoring information. At the same time, it records a list of pollution sources in the predefined area, including basic information such as the location, emission type, emission intensity, and emission period of various pollution sources within the area. Through the fusion of multi-dimensional data, a complete and dynamic digital twin environment for the predefined area is constructed.

[0019] Step S200: Obtain initial atmospheric survey data of the mobile inspection device in the preset area.

[0020] Specifically, by deploying mobile inspection equipment within a pre-defined area, on-site atmospheric environmental monitoring and data collection are carried out to accurately obtain initial atmospheric monitoring data within that area. This data comprehensively includes the specific types of air pollutants, real-time concentration values ​​of pollutants, the collection timestamps corresponding to each monitoring data point, and the precise geographical location information of the mobile inspection equipment at the time of collection of each set of monitoring data. This batch of measured data will complement the data from fixed monitoring stations in the digital twin environment, providing a core foundation of on-site measured data for subsequent atmospheric environmental anomaly determination, atmospheric diffusion probability inversion, and pollution trajectory analysis.

[0021] Step S300: When the initial atmospheric survey data shows an anomaly, based on the initial atmospheric survey data and the digital twin environment, perform probability inversion of atmospheric diffusion to generate multiple candidate pollution trajectories and their initial probability distributions.

[0022] Specifically, the initial atmospheric survey data, which includes information on pollutant type, concentration, timestamp, and corresponding geographical location, is first anomaly identified. When monitoring indicators show an abnormal atmospheric environment, the initial atmospheric survey data is input into a pre-constructed digital twin environment for a preset area. The atmospheric diffusion-driven branch in the environment is loaded, and the spatiotemporal diffusion process of pollutants released from potential source locations within the area is simulated in the digital twin environment, forming a diffusion simulation dataset. Then, a Bayesian inversion method is used to perform correlation analysis between the simulated concentration distribution of pollutants in the diffusion simulation dataset and the measured concentration distribution in the initial atmospheric survey data, inferring the posterior probability distribution of the pollution source location and release time. Subsequently, a Monte Carlo sampling method is used to extract multiple samples from this posterior probability distribution. Each sample corresponds to a set of pollution source parameters, and corresponding pollution trajectories are generated based on the digital twin environment. Each pollution trajectory contains core information on the pollution source location, release time, and diffusion path. The degree of fit between each pollution trajectory and the initial atmospheric survey data is calculated, and an initial probability is assigned to each trajectory based on the degree of fit. Finally, multiple candidate pollution trajectories are generated, and the initial probability distribution corresponding to each candidate pollution trajectory is determined.

[0023] Step S400: For the multiple candidate pollution trajectories, combine meteorological time series data, geographical blocking factors and pollution source emission characteristics, perform multi-round iterative analysis of the inspection path based on maximizing the confidence of causal inference, and control the mobile inspection equipment to perform multiple rounds of supplementary inspection data analysis based on the iterative analysis results until at least one pollution trajectory meets the preset convergence condition, and output the target pollution trajectory.

[0024] Specifically, for multiple candidate pollution trajectories already generated, a virtual intervention experiment is conducted in a digital twin environment, integrating meteorological time-series data, geographical blocking factors within the preset area, and pollution source emission characteristics. This involves removing hypothetical pollution source nodes, re-running the atmospheric diffusion-driven branch, and observing concentration changes at monitoring points to calculate the intervention consistency index, thus obtaining the causal confidence score for each pollution trajectory. Based on the causal confidence score, the expected increase in causal confidence score for different monitoring paths is simulated in the digital twin environment. The path with the largest increase is selected as the next monitoring path, and mobile monitoring equipment is controlled to conduct supplementary monitoring along this path. The supplementary monitoring data is then combined with the initial atmospheric monitoring data to recalculate the causal confidence score for each trajectory. The system updates the causal confidence and probability distribution of each candidate contaminated trajectory in one round. It then determines whether the updated indicators satisfy the preset convergence condition that the causal inference confidence is greater than a first threshold and the trajectory probability is greater than a second threshold. If not, it extracts taboo trajectories with updated causal confidence greater than the first threshold but trajectory probability less than the second threshold and removes them from the candidate trajectories. For the remaining trajectories, it repeats the above-mentioned iterative analysis of the survey path, supplementary surveys, and indicator updates. If the number of iterations reaches the preset maximum number of iterations and no trajectory still satisfies the convergence condition, it selects the trajectory with the highest current causal confidence and trajectory probability as the target contaminated trajectory. If at least one trajectory satisfies the convergence condition, it is directly output as the target contaminated trajectory.

