Geological disaster comprehensive monitoring and forecasting method based on double-satellite fusion

Abstract

The invention belongs to the technical field of geological disaster monitoring, and particularly relates to a geological disaster comprehensive monitoring and forecasting method based on double-satellite fusion, and the method comprises the following steps: developing a Beidou base station, a monitoring station and a high-precision board card, and obtaining the coordinate change data of a monitoring point; a point-line-plane three-dimensional monitoring network is established through networking of BDS / GNSS multiple monitoring points and a base station, Beidou high-precision positioning data and multi-source data are fused, and monitoring point displacement data are obtained; performing quantitative and refined forecasting on short-time approaching heavy rainfall by using an intelligent grid product, and training a heavy rainfall radar evolution process of an optimization model by using a reverse modeling sample; and constructing a multi-dimensional feature vector matrix, extracting important and key features in a multi-factor influence factor data set, and establishing a comprehensive monitoring and forecasting model of the long-short-term memory artificial neural network on the basis of the important and key features. The method can improve the accuracy and timeliness of early warning of geological disasters in a complex geological environment of Guangxi-East Alliance, and has a good market application prospect.

Description

Technical Field

[0001] The present invention belongs to the technical field of geological disaster monitoring, and in particular relates to a method for comprehensive monitoring and forecasting of geological disasters based on binary satellite fusion. Background Art

[0002] Guangxi is located in the central and southern subtropical monsoon climate zone. Heavy rainfall weather processes are frequent during the flood season, which often cause flash floods, rising rivers, and economic and population losses such as flooding of crops, roads, and houses. At present, there are relatively few studies on cloud detection tasks using geostationary meteorological satellites at home and abroad. In addition, the geological structure in Guangxi is relatively complex, and karst landforms are widely distributed. Coupled with the influence of heavy rainfall, geological disasters occur frequently, with hundreds of geological disasters every year, such as landslides, collapses, mudslides, ground collapses, and ground fissures. The direct economic losses caused each year range from tens of millions to hundreds of millions of yuan. For the monitoring, early warning and geological exploration of geological disasters, it is necessary to develop high-precision and highly reliable satellite navigation and positioning technology.

[0003] In order to improve the accuracy of geological disaster monitoring and early warning models, this application combines the advantages of Beidou's high monitoring accuracy and short early warning response time, as well as the rainfall monitoring capabilities of meteorological satellites in specific areas, to provide a comprehensive geological disaster monitoring and forecasting method based on dual-satellite fusion, which greatly extends the geological disaster early warning time and is of great significance to geological disaster monitoring, safety management and emergency response to sudden incidents in the complex geological environment of Guangxi-ASEAN.

[0004] The information disclosed in this background technology section is only intended to enhance the understanding of the overall background of the invention and should not be regarded as an acknowledgment or any form of suggestion that the information constitutes the prior art already known to a person skilled in the art. Summary of the invention

[0005] The purpose of the present invention is to provide a comprehensive monitoring and forecasting method for geological disasters based on binary satellite fusion, so as to improve the accuracy of the geological disaster monitoring and early warning model, extend the geological disaster early warning time, and gain more precious time for emergency rescue.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A method for comprehensive monitoring and prediction of geological disasters based on binary satellite fusion, comprising the following steps:

[0008] S1. Develop Beidou monitoring station integrated machine and Beidou base station integrated machine, as well as high-precision boards that support multi-system satellite positioning to obtain accurate measurement data of surface displacement;

[0009] S2. Based on BDS / GNSS multiple monitoring points and base station networking technology, build "master-slave" nodes and establish a "point-line-surface" three-dimensional monitoring network;

[0010] S3, based on the adaptive attenuated memory Kalman filter algorithm AFMKF and multi-source sensor information fusion technology, data noise reduction and fusion processing;

[0011] S4, using the three-dimensional data of "sky-air-ground" as the input factor of the CNN model, introducing the time dimension, and establishing a nonlinear precipitation estimation model from low-middle-high-level feature extraction to actual precipitation;

[0012] S5. Construct a modeling sample Sample: S_t+S+ S_+t based on the precipitation or heavy precipitation in the actual observation, and use the reconstructed modeling sample to train the heavy precipitation radar evolution process of the model;

[0013] S6. Construct a data set of landslide displacement prediction factors and multi-factor influencing factors in complex environments, and on this basis construct a multi-dimensional feature vector matrix to extract important and key features of environmental factors;

[0014] S7. Taking the characteristic vector constructed in S6 as a warning factor and combining it with the deformation monitoring curve, a medium- and long-term forecast model, a short-term forecast model and an impending slip forecast model are formed according to the characteristics of the time series of the data.

