Ground subsidence risk prediction method and system

Abstract

The invention relates to the technical field of land subsidence prediction, and discloses a land subsidence risk prediction method and system. The method comprises the steps of dividing a to-be-detected ground into a plurality of to-be-detected areas, obtaining correlation result data of target underground structure data and rainfall intensity data by adopting an Eclat algorithm, obtaining a rainfall predicted settlement value by adopting an SARIMA model based on the correlation result data, and obtaining a rainfall prediction settlement value according to the rainfall intensity data of a plurality of historical years and all historical settlement data. And drawing a time sequence diagram of rainfall intensity and land subsidence, determining a subsidence similar data set through a clustering algorithm, optimizing a rainfall predicted subsidence value based on a subsidence optimization factor, and obtaining a disaster subsidence trend between historical years and subsidence by adopting a random forest model. And determining the settlement level of the to-be-detected ground according to the target settlement value and the disaster settlement trend. According to the method, the rainfall data and the disaster data are integrated to comprehensively predict the land subsidence, and the method has good prediction accuracy and comprehensiveness.

Description

Technical Field

[0001] The present invention relates to the technical field of ground subsidence prediction, and in particular to a ground subsidence risk prediction method and system. Background Art

[0002] Land subsidence refers to the phenomenon of the ground sinking or sinking due to underground soil, rock layers or buildings. Land subsidence not only affects the structural safety of buildings, but also may have a significant impact on infrastructure such as transportation, water supply and power supply systems, and even trigger a series of geological disasters such as collapse and cracks. Therefore, accurately predicting the trend and extent of land subsidence is of great significance for disaster prevention and engineering construction.

[0003] Precipitation and disasters are the two main factors affecting ground subsidence. Changes in precipitation directly affect soil moisture and compaction, which in turn leads to ground subsidence. When there is precipitation, the soil absorbs water and expands, causing local subsidence. In addition, natural disasters can also aggravate ground subsidence. In arid environments, the soil will shrink, leading to ground subsidence, while floods and dust storms can also cause varying degrees of ground subsidence. However, due to the randomness and uncertainty of the occurrence of these disasters, current prediction technology cannot effectively predict ground subsidence using disaster data and precipitation data, and cannot comprehensively consider the impact on ground subsidence.

[0004] Therefore, it is necessary to design a ground subsidence hazard prediction method and system to solve the problems existing in current technology. Summary of the invention

[0005] In view of this, the present invention proposes a method and system for predicting ground subsidence danger, aiming to solve the problem that precipitation data and disaster data cannot be used to comprehensively predict ground subsidence.

[0006] In one aspect, the present invention provides a method for predicting land subsidence risk, comprising: The ground to be inspected is divided into several areas to be inspected, and the precipitation intensity data of several historical years of each area to be inspected are obtained based on big data. The initial underground structure data of each area to be inspected is obtained by using geological radar, and the initial underground structure data is preprocessed, and the target underground structure data is obtained based on the preprocessing results; The Eclat algorithm is used to obtain correlation result data between the target underground structure data and the precipitation intensity data, and the SARIMA model is used to obtain the precipitation predicted settlement value based on the correlation result data; Based on big data, all historical settlement data of each area to be detected are obtained. According to the precipitation intensity data of several historical years and all historical settlement data, a time series diagram of precipitation intensity and ground settlement is drawn. A settlement-similar data set is determined through a clustering algorithm. According to the settlement-similar data set, a settlement optimization factor is extracted. Based on the settlement optimization factor, the precipitation predicted settlement value is optimized to obtain a target settlement value. Obtain multiple historical disaster data and corresponding historical years in the local area, calculate the amount of ground settlement caused by the disaster based on the historical disaster data, and use a random forest model to derive the disaster settlement trend between the historical years and the settlement; The settlement level of the ground to be detected is determined according to the target settlement value and the disaster settlement trend.

[0007] Furthermore, when the initial underground structure data of each area to be detected is obtained by using geological radar, the initial underground structure data is preprocessed, including: Set the ground-penetrating radar frequency in each area to be inspected; Move along a preset route in each area to be inspected, and collect initial underground structure data of each area to be inspected at different frequencies; The initial underground structure data includes soil data and underground crack data; The initial underground structure data is preprocessed, and the preprocessing includes data cleaning and standardization.

[0008] Furthermore, when the Eclat algorithm is used to obtain the correlation result data of the target underground structure data and the precipitation intensity data, it includes: The Eclat algorithm is used to set the target underground structure data as the first transaction, and the precipitation intensity data as the second transaction. The intersection of the first transaction and the second transaction is found through an intersection operation to obtain an initial frequent item set, and the initial frequent item set is recursively intersected to obtain the frequent item sets of the target underground structure data and the precipitation intensity data. The correlation result data of the target underground structure data and the precipitation intensity data is determined according to the frequent item sets.

[0009] Further, when the SARIMA model is used to obtain the precipitation prediction settlement value based on the correlation result data, it includes: Dividing the correlation result data into a training set and a test set; Using cross validation combined with grid search, finding the optimal parameters of the SARIMA model to be established, using the training set to fit the SARIMA model to be established, and obtaining the SARIMA model; Substituting the test set into the SARIMA model and calculating the accuracy of precipitation deposition values; When the accuracy reaches a preset accuracy threshold, the precipitation prediction settlement value is obtained according to the correlation result data.