[0025] In one possible implementation, step S400 further includes: The convergence condition is that the causal inference confidence of the contaminated trajectory is greater than a first threshold and the trajectory probability is greater than a second threshold.

[0026] Specifically, the convergence condition is as follows: after multiple rounds of iterative analysis and supplementary surveys of candidate contamination trajectories, the contamination trajectory to be judged must simultaneously meet two quantitative indicator requirements: First, the causal reasoning confidence value calculated by the contamination trajectory must be greater than a pre-set first threshold to ensure the rationality and reliability of the trajectory at the causal logic level; Second, the trajectory probability value corresponding to the contamination trajectory must be greater than a pre-set second threshold to ensure the credibility and effectiveness of the trajectory at the probability distribution level. Only when both of these indicator requirements are met can the contamination trajectory be judged to meet the convergence condition and used as the basis for judging the target contamination trajectory.

[0027] In one possible implementation, step S300 further includes: Step S310: Input the initial atmospheric survey data into the digital twin environment, load the atmospheric diffusion driving branch, simulate the spatiotemporal diffusion process of pollutants after release from potential source locations in the digital twin environment, and construct a diffusion simulation dataset.

[0028] Step S320: Using the Bayesian inversion method, the simulated concentration distribution in the diffusion simulation dataset and the measured concentration distribution in the initial atmospheric survey data are combined to infer the posterior probability distribution of the pollution source location and release time.

[0029] Step S330: Based on the posterior probability distribution, generate the multiple candidate pollution trajectories in the digital twin environment by sampling pollution sources, and calculate the initial probability distribution.

[0030] Specifically, initial atmospheric survey data, carrying information on pollutant type, concentration, collection timestamp, and corresponding geographical location, are completely input into a pre-constructed digital twin environment for the pre-defined area. Simultaneously, a pre-configured atmospheric diffusion-driven branch within this environment is loaded. Based on the integrated 3D geographic information, meteorological field data, fixed monitoring station data, and pollution source inventory within the digital twin environment, the system accurately simulates the spatiotemporal diffusion process of pollutants released from various potential sources within the region, influenced by meteorological conditions and geographical obstructions. Key data such as pollutant concentration distribution, diffusion and transport paths, and diffusion rates at different time points and geographical locations are recorded concurrently during the simulation. After systematically integrating and organizing these simulation data, a complete and accurate diffusion simulation dataset is constructed, providing comprehensive simulation data support for subsequent back-analysis of pollution source-related parameters.

[0031] The Bayesian inversion method is used to back-calculate pollution source parameters. This method is based on Bayes' theorem. As the core structure, in which For parameters where the location and release time of the pollution source are unknown and need to be determined, The prior probability distribution of the parameters is assigned by the characteristics of the simulated pollutant concentration distribution corresponding to different potential source locations and different release times in the diffusion simulation dataset. Let be the likelihood function, representing the likelihood of a given pollution source parameter. Initial atmospheric survey data were obtained. The probability is constructed by calculating the deviation between the concentration distribution in the diffusion simulation and the measured concentration distribution in the initial atmospheric survey data. P(D) is the evidence factor, which is related to the parameter. Unrelated normalization constants; in the implementation process, the simulated concentration distribution of the diffusion simulation dataset is first transformed into a prior probability distribution of the pollution source location and release time. Then, based on the measured concentration distribution of the initial atmospheric survey data, different parameters are calculated. The corresponding likelihood function Then, the probability calculation was performed by substituting the values ​​into Bayes' theorem, and the prior probability distribution was corrected and updated to eliminate the deviation between the simulated concentration and the measured concentration. Finally, the posterior probability distribution of the pollution source location and release time, which integrates the simulated and measured data, was obtained. This distribution can quantify the probability of matching the location and release time of each potential pollution source with the measured data.