[0015] Preferably, in step S1, the specific features of the Beidou monitoring station integrated machine, the Beidou reference station integrated machine and the high-precision board are:

[0016] The high-precision board contains two BDS / GNSS multi-system chips, based on the carrier phase measurement algorithm, and uses a radio interferometer solution to invert the position coordinate values ​​of certain nodes and calculate the angle change through the coordinate values; the signal processing module of the high-precision board is designed with FPGA (5CEA7U19) + 0ARM (AM3225); the Beidou monitoring station all-in-one and Beidou base station all-in-one are developed based on BDS / GNSS multi-mode chips, and are used to monitor deformation detection quantities such as displacement and angle change.

[0017] As a preferred method, a "master-slave" node is constructed in S2 to build a "point-line-surface" three-dimensional monitoring network, which specifically includes the following steps:

[0018] A 1+N monitoring network is arranged in each monitoring area. Data is transmitted between each monitoring site and the reference site in the local area network and then transmitted back to the server through the reference site. Communication links are designed to complete the reception and transmission of satellite observation data, analyze the power in the data sending and receiving states, fuse the data in sleep mode, and process redundant data in the network to minimize the required data volume while meeting application requirements.

[0019] Preferably, data noise reduction and fusion processing is performed in S3, including the following steps:

[0020] The adaptive attenuated memory Kalman filter algorithm AFMKF is used to introduce the attenuation factor to dynamically track and monitor the deformation data and reduce the noise error affecting the system. A data fusion algorithm based on the AFMKF algorithm is designed for the CKF algorithm, and a variety of auxiliary sensors are used to carry out auxiliary monitoring of disaster points. The Beidou high-precision positioning data and multi-source sensor data are fused and processed.

[0021] Preferably, the nonlinear precipitation estimation model in S4 is characterized by:

[0022] Using the three-dimensional data of "sky-air-ground" as input factors, the time dimension is introduced, and the time window is the data of the past 7-8 frames. A nonlinear precipitation estimation model is established from the feature extraction of low-middle-high layers to the actual precipitation.

[0023] Preferably, in the modeling sample Sample: S_t+S+ S_+t constructed in S5, S_t is the radar echo sequence 1 to 2 hours before the precipitation process; S is the radar echo time series of the entire precipitation process; S_+t is the radar echo sequence 1 to 2 hours after the rain.

[0024] Preferably, constructing a data set of landslide displacement prediction factors and multi-factor influencing factors in a complex environment in S6 specifically includes the following steps:

[0025] The surface displacement, grid rainfall data, ground fissures, unstable slopes, and hydrogeological condition data obtained by Beidou satellites and Fengyun meteorological satellites are fused with multi-dimensional features. The specific process is as follows: first, the multi-sensor monitoring data of the initial state are centrally stored and managed to generate an initial set; a nonlinear function is applied to each member of the initial set to generate a prediction set; the predicted state is used to estimate the mean value and error covariance matrix of the state; the disturbance value is used to generate an observation set; based on the prediction set and the observation set, the Kalman filter gain is calculated to obtain the analysis set and extract the overall deformation.

[0026] As a preferred method, in S7, a comprehensive qualitative and quantitative analysis is conducted on the deformation stage of the disaster body, and the specific steps of the forecast are divided into medium- and long-term forecast, short-term forecast and impending landslide forecast according to the time scale of the deformation and destruction stage:

[0027] Taking the surface displacement and rainfall obtained by S3 as warning factors and combining with the deformation monitoring curve, a complex model of long and short-term memory artificial neural network based on various types of data is established according to the characteristics of the time series of data; the neural network method is used to predict the cumulative displacement in the medium and long-term forecast stage; the kinematic method is used to predict the time when the deformation enters the pre-slip stage in the short-term forecast stage; the Verhulst model is used to make a more accurate forecast of the time of disaster occurrence in the pre-slip stage.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] (1) The method for comprehensive monitoring and forecasting of geological disasters based on dual-satellite fusion of the present invention combines the advantages of BeiDou and Fengyun satellites (referred to as "dual satellites") in the application of geological disaster monitoring and early warning, solves the problem of insufficient information caused by single data, and through dual-satellite data fusion, is conducive to improving the accuracy and timeliness of the geological disaster monitoring and early warning model.