[0010] Furthermore, when all historical settlement data of each area to be detected are obtained based on big data, a time series diagram of precipitation intensity and ground settlement is drawn according to precipitation intensity data of several historical years and all historical settlement data, including: Obtaining precipitation intensity data and all historical settlement data for several historical years, the selected time periods correspond one to one, the x-axis of the time series graph represents the year, and the y-axis of the time series graph represents the number; The time series graph includes a precipitation intensity sequence graph and a ground subsidence sequence graph. The precipitation intensity sequence graph is obtained based on precipitation intensity data of several historical years, and the ground subsidence sequence graph is obtained based on all historical subsidence data. The precipitation intensity sequence graph is a bar graph, and the ground subsidence sequence graph is a line graph.

[0011] Furthermore, when a similar settlement data set is determined by a clustering algorithm, a settlement optimization factor is extracted according to the similar settlement data set, and the precipitation predicted settlement value is optimized based on the settlement optimization factor to obtain a target settlement value, the method includes: The precipitation intensity in the precipitation intensity sequence diagram and the ground settlement in the ground settlement sequence diagram are used as the data sets to be aggregated, the expected number of clusters k is determined to be 3, and the parameters of the Gaussian distribution are initialized, the probability that each data in the data sets to be aggregated belongs to each Gaussian distribution is calculated and the responsibility value is obtained, and the data sets corresponding to the precipitation intensity and the ground settlement are obtained according to the responsibility value, and the data sets are used as the settlement-similar data sets; The settlement optimization factor is obtained by taking the average of all the data in the settlement similar data set, and the target settlement value is the product of the settlement optimization factor and the precipitation predicted settlement value; Presetting a first sedimentation threshold and a second sedimentation threshold; When the target settlement value is greater than or equal to the first settlement threshold, the target settlement value is judged to be severe settlement; When the target sedimentation value is less than the first sedimentation threshold and greater than or equal to the second sedimentation threshold, the target sedimentation value is judged to be medium sedimentation; When the target settlement value is less than the second settlement threshold, the target settlement value is determined to be slight settlement.

[0012] Furthermore, when a plurality of historical disaster data and corresponding historical years of a local area are obtained and the amount of ground settlement caused by the disaster is calculated according to the historical disaster data, it includes: The historical disaster data include flood data, drought data and dust storm data; The settlement is given by the following formula: ; in, Indicates the amount of sedimentation, The water velocity of the flood, represents the sediment concentration in the flood, Indicates the duration of the flood, represents the average diameter of sediment particles deposited on the ground, represents the duration of drought, Indicates duration The amount of groundwater extracted from represents the baseline value for groundwater extraction, Indicates the soil type of the ground. represents the dust storm wind speed, Indicates the duration of the dust storm.

[0013] Furthermore, when the random forest model is used to obtain the disaster settlement trend between the historical years and the settlement amount, it includes: The historical years are arranged in chronological order into a disaster series, the features of the disaster series are extracted and substituted into the random forest model, and the disaster settlement trend is predicted based on the settlement amount, and the disaster settlement trend includes an upward trend, a downward trend and a stable trend.

[0014] Further, when determining the settlement level of the ground to be detected according to the target settlement value and the disaster settlement trend, it includes: The settlement levels include a first predicted settlement level, a second predicted settlement level, and a third predicted settlement level; When it is determined that the target settlement value is severe settlement and the disaster settlement is in an upward trend or a stable trend, it is determined as the first predicted settlement level; When it is judged that the target settlement value is the severe settlement and the disaster settlement trend is the downward trend, it is determined to be the second predicted settlement level; or, when it is judged that the target settlement value is the moderate settlement and the disaster settlement trend is the upward trend or the stable trend, it is determined to be the second predicted settlement level; or, when it is judged that the target settlement value is the slight settlement and the disaster settlement trend is the upward trend, it is determined to be the second predicted settlement level; When it is judged that the target settlement value is the medium settlement and the disaster settlement trend is the downward trend, it is determined to be the third predicted settlement level; or, when it is judged that the target settlement value is the slight settlement and the disaster settlement trend is the downward trend or the stable trend, it is determined to be the third predicted settlement level.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: by dividing the area to be detected into multiple subdivided areas, and obtaining historical precipitation intensity and settlement data based on big data, combined with the underground structure data of geological radar, the settlement risk of each area can be accurately analyzed, and the use of the Eclat algorithm and the SARIMA model can accurately predict the impact of precipitation on ground settlement, thereby improving the reliability and accuracy of the prediction results, and using the time series graph drawn by historical precipitation intensity and settlement data, and the settlement optimization factor extracted by the settlement similar data set determined by the clustering algorithm, the settlement optimization factor is used to optimize the precipitation prediction settlement value, thereby obtaining an accurate target settlement value, avoiding the error caused by a single prediction model, and improving the stability and accuracy of the prediction. By introducing historical disaster data and using the random forest model to analyze the impact of disasters on ground settlement, the potential threat of factors such as natural disasters to ground settlement can be taken into account. The prediction of disaster settlement trends provides comprehensive data for assessing the severity of ground settlement. According to the target settlement value and disaster settlement trend, the accurate division of settlement levels can be achieved, the settlement danger of the ground to be detected can be clearly understood, and the accuracy of the prediction is improved.