[0032] Based on the posterior probability distribution of pollution source location and release time, Monte Carlo sampling of pollution source parameters is carried out in a digital twin environment. Multiple sets of pollution source parameter samples are randomly selected from the posterior probability distribution. Each set of samples corresponds to a specific set of pollution source location and release time parameters. Based on the atmospheric diffusion driving branch of the digital twin environment for each set of parameters, corresponding pollution trajectories are simulated and generated. Each pollution trajectory fully presents the spatiotemporal diffusion path of pollutants after release from the corresponding pollution source location. Subsequently, each candidate pollution trajectory is compared with the initial atmospheric survey data in multiple dimensions. The degree of consistency between the diffusion concentration distribution, spatiotemporal characteristics and measured concentration distribution and monitoring spatiotemporal information corresponding to the trajectory is calculated. According to the degree of consistency, each candidate pollution trajectory is assigned a corresponding initial probability value. All candidate pollution trajectories and their corresponding initial probability values ​​together constitute multiple candidate pollution trajectories and their initial probability distribution.

[0033] In one possible implementation, step S330 further includes: Step S331: Use the Monte Carlo sampling method to extract multiple samples from the posterior probability distribution, each sample corresponding to a pollution source parameter, and generate a corresponding pollution trajectory based on the digital twin environment.

[0034] Step S332: Calculate the degree of agreement between each pollution trajectory and the initial atmospheric survey data, and assign an initial probability to each trajectory based on the degree of agreement to form the multiple candidate pollution trajectories and the initial probability distribution.

[0035] Specifically, the Monte Carlo sampling method is used to randomly sample the posterior probability distribution of the obtained pollution source location and release time. Multiple independent samples are drawn from this probability distribution, and each sample corresponds to a unique set of pollution source parameters. These parameters include core back-inference information such as the specific geographical location of the pollution source and the release time of the pollutants. Each set of pollution source parameters is input into the digital twin environment of the preset area, and the atmospheric diffusion driving branch in the environment is loaded simultaneously. Based on the basic data such as three-dimensional geographic information, meteorological field data, and geographic blocking factors integrated in the digital twin environment, the spatiotemporal diffusion process of pollutants in the atmosphere after being released from the pollution source location and release time corresponding to the sample parameter is accurately simulated. The characteristics of pollutant transport path and spatiotemporal concentration changes are restored, and finally a corresponding pollution trajectory is generated for each set of pollution source parameters in the digital twin environment.

[0036] For each pollution trajectory generated in the digital twin environment, a full-dimensional matching analysis is performed with the initial atmospheric survey data, which includes information on pollutant type, concentration, collection timestamp, and corresponding geographical location. First, each spatiotemporal node of the pollution trajectory is matched one-to-one with the monitoring node of the survey data. If the pollutant type is inconsistent, the matching degree of that node is recorded as 0; if the types are consistent, the matching degree is determined by a formula. Calculate the concentration fit at a single node. The concentration match of the i-th node. To simulate concentration, For the measured concentration, a negative result is taken as 0; then through... The overall trajectory matching degree is calculated, where R is the spatiotemporal point overlap coverage rate, i.e., the number of effective matching nodes / the total number of survey nodes. A weight is preset for the i-th node, and the sum of the weights is 1. n is the number of valid nodes that match the type. The single-node concentration match is then normalized. The normalized value is directly used as the initial probability of the corresponding trajectory. The higher the match, the greater the initial probability. Finally, all pollution trajectories with initial probability assignments are integrated to form multiple candidate pollution trajectories. The initial probabilities of each trajectory together constitute a standard initial probability distribution.

[0037] In one possible implementation, step S400 further includes: Step S410: For each candidate pollution trajectory, conduct a virtual intervention experiment using meteorological time series data, geographical blocking factors, and pollution source emission inputs to obtain the causal confidence level of each pollution trajectory.