[0030] (2) In the method for comprehensive monitoring and forecasting of geological disasters based on dual-satellite fusion of the present invention, the transmission of ground data is based on a "point-line-surface" three-dimensional monitoring network of monitoring points and base stations, which ensures the normal transmission of data in a weak signal environment.

[0031] (3) The method for comprehensive monitoring and forecasting of geological disasters based on dual-satellite fusion of the present invention fuses the Beidou high-precision positioning data and multi-source sensor data, completes the in-network processing of redundant data, and solves the problem of isolated deformation data in traditional disaster monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic flow chart of the method of the present invention;

[0033] Figure 2 It is a schematic diagram of power consumption in data sending and receiving states;

[0034] Figure 3 It is a schematic diagram of data denoising and fusion processing using the AFMKF algorithm;

[0035] Figure 4 It is a schematic diagram of a series of characteristics of heavy precipitation shown on radar and satellite images;

[0036] Figure 5 It is a schematic diagram of establishing a precipitation correction forecast model for a graded numerical model;

[0037] Figure 6 It is a schematic diagram of multi-sensor data fusion;

[0038] Figure 7 It is a flowchart of the creation, training and prediction of BP neural network. DETAILED DESCRIPTION

[0039] The following is a clear and complete description of the technical solution of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0040] In the description of the present invention, it should be noted that the orientations or positional relationships indicated by terms such as "center", "up", "down", "left", "right", "vertical", "horizontal", "inside" and "outside" are based on the orientations or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the invention.

[0041] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0042] See attached Figure 1 , a method for comprehensive monitoring and prediction of geological disasters based on binary satellite fusion, comprising the following steps:

[0043] S1. Develop Beidou monitoring station and Beidou base station integrated machine, as well as high-precision boards and other terminal equipment that support multi-system satellite positioning such as BDS and GPS to obtain accurate measurement data of surface displacement.

[0044] Based on the sustainable landslide monitoring method of BDS / GNSS, the change data between the real-time three-dimensional coordinates and the initial coordinates of each monitoring point are studied and analyzed to obtain the change amount of the monitoring point. The multi-dimensional deformation monitoring technology combining short baseline and long baseline is used to perform error correction and real-time high-precision positioning solution to obtain accurate measurement data of surface displacement. A local solution solution based on the local area network is designed to transmit information data.

[0045] Based on the carrier phase measurement algorithm, a BDS / GNSS high-precision board is developed to achieve static mm-level precision measurement. Specifically, a pair of two BDS / GNSS multi-system chips are used to obtain the pseudo-range and carrier phase values ​​of BDS / GNSS multi-system satellites; a radio interferometer solution is used to take the distance between the phase centers of the two receiving antennas as the measurement baseline and the length measurement value; when the two antennas receive the RF signal, after low-noise amplification, frequency conversion and filtering, coherent measurement is performed in the receiver, and the coherent carrier phase difference data can be used to invert the baseline length through the algorithm. The intersection of multiple lengths can invert the position coordinate values ​​of certain nodes, and the angle change can be calculated through the coordinate values. The signal processing module of the high-precision board receiver adopts FPGA (5CEA7U19) + 0ARM (AM3225) design, receives the intermediate frequency digital signal down-converted by the RF module, completes the processing of the digital signal, and realizes positioning solution and integrated navigation.

[0046] Based on the BDS / GNSS multi-mode chip, the Beidou monitoring station integrated machine and Beidou base station integrated machine are developed and deployed on deformation monitoring targets such as geological disaster bodies and buildings to monitor deformation monitoring quantities such as displacement and angle change. The deformation monitoring station is connected to the precision positioning base station to obtain high-precision base position data. At the same time, a local solution solution based on the local area network is designed to transmit information data. Among them, the Beidou base station integrated machine consists of an AC-220V surge protector, a 220V to 12V switching power supply, an industrial computer, a 4G router, a base station receiver, an antenna lightning arrester, etc. The Beidou monitoring station integrated machine is mainly composed of a power module, a receiver module, a data transmission module and a heat dissipation module. According to different power supply methods, it can be divided into two versions: a 220V mains power supply version and a solar power supply version.