[0016] On the other hand, the present application also provides a ground subsidence risk prediction system for applying the above-mentioned ground subsidence risk prediction method, comprising: Data acquisition module, settlement prediction module, settlement correction module, disaster settlement module and settlement level module; The data acquisition module is configured to divide the ground to be detected into a number of areas to be detected, obtain precipitation intensity data of a number of historical years for each area to be detected based on big data, use geological radar to obtain initial underground structure data for each area to be detected, pre-process the initial underground structure data, and obtain target underground structure data based on the pre-processing results; The settlement prediction module is configured to use an Eclat algorithm to obtain correlation result data between the target underground structure data and the precipitation intensity data, and use a SARIMA model based on the correlation result data to obtain a precipitation predicted settlement value; The settlement correction module is configured to obtain all historical settlement data of each area to be detected based on big data, draw a time series diagram of precipitation intensity and ground settlement according to precipitation intensity data of several historical years and all historical settlement data, determine a settlement similar data set through a clustering algorithm, extract a settlement optimization factor according to the settlement similar data set, optimize the precipitation predicted settlement value based on the settlement optimization factor, and obtain a target settlement value; The disaster settlement module is configured to obtain a plurality of historical disaster data and corresponding historical years of the local area, calculate the settlement amount caused by the disaster on the ground according to the historical disaster data, and use a random forest model to obtain the disaster settlement trend between the historical years and the settlement amount; The settlement level module is configured to determine the settlement level of the ground to be detected according to the target settlement value and the disaster settlement trend.

[0017] It is understandable that the above-mentioned method and system for predicting ground subsidence danger have the same beneficial effects, which will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings: Figure 1 A flow chart of a method for predicting land subsidence risk provided by an embodiment of the present invention; Figure 2 A functional block diagram of a ground subsidence hazard prediction system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0019] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features described in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0020] In some embodiments of the present application, see Figure 1 As shown, a method for predicting land subsidence danger includes: S100: Divide the ground to be inspected into several areas to be inspected, obtain precipitation intensity data of several historical years for each area to be inspected based on big data, use geological radar to obtain initial underground structure data for each area to be inspected, pre-process the initial underground structure data, and obtain target underground structure data based on the pre-processing results; S200: Using the Eclat algorithm to obtain the correlation result data of the target underground structure data and the precipitation intensity data, and using the SARIMA model based on the correlation result data to obtain the precipitation predicted settlement value; S300: Based on big data, all historical settlement data of each area to be detected are obtained. According to the precipitation intensity data of several historical years and all historical settlement data, a time series diagram of precipitation intensity and ground settlement is drawn. A settlement-similar data set is determined through a clustering algorithm. According to the settlement-similar data set, a settlement optimization factor is extracted. Based on the settlement optimization factor, the precipitation prediction settlement value is optimized to obtain a target settlement value. S400: obtaining multiple historical disaster data and corresponding historical years in the local area, calculating the amount of ground settlement caused by the disaster based on the historical disaster data, and using a random forest model to derive the disaster settlement trend between the historical years and the settlement amount; S500: Determine the settlement level of the ground to be detected according to the target settlement value and the disaster settlement trend.

[0021] Specifically, the ground to be detected is divided into several areas to be detected, and the number of areas to be detected is preferably 20. Based on big data technology, the precipitation intensity data of several historical years for each area to be detected is obtained. The several historical years are preferably the past 5 years of the current year, including the past 5 years. Through the big data platform, the precipitation intensity data of the past 5 years are collected to ensure that the time span of the data is long enough to capture the potential relationship between precipitation and settlement in the future. The geological radar can provide comprehensive soil data and underground crack data under the ground. Lay the data foundation for subsequent processing. The Eclat algorithm is a frequent item set mining algorithm. By analyzing the frequent item sets in the data, it can reveal the potential relationship between precipitation intensity data and target underground structure data. Once the correlation result data between precipitation intensity data and target underground structure data is obtained, the next step is to use the SARIMA model based on these correlation data to obtain the precipitation predicted settlement value. The SARIMA model is a prediction model for time series analysis. By combining the SARIMA model with correlation data, the precipitation predicted settlement value can be obtained. A prediction result based on precipitation intensity data is provided for the prediction of ground subsidence. Big data technology is used again to obtain all historical subsidence data for each area to be tested. The historical subsidence data includes the actual subsidence in the past five years, including the fifth year. By combining the precipitation intensity data and historical subsidence data of historical years, a time series diagram between precipitation intensity and ground subsidence is drawn. This series diagram can show the correlation between the fluctuation of precipitation intensity and the change of subsidence, providing an intuitive basis for the optimization of subsequent precipitation prediction subsidence values. Based on time series data, a clustering algorithm is used to group historical subsidence data, and data with similar subsidence trends are classified into one category. The subsidence data sets with similar subsidence are determined and subsidence optimization factors are extracted. Through in-depth analysis of historical data, the prediction results are adjusted to improve the accuracy of the target subsidence value.