[0038] Step S420: Based on the causal confidence level, in the digital twin environment, simulate the expected growth rate of the causal confidence level for different inspection paths, and select the path with the largest growth rate as the next round of inspection path.

[0039] Step S430: Control the mobile inspection equipment to conduct supplementary inspections according to the next round of inspection path, and recalculate the one-round update causal confidence and one-round update probability distribution of each pollution trajectory based on the supplementary inspections and the initial atmospheric inspection data.

[0040] Step S440: Determine whether the causal confidence of the first round of updates and the probability distribution of the first round of updates satisfy the convergence condition. If not, repeat the iterative path survey.

[0041] Specifically, for each candidate pollution trajectory, a combination of virtual node intervention and index quantification is employed. First, the meteorological time-series data, geographical blocking factors, and pollution source emission characteristics of the preset area are fully input into the digital twin environment and fused with the corresponding candidate pollution trajectory. Then, a virtual intervention operation is performed on each trajectory, namely, removing the hypothetical pollution source nodes in the trajectory, reloading the atmospheric diffusion driving branch, and running the atmospheric diffusion simulation program. Pollutant concentration data of monitoring points before and after the intervention are collected simultaneously. The causal confidence is quantified by calculating the intervention consistency index of concentration change. Specifically, the deviation rate between the simulated and measured concentration values ​​after intervention and the matching degree of concentration change trend are used as the core calculation dimensions. The results of the two dimensions are weighted and summed to obtain the quantified intervention consistency index. The value of this index is the causal confidence of the corresponding candidate pollution trajectory. The causal confidence calculation of all candidate pollution trajectories is completed.

[0042] This method employs path simulation and quantitative assessment of the increase in causal confidence. Using the calculated causal confidence scores of each candidate pollution trajectory as a baseline, and relying on fundamental data such as 3D geographic information and meteorological time-series data of a pre-defined area within a digital twin environment, multiple candidate monitoring paths covering key areas of potential pollution diffusion are planned and generated. Virtual simulations are performed on each candidate monitoring path, simulating the process of mobile monitoring equipment collecting supplementary monitoring data along the path. The simulated supplementary data is then merged with the original data, and the simulated causal confidence scores of each candidate pollution trajectory are recalculated. The expected increase in causal confidence scores for each candidate monitoring path is obtained by calculating the difference between the simulated value and the baseline value. The expected increase in causal confidence scores for all candidate monitoring paths is then quantitatively ranked, and the monitoring path with the largest expected increase is selected as the supplementary monitoring path for the next round of actual execution.

[0043] Based on the selected next round of monitoring path, a monitoring command containing path locations, monitoring frequency, and pollutant detection type is issued to the mobile monitoring equipment. The equipment is then controlled to conduct supplementary atmospheric monitoring along the path in a preset area, simultaneously collecting supplementary monitoring data including pollutant concentration, collection timestamp, and corresponding geographical location, and completing data verification. The verified supplementary monitoring data is then fused with the initial atmospheric monitoring data to form a complete atmospheric monitoring dataset. Relying on the digital twin environment, the virtual intervention experiment calculation method for causal confidence and the probability distribution consistency calculation method are used to re-quantify and calculate the causal confidence and probability distribution of each candidate pollution trajectory based on the fused complete dataset. This yields the updated causal confidence and updated probability distribution for each trajectory, providing the latest quantitative indicators for judging subsequent convergence conditions.

[0044] The recalculated causal confidence and probability of each candidate contamination trajectory are compared one by one with the preset convergence criteria. The convergence criteria are that the updated causal confidence of the contamination trajectory is greater than the first threshold and the updated trajectory probability is greater than the second threshold. If at least one candidate contamination trajectory meets both threshold requirements, the convergence standard is met, subsequent inspection path iterations are stopped, and this trajectory is output as the target contamination trajectory. If none of the candidate contamination trajectories meet both threshold requirements, the convergence standard is not met. The inspection path iteration operation is repeated for the current set of candidate contamination trajectories. The entire process from causal confidence calculation, inspection path planning, supplementary inspection execution to index recalculation is carried out again until a contamination trajectory that meets the convergence criteria appears.