[0047] S2. Based on BDS / GNSS multiple monitoring points and base station networking technology, build a "master-slave" node and establish a "point-line-surface" three-dimensional monitoring network.

[0048] First, a 1+N mode monitoring network is deployed in each monitoring area. Data is transmitted via the local area network between each monitoring site and the reference site, and finally transmitted back to the server through the reference site.

[0049] Secondly, design the communication link to complete the reception and transmission of satellite observation data. Figure 2As shown in the figure, the power under data sending and receiving state is analyzed, the sleep mode is used to merge the data, the redundant data is processed in the network, and the required data volume is minimized under the premise of meeting the application requirements. The aopre wireless bridge is used for point-to-point wireless transmission, in which the wireless frequency band is 5.8G and the transmission speed is 300M. When in the mountains or in the wild, the 4G DTU communication solution is used to use the 4G network to achieve long-distance wireless transmission. The module uses a hardware watchdog circuit to monitor the module operation status in real time, and the voltage input end adds an anti-short circuit design. The serial data rate of this solution is 300~115200bps, supports RS232 and RS485 interfaces, and is compatible with the RS232 circuit and RS485 circuit expanded on the STM32F767ZFT6 core processor. When the transmission distance is hundreds of meters or even kilometers, a digital radio is used for data transmission.

[0050] S3. Data denoising and fusion processing are performed based on the adaptive attenuated memory Kalman filter algorithm AFMKF and multi-source sensor information fusion technology.

[0051] As attached Figure 3 As shown in the figure, an adaptive attenuated memory Kalman filter algorithm AFMKF is used to introduce an attenuation factor to dynamically track and monitor the deformation data, reduce the noise error affecting the system, and accelerate the convergence of data monitoring. In view of the uncertainty of the attenuated memory factor, periodic high-precision observations are used to update the new information sequence, and the calculation of the attenuated memory factor is constrained.

[0052] The AFMKF data fusion algorithm was designed for the CKF algorithm, and a variety of auxiliary sensors were used to assist in monitoring the disaster points, including: intelligent rainfall monitors, inclination / acceleration monitors, intelligent crack displacement monitors, mud and water level depth monitors, remote alarms and other universal geological disaster monitoring and early warning equipment. The Beidou high-precision positioning data and multi-source sensor data were fused to complete the in-network processing of redundant data while solving the problem of isolated deformation data in traditional disaster monitoring.

[0053] S4. The three-dimensional data of "sky-air-ground" is used as the input factor of the CNN model, and the time dimension is introduced to establish a nonlinear precipitation estimation model from low-middle-high-level feature extraction to actual precipitation.

[0054] The CNN model uses the three-dimensional data of "sky-air-ground" as the input factor of the model. The model input not only considers the real-time changes of surface elements, but also considers the high-altitude radar echoes and satellite cloud map trends. The time dimension is further introduced, and the time window is the data of the past 7-8 frames. A nonlinear precipitation estimation model is established from low-middle-high-level feature extraction to actual precipitation. Among them, the nonlinear precipitation estimation model can be used for:

[0055] (1) Mesoscale characteristics and automatic identification of heavy rainfall;

[0056] By finding typical short-term heavy rainfall cases, combining the new generation of Doppler radar, wind profiler radar, meteorological wind and cloud geostationary satellite, mesoscale ground station, single-point sounding detection and numerical model data, the popular learning method is used to extract the characteristics of the circulation physical field, and the characteristic information of the precipitation formation stage, mature stage and weakening and dissipation stage is obtained. Through the analysis of the mesoscale characteristic information of heavy rainfall, the image pattern recognition and other technical methods are used to automatically identify a series of characteristics of heavy rainfall on radar and satellite images, such as the attached Figure 4 shown.