[0022] It is understandable that regional disasters can also cause ground subsidence. Based on historical disaster data, the amount of ground subsidence caused by disaster events each year is calculated. By analyzing the amount of subsidence before and after the disaster, the random forest model can be used to quantitatively evaluate the disaster subsidence trend of ground subsidence caused by disaster events each year. Based on the target subsidence value and disaster subsidence trend obtained from big data and models, combined with the derived subsidence level, the subsidence risk of the ground to be tested can be comprehensively predicted, which improves the accuracy and scientificity of hazard prediction.

[0023] In some embodiments of the present application, when using geological radar to obtain the initial underground structure data of each area to be detected and preprocessing the initial underground structure data, it includes: setting the geological radar frequency in each area to be detected, moving along a preset route in each area to be detected, and collecting the initial underground structure data of each area to be detected at different frequencies, the initial underground structure data including soil data and underground crack data, and preprocessing the initial underground structure data, and the preprocessing includes data cleaning and standardization.

[0024] It is understandable that two frequencies are set according to the geological conditions of the area to be detected. High-frequency radar waves can detect deeper underground structures below the area to be detected and obtain underground crack data. Low-frequency radar waves can detect shallower underground structures below the area to be detected, thereby obtaining soil data. Moving along the center of the area to be detected to the edge of the area to be detected can effectively obtain underground crack data in the area to be detected, thereby improving the accuracy of subsequent predictions. Preprocessing can ensure the accuracy and quality of the target underground structure data. Data cleaning removes erroneous or incomplete data in the initial underground structure data, and then performs data standardization processing on it to ensure the consistency of the target underground structure data format for subsequent processing and analysis.

[0025] In some embodiments of the present application, when the Eclat algorithm is used to obtain the correlation result data of the target underground structure data and the precipitation intensity data, it includes: using the Eclat algorithm to set the target underground structure data as the first transaction, setting the precipitation intensity data as the second transaction, finding the intersection of the first transaction and the second transaction through an intersection operation to obtain an initial frequent item set, and recursively performing an intersection operation on the initial frequent item set to obtain a frequent item set of the target underground structure data and the precipitation intensity data, and determining the correlation result data of the target underground structure data and the precipitation intensity data according to the frequent item set.

[0026] It can be understood that the Eclat algorithm is an algorithm for frequent item set mining, which finds frequent item sets in data transactions through operations based on set intersection. In order to deeply analyze the relationship between the target underground structure data and the precipitation intensity data, the Eclat algorithm is used to set the target underground structure data and the precipitation intensity data as the first transaction and the second transaction respectively, and find the initial frequent item set between the two through intersection operations. First, the target underground structure data is taken as the first transaction, and the precipitation intensity data is taken as the second transaction. The intersection of the first transaction and the second transaction is obtained by intersection. For example, when the precipitation is 400mm, underground cracks in sandy soil frequently extend by 0.5mm. Thus, the initial frequent item set is obtained. After determining the preliminary frequent item set, the Eclat algorithm will further perform intersection operations on the obtained initial frequent item set in a recursive manner. By continuously performing recursive operations, high-order frequent item sets can be screened out. Compared with traditional algorithms, the Eclat algorithm improves the mining efficiency of frequent item sets through intersection operations. It can quickly extract effective information from a large amount of data and reveal the potential correlation between the target underground structure data and precipitation intensity data, and obtain correlation result data, providing a scientific basis for subsequent further analysis and prediction.

[0027] In some embodiments of the present application, when the SARIMA model is used to obtain the precipitation prediction sedimentation value based on the correlation result data, it includes: dividing the correlation result data into a training set and a test set, using cross-validation and combining grid search to find the optimal parameters of the SARIMA model to be established, using the training set to fit the SARIMA model to be established to obtain the SARIMA model, substituting the test set into the SARIMA model and calculating the accuracy of the precipitation precipitation value, when the accuracy reaches a preset accuracy threshold, obtaining the precipitation prediction sedimentation value according to the correlation result data.

[0028] It is understandable that after obtaining the correlation result data, the correlation result data is divided into a training set and a test set. 50% to 70% of the correlation result data is used as the training set, and the rest of the data is used as the test set. It is ensured that both the training set and the test set contain a variety of data to improve the generalization ability of the SARIMA model, and cross-validation and grid search are used to find the best parameters of the SARIMA model. Cross-validation divides the data into several parts and trains the SARIMA model multiple times to ensure the stability and reliability of the obtained SARIMA model. In the absence of suitable hyperparameters, the SARIMA model may be overfitted or underfitted. Overfitting means that the model performs well on the training set but performs poorly on the test set, while underfitting means that the model cannot capture the complex patterns in the data. Finding suitable hyperparameters through grid search can help improve the generalization ability of the model and avoid too large performance differences between the training set and the test set. After reaching the preset accuracy threshold, the SARIMA model can predict the correlation result data in real time and obtain the precipitation prediction settlement value, thereby providing efficient and accurate prediction of the impact of precipitation on ground settlement. Improved the accuracy of ground subsidence hazard prediction.