[0045] In one possible implementation, step S410 further includes: By removing the hypothetical pollution source nodes and rerunning the atmospheric diffusion-driven branch, we observed the concentration changes at monitoring points, calculated the intervention consistency index, and obtained the causal confidence of each pollution trajectory.

[0046] Specifically, for each pollution trajectory, the corresponding hypothetical pollution source node is first located in the digital twin environment and a virtual removal operation is performed to remove the node from the diffusion simulation parameter system. Then, the atmospheric diffusion driving branch is reloaded, and the atmospheric diffusion simulation program is run in conjunction with the three-dimensional geographic information, meteorological time-series data, geographical blocking factors, and pollution source emission characteristics of the preset area to generate pollutant diffusion concentration distribution data without the pollution source node. Simultaneously, the simulated concentrations after intervention at each monitoring point are collected. At the same time, the original simulated concentrations before intervention at each monitoring point before the node removal are retrieved, and the two sets of concentration data are compared point by point to analyze the concentration changes. The three core dimensions of concentration change rate, concentration deviation convergence, and spatiotemporal trend consistency are used to quantify the intervention consistency index through a weighted summation method. The concentration change rate is the ratio of the concentration difference before and after the intervention to the concentration before the intervention. The concentration deviation convergence is the improvement ratio of the deviation between the simulated concentration and the measured concentration after the intervention compared to the concentration before the intervention. The spatiotemporal trend consistency is the matching coefficient of the spatiotemporal trend of concentration before and after the intervention. The weights of each dimension are preset according to the core analysis needs of the monitoring data. The final calculated intervention consistency index value is the causal confidence of the pollution trajectory. The higher the index value, the stronger the causal relationship credibility of the trajectory.

[0047] In one possible implementation, step S440 further includes: Step S441: Based on the first-round update causal confidence and the first-round update probability distribution, extract taboo trajectories whose update causal confidence is greater than the first threshold but whose trajectory probability is less than the second threshold.

[0048] Step S442: After deleting the taboo trajectory from the multiple candidate contaminated trajectories, repeat the survey path iteration for the remaining trajectory.

[0049] Specifically, using a preset first threshold and a second threshold as dual judgment criteria, the updated causal confidence and updated probability distribution of all candidate pollution trajectories are checked and quantitatively screened track by track. First, the updated causal confidence value and updated trajectory probability value corresponding to each trajectory are extracted. Then, the relationship between the two values ​​and the corresponding thresholds is compared one by one to accurately screen out candidate pollution trajectories whose updated causal confidence value is greater than the first threshold but whose updated trajectory probability value is less than the second threshold. Such trajectories are defined as taboo trajectories. Although such trajectories have a certain causal correlation validity, their matching degree with the measured and surveyed data is insufficient, and they are not worth further iterative analysis. This clarifies the specific targets for subsequent trajectory elimination operations.

[0050] All taboo trajectories identified through screening are removed from the current candidate contaminated trajectory set. After cleaning, only the remaining candidate contaminated trajectories that have not triggered taboo conditions are retained to form new candidate trajectory analysis objects. Subsequently, for this remaining trajectory set, the entire process of iterative operation of the inspection path is repeated, from causal confidence calculation, inspection path simulation planning and selection, supplementary inspection execution by mobile inspection equipment, to recalculating and updating causal confidence and probability distribution by fusing data, and judging convergence conditions. The process is continuously iterated and optimized until a trajectory that simultaneously satisfies the condition of updating causal confidence greater than the first threshold and trajectory probability greater than the second threshold appears in the remaining trajectory, thus completing the screening and output of the target contaminated trajectory.

[0051] In one possible implementation, step S440 further includes: If the number of iterations reaches the preset maximum number of iterations, and there is no contaminated trajectory that satisfies the preset convergence condition, the trajectory with the highest current causal confidence and trajectory probability is selected as the target contaminated trajectory.