[0057] (2) Short-term and nowcasting of heavy rainfall areas;

[0058] Construct a satellite precipitation cloud cluster time series forecast model for heavy precipitation CI based on deep convolutional neural network, and effectively integrate and use numerical model data with atmospheric background field information to construct multi-dimensional original input, effectively track heavy precipitation satellite cloud cluster signals through learning, and realize the forecast of candidate areas near heavy precipitation in a short time. Construct a radar video information time series extrapolation forecast model for heavy precipitation areas based on deep learning technology (including RNN model, two-dimensional CNN model and three-dimensional CNN model, etc.), and combine the indicative role of mesoscale variable pressure field to further compare, correct and calibrate future heavy precipitation areas.

[0059] (3) Quantitative and refined forecast of rainfall levels of heavy rainfall;

[0060] A quantitative evaluation is conducted on the radar and satellite imagery of heavy precipitation obtained from the above forecast. A cross-channel three-dimensional columnar convolution is used, that is, the network input is formed in the form of multiple channels, fusion of multiple layers of data, and multiple feature quantities, and through convolution learning evolution, the magnitude of heavy precipitation in six time periods of 1 hour each in the next 0-6 hours is obtained. The heavy precipitation location network and parameters of various data sources obtained from the research are used as the front-end network and starting point, and the back-end network is established by deconvolution operation, and then the front-end and back-end network fusion training is performed to achieve accurate forecasts of component-level heavy precipitation in a short period of time in the future. If there is no data on heavy precipitation weather processes of this type in the training set, the deep adaptation network (DAN) in the transfer learning technology is used to adaptively adjust the heavy precipitation forecast results.

[0061] (4) Conduct nonlinear precipitation forecast based on multi-model super ensemble forecast;

[0062] For different levels of precipitation, the European fine grid precipitation is used as the forecast factor, combined with the physical quantity characteristic factor set, and the fuzzy neural network is used to forecast precipitation for each station. Through the rainfall forecast of each station in Guangxi and the precipitation field forecast of all numerical model forecast members, the research on using the random forest algorithm to achieve station-to-grid forecast is carried out. A graded numerical model precipitation correction forecast model is established, as shown in the attached figure. Figure 5 shown.

[0063] S5 constructs modeling samples based on the occurrence of precipitation or heavy precipitation in actual observations: Sample: S_t+S+ S_+t, and uses the reconstructed modeling samples to train the model's radar evolution process of heavy precipitation.

[0064] The occurrence of short-term heavy precipitation events is a low-probability event in weather phenomena. In the entire radar echo time series, the distribution of its samples is very uneven. To this end, a modeling sample based on precipitation or heavy precipitation in actual observation is constructed: Sample: S_t+S+ S_+t, where S_t is the radar echo sequence 1 to 2 hours before the precipitation process; S is the radar echo time series of the entire precipitation process; S_+t is the radar echo sequence 1 to 2 hours after the rain. When the model simulates the radar evolution process of heavy precipitation, the reconstructed modeling samples are used for training.

[0065] S6. Construct a data set of landslide displacement prediction factors and multi-factor influencing factors in complex environments, and on this basis construct a multi-dimensional feature vector matrix to extract important and key features of environmental factors.

[0066] The surface displacement and grid rainfall data obtained by Beidou satellite and Fengyun meteorological satellite in S1 and S2, as well as the ground fissures, unstable slopes, hydrogeological conditions and other data obtained by SPOT7 and Landsat8 images combined with field measurement methods, are centrally stored and managed. First, RabbitMQ message queue is used to receive data, and RabbitMQ cluster is built in the server, equipped with haproxy load balancing, and a mirror cluster is established. The publish / subscribe mode is adopted. After the data is stored in RabbitMQ, the early warning model receives real-time data through the subscription mode. The database is constructed to classify, collect and store the multi-source data of each monitoring system. The data tables mainly include the basic information table of monitoring stations, the basic information table of monitoring points, the monitoring data table, the early warning object information table, the early warning log table, the file upload information table, the file upload record table, the benchmark port information table, the sensor basic information table, the sensor data table, the system log table, and the scheduled task table. According to the physical design plan of the database, the hardware and software selected for database construction should be organically integrated.