[0029] In some embodiments of the present application, when all historical settlement data of each area to be detected is obtained based on big data, and a time series graph of precipitation intensity and ground settlement is drawn according to the precipitation intensity data of several historical years and all historical settlement data, it includes: obtaining the precipitation intensity data of several historical years and all historical settlement data, the selected time periods correspond one to one, the x-axis of the time series graph represents the year, the y-axis of the time series graph represents the number, the time series graph includes a precipitation intensity sequence graph and a ground settlement sequence graph, a precipitation intensity sequence graph is obtained based on the precipitation intensity data of several historical years, and a ground settlement sequence graph is obtained based on all historical settlement data, the precipitation intensity sequence graph is a bar graph, and the ground settlement sequence graph is a line graph.

[0030] It is understandable that all historical settlement data of each area to be detected are obtained again based on the big data platform. All historical settlement data are also data from the past 5 years, including the 5th year. The historical settlement data corresponds to the selected past time period one by one to ensure that the precipitation intensity data and historical settlement data are at the same time point. The precipitation intensity data and ground settlement data are presented in the form of a time series graph, which provides a visual image for subsequent prediction and analysis of ground settlement. The x-axis of the time series graph is arranged by year, with each year as 1 unit, such as: 2021 and 2022, which intuitively shows the past 5 years, while the y-axis is arranged in numbers, based on 100mm and 1mm as units, representing the actual values ​​of precipitation intensity and ground settlement, respectively. The precipitation intensity sequence graph uses a bar graph to facilitate the display of precipitation fluctuations in different years, and the ground settlement sequence graph uses a line graph to show the trend of settlement changes. Through the combination of bar graphs and line graphs, the changing trends of precipitation intensity and ground settlement can be clearly displayed, providing data support for subsequent clustering analysis.

[0031] In some embodiments of the present application, when a similar settlement data set is determined by a clustering algorithm, a settlement optimization factor is extracted according to the similar settlement data set, and the precipitation predicted settlement value is optimized based on the settlement optimization factor to obtain the target settlement value, the method includes: taking the precipitation intensity in the precipitation intensity sequence diagram and the ground settlement in the ground settlement sequence diagram as the data set to be aggregated, determining the expected number of clusters k to be 3, and initializing the parameters of the Gaussian distribution, calculating the probability that each data in the data set to be aggregated belongs to each Gaussian distribution and obtaining the responsibility value, and obtaining the data corresponding to the precipitation intensity and the ground settlement according to the responsibility value. The data set is taken as a settlement similar data set, and the settlement optimization factor is obtained by averaging all the data in the settlement similar data set. The target settlement value is the product of the settlement optimization factor and the precipitation predicted settlement value. The first settlement threshold and the second settlement threshold are set in advance. When the target settlement value is greater than or equal to the first settlement threshold, the target settlement value is judged to be severe settlement. When the target settlement value is less than the first settlement threshold and greater than or equal to the second settlement threshold, the target settlement value is judged to be moderate settlement. When the target settlement value is less than the second settlement threshold, the target settlement value is judged to be slight settlement.

[0032] It is understandable that by analyzing the precipitation intensity data and historical settlement data through clustering algorithms, close settlement data sets are found, and all data in the settlement data sets are averaged to obtain the settlement optimization factor. The settlement optimization factor obtained is derived from a variety of data, which improves the compatibility and accuracy of the settlement optimization factor. In addition, the optimization mechanism of the settlement optimization factor for the precipitation prediction settlement value improves the accuracy of the target settlement value, reduces the error caused by manually obtained data and model prediction, and improves the reliability of ground settlement prediction. The first settlement threshold is greater than the second settlement threshold. Real-time comparison is performed according to the target settlement value, which can preliminarily predict the ground settlement situation and lay a theoretical basis for subsequent processing.

[0033] In some embodiments of the present application, when obtaining a plurality of historical disaster data and corresponding historical years of a local area, and calculating the amount of ground settlement caused by the disaster according to the historical disaster data, the historical disaster data includes flood data, drought data, and dust storm data, and the settlement amount is obtained by the following formula: ; in, Indicates the amount of sedimentation, The water velocity of the flood, represents the sediment concentration in the flood, Indicates the duration of the flood, represents the average diameter of sediment particles deposited on the ground, represents the duration of drought, Indicates duration The amount of groundwater extracted from indicates the baseline value for groundwater extraction, Indicates the soil type of the ground. Indicates the dust storm wind speed, Indicates the duration of the dust storm.

[0034] It is understandable that obtaining multiple historical years in the local area reflects the comprehensiveness of disaster data for ground subsidence prediction. The historical years are preferably the past 10 years, including the 10th year. Soil types are represented as sandy soil, clay soil and loam. The value of sandy soil is preferably 1.5, the value of clay soil is preferably 1, and the value of loam is preferably 0.8. Disaster data are data on floods, droughts and dust storms in the local area. By comprehensively analyzing the impact of different disaster types on ground subsidence, the reliability and comprehensiveness of ground subsidence prediction are improved.