[0052] Specifically, during the iterative process of the inspection path, each iteration operation is counted and compared in real time with the preset maximum number of iterations. If the number of iterations reaches the preset maximum number of iterations, and after verification, there is still no trajectory among all the current candidate contaminated trajectories that can simultaneously meet the preset convergence conditions of causal confidence greater than the first threshold and trajectory probability greater than the second threshold, then the subsequent inspection path iteration operation is stopped. Subsequently, all the remaining candidate contaminated trajectories are sorted by indicators, and the two core quantitative indicators of the current causal confidence and trajectory probability of each trajectory are extracted comprehensively. Through the dual-indicator comprehensive evaluation method, the candidate contaminated trajectory with the largest causal confidence value and trajectory probability value is selected and directly selected as the final target contaminated trajectory, thus completing the output of the contaminated trajectory determination.

[0053] Example 2, based on the same inventive concept as the digital twin-based survey data analysis method in the previous examples, such as... Figure 2 As shown, this application provides a survey data analysis device based on digital twins. The device and method embodiments in this application are based on the same inventive concept. The device includes: The digital twin environment construction module 10 is used to construct a digital twin environment for a preset area, including three-dimensional geographic information, meteorological field data, fixed monitoring station data, and a pollution source list.

[0054] The atmospheric survey data acquisition module 20 is used to acquire the initial atmospheric survey data of the mobile inspection equipment in the preset area.

[0055] The probability inversion module 30 is used to perform probability inversion of atmospheric diffusion based on the initial atmospheric survey data and the digital twin environment when the initial atmospheric survey data shows anomalies, and generate multiple candidate pollution trajectories and their initial probability distributions.

[0056] The target pollution trajectory output module 40 is used to perform multi-round iterative analysis of the multiple candidate pollution trajectories based on maximizing the confidence of causal inference, in combination with meteorological time series data, geographical blocking factors and pollution source emission characteristics, and control the mobile inspection equipment to perform multiple rounds of supplementary inspection data analysis based on the iterative analysis results until at least one pollution trajectory meets the preset convergence condition, and outputs the target pollution trajectory.

[0057] Furthermore, the device is also used to perform the following functions: The convergence condition is that the causal inference confidence of the contaminated trajectory is greater than a first threshold and the trajectory probability is greater than a second threshold.

[0058] Furthermore, the device is also used to perform the following functions: The initial atmospheric survey data is input into the digital twin environment, and an atmospheric diffusion-driven branch is loaded. The spatiotemporal diffusion process of pollutants after release from potential source locations is simulated in the digital twin environment to construct a diffusion simulation dataset. Using a Bayesian inversion method, the simulated concentration distribution in the diffusion simulation dataset and the measured concentration distribution in the initial atmospheric survey data are combined to infer the posterior probability distribution of the pollution source location and release time. Based on the posterior probability distribution, multiple candidate pollution trajectories are generated in the digital twin environment through pollution source sampling, and the initial probability distribution is calculated.

[0059] Furthermore, the device is also used to perform the following functions: Multiple samples are extracted from the posterior probability distribution using the Monte Carlo sampling method, each sample corresponding to a pollution source parameter, and a corresponding pollution trajectory is generated based on the digital twin environment; the degree of fit between each pollution trajectory and the initial atmospheric survey data is calculated, and an initial probability is assigned to each trajectory according to the degree of fit, thus forming the multiple candidate pollution trajectories and the initial probability distribution.

[0060] Furthermore, the device is also used to perform the following functions: For each candidate pollution trajectory, a virtual intervention experiment is conducted using meteorological time-series data, geographical blocking factors, and pollution source emission inputs to obtain the causal confidence score for each pollution trajectory. Based on the causal confidence score, in a digital twin environment, the expected growth rate of the causal confidence score for different inspection paths is simulated, and the path with the largest growth rate is selected as the next round of inspection path. The mobile inspection device is controlled to conduct supplementary inspections based on the next round of inspection path, and the updated causal confidence score and updated probability distribution for each pollution trajectory are recalculated based on the supplementary inspection and initial atmospheric inspection data. It is determined whether the updated causal confidence score and updated probability distribution satisfy the convergence condition; if not, the inspection path iteration is repeated.