[0067] The influencing factors of the disaster body are adjusted and judged, and real-time monitoring of the corresponding more important factors is established, and data is transmitted back. In the centralized fusion structure, the multi-sensor information fusion system will use multiple sensors to obtain measurement information containing various target characteristics, and send it directly to the fusion center to perform processing such as spatial registration, data correlation, tracking filtering and prediction, and appropriately combine and synthesize this information according to the optimization criteria. Among them, spatial registration refers to the coordination of synchronous and asynchronous data through the registration algorithm to estimate and compensate for sensor deviations. The process is shown in the attached figure. Figure 6 As shown in the figure, the target tracking algorithm of integrated Kalman filter (ENKF) is used to perform data fusion of multi-dimensional features. The specific process is as follows: first, a series of disturbances are applied to the multi-sensor monitoring data of the initial state to generate the initial set; a nonlinear function is applied to each member in the initial set to generate a prediction set; the predicted state is used to estimate the mean value and error covariance matrix of the state; the disturbance value is used to generate the observation set. Based on the prediction set and the observation set, the Kalman filter gain is calculated to obtain the analysis set and extract the overall deformation.

[0068] S7. Taking the characteristic vectors such as surface displacement and rainfall in S6 as early warning factors and combining them with the deformation monitoring curve, a medium- and long-term forecast model, a short-term forecast model and a pre-slip forecast model are formed according to the characteristics of the time series of the data.

[0069] A comprehensive qualitative and quantitative analysis is conducted on the deformation stage of the disaster body; at the same time, according to the time scale of the deformation and destruction stage, it can be divided into medium- and long-term forecasts, short-term forecasts and impending landslide forecasts.

[0070] (1) In the medium and long term, the neural network method is used to predict the cumulative displacement. The neural network learning process is a process of error correction while backpropagating, which is divided into two stages: the first stage is carried out in the forward propagation direction, and the output value of each neuron can be obtained from the direction from the input layer to the hidden layer to the output layer, until the output value of the final output layer is obtained; the second stage is carried out in the reverse propagation direction, and the connection weights between nodes are adjusted according to the error between the actual output of the output layer and the expected output value, so that the error is minimized. See the attached Figure 7 . There are study samples ( is the number of samples), given the network vector After the weights are calculated, the formula can be used:

[0071]

[0072]

[0073] The output value of the network is calculated through the basic neural network hidden layer neuron node output formula , and define the output error of the network as:

[0074]

[0075] The error function is defined as:

[0076]

[0077] In the back propagation algorithm, it is along the error function Follow The changing negative gradient direction is To be corrected. The correction value is ,Pick

[0078]

[0079] in, is the learning rate, which is a number between 0 and 1. Substitute dp and ep into the formula In, we have:

[0080]

[0081] This formula can be used as the basic formula for iterative calculation of the sorted samples in the next step.

[0082] In order to meet the actual needs of forecasting, The nonlinear mapping ability of the neural network is used for modeling. Before training the network, the weights and thresholds of the network must be initialized. The principle is as follows: If the network size is small, the selected learning rate and the number of learning steps can be smaller; if the network scale is large, then take a larger value accordingly until it is adjusted appropriately.

[0083] (2) In the short-term prediction stage, the time when the deformation enters the pre-slip stage is predicted by using kinematic methods. As a reference, the landslide after removing the constant velocity part Initial acceleration rate of the curve, confirming the entry into the forestation stage The velocity value of the point; make the landslide and Curve Find the time interval of constant acceleration in the curve and set it as , remove the uniform speed part ,but The corresponding time interval of constant acceleration motion in the curve is day; in Selected of point (that is, when the landslide deforms to this point, the selection of this point is not fixed) as the forecast starting point, then In the curve The point coordinates are ,in for The actual .

[0084] In order to facilitate the parabola quantitative prediction method, the coefficient for The parabola To create a graph, follow these steps:

[0085] ① Make and Curve, when the deformation enters the initial acceleration stage, Curve conversion Line, using The line determines the initial acceleration rate value after removing the uniform speed part;

[0086] ②According to and The curve determines the constant acceleration stage and selects the prediction starting point. Compare the curve with the drawn figure to determine the equation of the fitted parabola;

[0087] ③Set up multiple parabola equations to intercept Curve, proceed The established parabola coefficient is slightly larger than the fitted parabola coefficient;

[0088] ④Use , Curves and macroscopic deformation signs eliminate erroneous results and determine the correct point.