[0035] In some embodiments of the present application, when using the random forest model to derive the disaster settlement trend between historical years and settlement amounts, it includes: arranging the historical years into a disaster series in chronological order, extracting the characteristics of the disaster series and substituting them into the random forest model, and predicting the disaster settlement trend based on the settlement amount, the disaster settlement trend includes an upward trend, a downward trend and a stable trend.

[0036] In some embodiments of the present application, when determining the settlement level of the ground to be detected according to the target settlement value and the disaster settlement trend, it includes: the settlement level includes a first predicted settlement level, a second predicted settlement level and a third predicted settlement level. When the target settlement value is judged to be severe settlement and the disaster settlement is an upward trend or a stable trend, it is determined to be the first predicted settlement level. When the target settlement value is judged to be severe settlement and the disaster settlement trend is a downward trend, it is determined to be the second predicted settlement level. Or, when the target settlement value is judged to be moderate settlement and the disaster settlement trend is an upward trend or a stable trend, it is determined to be the second predicted settlement level. Or, when the target settlement value is judged to be slight settlement and the disaster settlement trend is an upward trend, it is determined to be the second predicted settlement level. When the target settlement value is judged to be moderate settlement and the disaster settlement trend is a downward trend, it is determined to be the third predicted settlement level. Or, when the target settlement value is judged to be slight settlement and the disaster settlement trend is a downward trend or a stable trend, it is determined to be the third predicted settlement level.

[0037] It is understandable that the same soil type as the ground to be tested is selected to organize the historical years into a disaster series in chronological order. The series uses the year as a time series input, extracts the time features from the disaster series, takes the extracted features as input, and combines the settlement amount corresponding to each year as input data. It is substituted into the random forest model to generate multiple decision trees and combines the output results, and the output results are trained to predict the disaster settlement trend. The danger level of the first predicted settlement level, the second predicted settlement level, and the third predicted settlement level is reduced. By combining the target settlement value with the disaster settlement trend, the potential ground settlement risk can be accurately predicted. For example: Even if the target settlement value is judged to be a slight settlement, if the disaster settlement trend is an upward trend, it can still be predicted as the second predicted settlement level, avoiding the prediction of a single data and affecting the overall prediction level.

[0038] In summary, the beneficial effects of the present invention are: by dividing the area to be detected into multiple subdivided areas, and obtaining historical precipitation intensity and settlement data based on big data, combined with the underground structure data of geological radar, the settlement risk of each area can be accurately analyzed, and the use of the Eclat algorithm and the SARIMA model can accurately predict the impact of precipitation on ground settlement, thereby improving the reliability and accuracy of the prediction results, and using the time series graph drawn by historical precipitation intensity and settlement data, and the settlement optimization factor extracted by the settlement similar data set determined by the clustering algorithm, the settlement optimization factor is used to optimize the precipitation prediction settlement value, thereby obtaining an accurate target settlement value, avoiding the error caused by a single prediction model, and improving the stability and accuracy of the prediction. By introducing historical disaster data and using the random forest model to analyze the impact of disasters on ground settlement, the potential threat of factors such as natural disasters to ground settlement can be taken into account. The prediction of disaster settlement trends provides comprehensive data for assessing the severity of ground settlement. According to the target settlement value and disaster settlement trend, the accurate division of settlement levels can be achieved, the settlement danger of the ground to be detected can be clearly understood, and the accuracy of the prediction is improved.

[0039] In another preferred embodiment based on the above embodiment, refer to Figure 2 As shown, this embodiment provides a ground subsidence risk prediction system, which is used to apply the above-mentioned ground subsidence risk prediction method, including: Data acquisition module, settlement prediction module, settlement correction module, disaster settlement module and settlement level module; The data acquisition module is configured to divide the ground to be detected into several areas to be detected, obtain the precipitation intensity data of several historical years for each area to be detected based on big data, use geological radar to obtain the initial underground structure data of each area to be detected, pre-process the initial underground structure data, and obtain the target underground structure data based on the pre-processing results; The settlement prediction module is configured to use the Eclat algorithm to obtain correlation result data of the target underground structure data and the precipitation intensity data, and use the SARIMA model to obtain the precipitation prediction settlement value based on the correlation result data; The settlement correction module is configured to obtain all historical settlement data of each area to be detected based on big data, draw a time series diagram of precipitation intensity and ground settlement based on precipitation intensity data of several historical years and all historical settlement data, determine similar settlement data sets through clustering algorithms, extract settlement optimization factors based on similar settlement data sets, optimize precipitation predicted settlement values ​​based on settlement optimization factors, and obtain target settlement values; The disaster settlement module is configured to obtain multiple historical disaster data and corresponding historical years in the local area, calculate the settlement amount caused by the disaster on the ground based on the historical disaster data, and use the random forest model to obtain the disaster settlement trend between the historical years and the settlement amount; The settlement level module is configured to determine the settlement level of the ground to be detected according to the target settlement value and the disaster settlement trend.