[0061] Furthermore, the device is also used to perform the following functions: By removing the hypothetical pollution source nodes and rerunning the atmospheric diffusion-driven branch, we observed the concentration changes at monitoring points, calculated the intervention consistency index, and obtained the causal confidence of each pollution trajectory.

[0062] Furthermore, the device is also used to perform the following functions: Based on the first-round update causal confidence and the first-round update probability distribution, taboo trajectories with update causal confidence greater than the first threshold but trajectory probability less than the second threshold are extracted; after deleting the taboo trajectories from the multiple candidate contaminated trajectories, the inspection path iteration is repeatedly performed for the remaining trajectories.

[0063] Furthermore, the device is also used to perform the following functions: If the number of iterations reaches the preset maximum number of iterations, and there is no contaminated trajectory that satisfies the preset convergence condition, the trajectory with the highest current causal confidence and trajectory probability is selected as the target contaminated trajectory.

[0064] Example 3, Figure 3 This is a schematic diagram of the electronic device provided by the digital twin-based survey data analysis method of the present invention, showing an exemplary electronic device suitable for implementing the embodiments of the present invention. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality or scope of the embodiments of the present invention. Figure 3As shown, the electronic device includes a processor 21, a memory 22, an input device 23, and an output device 24; the number of processors 21 in the electronic device can be one or more. Figure 3 Taking a processor 21 as an example, the processor 21, memory 22, input device 23, and output device 24 in an electronic device can be connected via a bus or other means. Figure 3 Taking the example of a connection between China and Israel via a bus.

[0065] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0066] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0067] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A survey data analysis method based on digital twins, characterized in that, include: Construct a digital twin environment for the pre-defined area, including three-dimensional geographic information, meteorological field data, fixed monitoring station data, and a pollution source inventory; Acquire initial atmospheric survey data of the mobile inspection equipment in the preset area; When the initial atmospheric survey data shows anomalies, based on the initial atmospheric survey data and the digital twin environment, a probability inversion of atmospheric diffusion is performed to generate multiple candidate pollution trajectories and their initial probability distributions. For the multiple candidate pollution trajectories, combined with meteorological time series data, geographical blocking factors and pollution source emission characteristics, a multi-round iterative analysis of the inspection path based on maximizing the confidence of causal inference is performed. The mobile inspection equipment is controlled to perform multiple rounds of supplementary inspection data analysis based on the iterative analysis results until at least one pollution trajectory meets the preset convergence condition, and the target pollution trajectory is output.

2. The survey data analysis method based on digital twin as described in claim 1, characterized in that, The convergence condition is that the causal inference confidence of the contaminated trajectory is greater than a first threshold and the trajectory probability is greater than a second threshold.

3. The survey data analysis method based on digital twin as described in claim 1, characterized in that, Based on the initial atmospheric survey data and the digital twin environment, a probability inversion of atmospheric diffusion is performed to generate multiple candidate pollution trajectories and their initial probability distributions, including: The initial atmospheric survey data is input into the digital twin environment, the atmospheric diffusion driving branch is loaded, and the spatiotemporal diffusion process of pollutants after release from potential source locations is simulated in the digital twin environment to construct a diffusion simulation dataset; By employing the Bayesian inversion method and combining the simulated concentration distribution in the diffusion simulation dataset with the measured concentration distribution in the initial atmospheric survey data, the posterior probability distribution of the pollution source location and release time is inferred. Based on the posterior probability distribution, multiple candidate pollution trajectories are generated in the digital twin environment by sampling pollution sources, and the initial probability distribution is calculated.

4. The survey data analysis method based on digital twin as described in claim 3, characterized in that, Based on the posterior probability distribution, multiple candidate pollution trajectories are generated in the digital twin environment through pollution source sampling, and the initial probability distribution is calculated, including: Multiple samples are extracted from the posterior probability distribution using the Monte Carlo sampling method, each sample corresponding to a pollution source parameter, and a corresponding pollution trajectory is generated based on the digital twin environment; The degree of agreement between each pollution trajectory and the initial atmospheric survey data is calculated, and an initial probability is assigned to each trajectory based on the degree of agreement, thus forming the multiple candidate pollution trajectories and the initial probability distribution.