[0040] It can be understood that each module in a ground subsidence hazard prediction system works together to achieve accurate prediction and effective correction of ground subsidence. The data acquisition module first divides the ground to be detected into multiple areas, and collects the historical precipitation intensity data of each area through big data technology. The initial underground structure data of each area to be detected is obtained using geological radar technology, and these data are preprocessed to obtain the target underground structure data, which lays a data foundation for subsequent analysis. The settlement prediction module uses the Eclat algorithm to calculate the correlation between the target underground structure data and the precipitation intensity data to obtain the correlation result data. Based on this data, the SARIMA model is used to calculate the precipitation predicted settlement value, which provides a prediction basis for subsequent settlement analysis. After obtaining the precipitation predicted settlement value, the settlement correction module draws a time series diagram of precipitation intensity and ground subsidence by combining historical settlement data and precipitation intensity data, clusters the data using a clustering algorithm to avoid extreme data, and extracts settlement optimization factors to optimize the precipitation predicted settlement value. Based on the extraction of multiple data, the historical settlement data and precipitation intensity data are comprehensively analyzed to improve the accuracy of the target settlement value. The disaster settlement module obtains historical disaster data of the local area, and combines it with the corresponding years, calculates the settlement caused by the disaster, and uses the random forest model to analyze the disaster settlement trend between historical years and settlement, reflecting the impact of disasters on ground settlement and improving the comprehensiveness and accuracy of ground settlement risk prediction. The settlement level module determines the settlement level of the ground to be tested based on the target settlement value and the disaster settlement trend, avoiding the impact of single data on the accuracy of the prediction and improving the accuracy of the prediction.

[0041] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0042] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0043] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0044] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0045] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting land subsidence risk, characterized in that: include: The ground to be inspected is divided into several areas to be inspected, and the precipitation intensity data of several historical years of each area to be inspected are obtained based on big data. The initial underground structure data of each area to be inspected is obtained by using geological radar, and the initial underground structure data is preprocessed, and the target underground structure data is obtained based on the preprocessing results; The Eclat algorithm is used to obtain correlation result data between the target underground structure data and the precipitation intensity data, and the SARIMA model is used to obtain the precipitation predicted settlement value based on the correlation result data; Based on big data, all historical settlement data of each area to be detected are obtained. According to the precipitation intensity data of several historical years and all historical settlement data, a time series diagram of precipitation intensity and ground settlement is drawn. A settlement-similar data set is determined through a clustering algorithm. According to the settlement-similar data set, a settlement optimization factor is extracted. Based on the settlement optimization factor, the precipitation predicted settlement value is optimized to obtain a target settlement value. Obtain multiple historical disaster data and corresponding historical years in the local area, calculate the amount of ground settlement caused by the disaster based on the historical disaster data, and use a random forest model to derive the disaster settlement trend between the historical years and the settlement; The settlement level of the ground to be detected is determined according to the target settlement value and the disaster settlement trend.

2. The method for predicting land subsidence danger according to claim 1, characterized in that: When the initial underground structure data of each area to be detected is obtained by using geological radar, the initial underground structure data is preprocessed, including: Set the ground-penetrating radar frequency in each area to be inspected; Move along a preset route in each area to be inspected, and collect initial underground structure data of each area to be inspected at different frequencies; The initial underground structure data includes soil data and underground crack data; The initial underground structure data is preprocessed, and the preprocessing includes data cleaning and standardization.

3. The method for predicting land subsidence danger according to claim 2, characterized in that: When the Eclat algorithm is used to obtain the correlation result data of the target underground structure data and the precipitation intensity data, it includes: The Eclat algorithm is used to set the target underground structure data as the first transaction, and the precipitation intensity data as the second transaction. The intersection of the first transaction and the second transaction is found through an intersection operation to obtain an initial frequent item set, and the initial frequent item set is recursively intersected to obtain the frequent item sets of the target underground structure data and the precipitation intensity data. The correlation result data of the target underground structure data and the precipitation intensity data is determined according to the frequent item sets.

4. The method for predicting land subsidence danger according to claim 3, characterized in that: The SARIMA model is used to obtain the precipitation prediction settlement value based on the correlation result data, including: Dividing the correlation result data into a training set and a test set; Using cross validation combined with grid search, finding the optimal parameters of the SARIMA model to be established, using the training set to fit the SARIMA model to be established, and obtaining the SARIMA model; Substituting the test set into the SARIMA model and calculating the accuracy of precipitation deposition values; When the accuracy reaches a preset accuracy threshold, the precipitation prediction settlement value is obtained according to the correlation result data.

5. The method for predicting land subsidence danger according to claim 4, characterized in that: When all historical settlement data of each area to be detected are obtained based on big data, and a time series diagram of precipitation intensity and ground settlement is drawn based on precipitation intensity data of several historical years and all historical settlement data, it includes: Obtaining precipitation intensity data and all historical settlement data for several historical years, the selected time periods correspond one to one, the x-axis of the time series graph represents the year, and the y-axis of the time series graph represents the number; The time series graph includes a precipitation intensity sequence graph and a ground subsidence sequence graph. The precipitation intensity sequence graph is obtained based on precipitation intensity data of several historical years, and the ground subsidence sequence graph is obtained based on all historical subsidence data. The precipitation intensity sequence graph is a bar graph, and the ground subsidence sequence graph is a line graph.

6. The method for predicting land subsidence danger according to claim 5, characterized in that: When a similar settlement data set is determined by a clustering algorithm, a settlement optimization factor is extracted according to the similar settlement data set, and the precipitation predicted settlement value is optimized based on the settlement optimization factor to obtain a target settlement value, the method includes: The precipitation intensity in the precipitation intensity sequence diagram and the ground settlement in the ground settlement sequence diagram are used as the data sets to be aggregated, the expected number of clusters k is determined to be 3, and the parameters of the Gaussian distribution are initialized, the probability that each data in the data sets to be aggregated belongs to each Gaussian distribution is calculated and the responsibility value is obtained, and the data sets corresponding to the precipitation intensity and the ground settlement are obtained according to the responsibility value, and the data sets are used as the settlement-similar data sets; The settlement optimization factor is obtained by taking the average of all the data in the settlement similar data set, and the target settlement value is the product of the settlement optimization factor and the precipitation predicted settlement value; Presetting a first sedimentation threshold and a second sedimentation threshold; When the target settlement value is greater than or equal to the first settlement threshold, the target settlement value is judged to be severe settlement; When the target sedimentation value is less than the first sedimentation threshold and greater than or equal to the second sedimentation threshold, the target sedimentation value is judged to be medium sedimentation; When the target settlement value is less than the second settlement threshold, the target settlement value is determined to be slight settlement.

7. The method for predicting land subsidence danger according to claim 6, characterized in that: When obtaining a plurality of historical disaster data and corresponding historical years of a local area and calculating the amount of ground settlement caused by the disaster according to the historical disaster data, it includes: The historical disaster data include flood data, drought data and dust storm data; The settlement is given by the following formula: ； in, Indicates the amount of sedimentation, The water velocity of the flood, represents the sediment concentration in the flood, Indicates the duration of the flood, represents the average diameter of sediment particles deposited on the ground, represents the duration of drought, Indicates duration The amount of groundwater extracted from represents the baseline value for groundwater extraction, Indicates the soil type of the ground. represents the dust storm wind speed, Indicates the duration of the dust storm.

8. The method for predicting land subsidence danger according to claim 7, characterized in that: When the random forest model is used to derive the disaster settlement trend between the historical years and the settlement amount, it includes: The historical years are arranged in chronological order into a disaster series, the features of the disaster series are extracted and substituted into the random forest model, and the disaster settlement trend is predicted based on the settlement amount, and the disaster settlement trend includes an upward trend, a downward trend and a stable trend.

9. The method for predicting land subsidence danger according to claim 8, characterized in that: When determining the settlement level of the ground to be detected according to the target settlement value and the disaster settlement trend, it includes: The settlement levels include a first predicted settlement level, a second predicted settlement level, and a third predicted settlement level; When it is determined that the target settlement value is severe settlement and the disaster settlement is in an upward trend or a stable trend, it is determined as the first predicted settlement level; When it is judged that the target settlement value is the severe settlement and the disaster settlement trend is the downward trend, it is determined to be the second predicted settlement level; or, when it is judged that the target settlement value is the moderate settlement and the disaster settlement trend is the upward trend or the stable trend, it is determined to be the second predicted settlement level; or, when it is judged that the target settlement value is the slight settlement and the disaster settlement trend is the upward trend, it is determined to be the second predicted settlement level; When it is judged that the target settlement value is the medium settlement and the disaster settlement trend is the downward trend, it is determined to be the third predicted settlement level; or, when it is judged that the target settlement value is the slight settlement and the disaster settlement trend is the downward trend or the stable trend, it is determined to be the third predicted settlement level.

10. A land subsidence risk prediction system, used for applying the land subsidence risk prediction method according to any one of claims 1 to 9, characterized in that: include: Data acquisition module, settlement prediction module, settlement correction module, disaster settlement module and settlement level module; The data acquisition module is configured to divide the ground to be detected into a number of areas to be detected, obtain precipitation intensity data of a number of historical years for each area to be detected based on big data, use geological radar to obtain initial underground structure data for each area to be detected, pre-process the initial underground structure data, and obtain target underground structure data based on the pre-processing results; The settlement prediction module is configured to use an Eclat algorithm to obtain correlation result data between the target underground structure data and the precipitation intensity data, and use a SARIMA model based on the correlation result data to obtain a precipitation predicted settlement value; The settlement correction module is configured to obtain all historical settlement data of each area to be detected based on big data, draw a time series diagram of precipitation intensity and ground settlement according to precipitation intensity data of several historical years and all historical settlement data, determine a settlement similar data set through a clustering algorithm, extract a settlement optimization factor according to the settlement similar data set, optimize the precipitation predicted settlement value based on the settlement optimization factor, and obtain a target settlement value; The disaster settlement module is configured to obtain a plurality of historical disaster data and corresponding historical years of the local area, calculate the settlement amount caused by the disaster on the ground according to the historical disaster data, and use a random forest model to obtain the disaster settlement trend between the historical years and the settlement amount; The settlement level module is configured to determine the settlement level of the ground to be detected according to the target settlement value and the disaster settlement trend.

Images