5. The survey data analysis method based on digital twin as described in claim 1, characterized in that, For the multiple candidate pollution trajectories, combining meteorological time-series data, geographical blocking factors, and pollution source emission characteristics, a multi-round iterative analysis of the inspection path based on maximizing the confidence of causal inference is performed. The mobile inspection equipment is controlled to perform multiple rounds of supplementary inspection data analysis based on the iterative analysis results until at least one pollution trajectory meets the preset convergence condition, and the target pollution trajectory is output, including: For each candidate pollution trajectory, a virtual intervention experiment was conducted using meteorological time-series data, geographical blocking factors, and pollution source emission inputs to obtain the causal confidence level of each pollution trajectory. Based on the causal confidence level, in the digital twin environment, the expected growth rate of the causal confidence level for different inspection paths is simulated, and the path with the largest growth rate is selected as the next round of inspection path. The mobile inspection equipment is controlled to conduct supplementary inspections according to the next round of inspection path, and the updated causal confidence and updated probability distribution of each pollution trajectory are recalculated based on the supplementary inspections and the initial atmospheric inspection data. Determine whether the causal confidence score and probability distribution of the first round of updates satisfy the convergence condition. If not, repeat the iterative path survey.

6. The survey data analysis method based on digital twin as described in claim 5, characterized in that, For each candidate pollution trajectory, a virtual intervention experiment was conducted in a digital twin environment using meteorological time-series data, geographical blocking factors, and pollution source emission characteristics to obtain the causal confidence score for each pollution trajectory, including: By removing the hypothetical pollution source nodes and rerunning the atmospheric diffusion-driven branch, we observed the concentration changes at monitoring points, calculated the intervention consistency index, and obtained the causal confidence of each pollution trajectory.

7. The survey data analysis method based on digital twin as described in claim 5, characterized in that, Repeatedly execute the iterative survey path, including: Based on the first-round update causal confidence and the first-round update probability distribution, extract taboo trajectories whose update causal confidence is greater than the first threshold but whose trajectory probability is less than the second threshold; After removing the forbidden trajectory from the multiple candidate contaminated trajectories, the survey path iteration is repeated for the remaining trajectories.

8. The survey data analysis method based on digital twin as described in claim 1, characterized in that, If the number of iterations reaches the preset maximum number of iterations, and there is no contaminated trajectory that satisfies the preset convergence condition, the trajectory with the highest current causal confidence and trajectory probability is selected as the target contaminated trajectory.

9. A survey data analysis device based on digital twins, characterized in that, The apparatus is used to implement the survey data analysis method based on digital twins as described in any one of claims 1-8, and the apparatus comprises: The digital twin environment construction module is used to construct a digital twin environment for a preset area, including three-dimensional geographic information, meteorological field data, fixed monitoring station data, and a pollution source list; The atmospheric survey data acquisition module is used to acquire the initial atmospheric survey data of the mobile inspection equipment in the preset area; The probability inversion module is used to perform probability inversion of atmospheric diffusion based on the initial atmospheric survey data and the digital twin environment when the initial atmospheric survey data shows anomalies, and generate multiple candidate pollution trajectories and their initial probability distributions. The target pollution trajectory output module is used to perform multi-round iterative analysis of the multiple candidate pollution trajectories, combining meteorological time series data, geographical blocking factors and pollution source emission characteristics, and to control the mobile inspection equipment to perform multiple rounds of supplementary inspection data analysis based on the iterative analysis results, until at least one pollution trajectory meets the preset convergence condition, and outputs the target pollution trajectory.

10. An electronic device, characterized in that, The electronic device includes: processor; Memory used to store the processor's executable instructions; The processor is used to execute the survey data analysis method based on digital twins as described in any one of claims 1 to 8.

Images