[0089] (3) The Verhulst model is used to make a more accurate prediction of the disaster occurrence time during the pre-slip stage. Suppose the original equally spaced monitoring data sequence is ;

[0090]

[0091] right Perform an AGO transformation and we get

[0092]

[0093] by Fit the Verhulst first-order whitened nonlinear differential equation:

[0094]

[0095] In the formula, and is the unknown coefficient, which can be obtained by the least square method, including:

[0096]

[0097]

[0098]

[0099] Substitute the unknown coefficients into the first-order whitened nonlinear differential equation and solve the solution of the nonlinear differential equation:

[0100]

[0101] The above formula is the established time Verhulst nonlinear differential dynamic forecast model. Among them, is the initial moment.

[0102] like If it is non-equally spaced time series data, it should be first converted into an equivalent equal-spaced series before modeling. Since the evolution of the disaster body is very similar to the process of biological reproduction to extinction, the critical value (inflection point value) of the biological transformation from maturity (rapid growth) to extinction (slow growth) can be As the critical displacement value. Alternative The moment to solve the landslide :

[0103]

[0104] Since the initial time is generally 0, the above formula becomes

[0105]

[0106] It is actually a time series number, the actual slope failure time Should be

[0107]

[0108] Where △t is the average interval time of monitoring data.

[0109] In addition, this embodiment releases geological disaster warning information according to the development stage, urgency, unstable development trend and possible harm degree of geological disasters. The warning levels are divided into level 1, level 2, level 3 and level 4, corresponding to different levels of geological disaster risk, high risk, relatively high risk and general risk, respectively, and marked in red, orange, yellow and blue. Level 1 is the highest level.

[0110] The foregoing description of specific exemplary embodiments of the present invention is for the purpose of illustration and demonstration. These descriptions are not intended to limit the present invention to the precise form disclosed, and it is clear that many changes and variations can be made based on the above teachings. The purpose of selecting and describing the exemplary embodiments is to explain the specific principles of the present invention and its practical application, so that those skilled in the art can realize and utilize various different exemplary embodiments of the present invention and various different selections and changes. The scope of the present invention is intended to be limited by the claims and their equivalents.

Claims

1. A method for comprehensive monitoring and prediction of geological disasters based on binary satellite fusion, characterized in that: The following steps are involved: S1. Develop Beidou monitoring station integrated machine and Beidou base station integrated machine, as well as high-precision boards that support multi-system satellite positioning to obtain accurate measurement data of surface displacement; S2. Based on BDS / GNSS multiple monitoring points and base station networking technology, build "master-slave" nodes and establish a "point-line-surface" three-dimensional monitoring network; S3, based on the adaptive attenuated memory Kalman filter algorithm AFMKF and multi-source sensor information fusion technology, data noise reduction and fusion processing; S4, using the "sky-air-ground" three-dimensional data as the input factor of the CNN model, introducing the time dimension, and establishing a nonlinear precipitation estimation model from low-middle-high-level feature extraction to actual precipitation; S5. Construct a modeling sample Sample: S_t+S+ S_+t based on the precipitation or heavy precipitation in the actual observation, and use the reconstructed modeling sample to train the heavy precipitation radar evolution process of the model; S6. Construct a data set of landslide displacement prediction factors and multi-factor influencing factors in complex environments, and on this basis construct a multi-dimensional feature vector matrix to extract important and key features of environmental factors; S7. Taking the characteristic vector constructed in S6 as a warning factor and combining it with the deformation monitoring curve, a medium- and long-term forecast model, a short-term forecast model and an impending slip forecast model are formed according to the characteristics of the time series of the data.

2. The method for comprehensive monitoring and prediction of geological disasters based on dual-satellite fusion according to claim 1 is characterized in that: In step S1, the specific features of the Beidou monitoring station integrated machine, Beidou reference station integrated machine and high-precision board are: The high-precision board contains two BDS / GNSS multi-system chips, based on the carrier phase measurement algorithm and the radio interferometer solution, which can invert the position coordinate values ​​of certain nodes and calculate the angle change through the coordinate values; The signal processing module of the high-precision board adopts FPGA (5CEA7U19) + 0ARM (AM3225) design; the Beidou monitoring station integrated machine and Beidou base station integrated machine are developed based on BDS / GNSS multi-mode chips, which are used to monitor deformation detection quantities, including displacement and angle changes.

3. The method for comprehensive monitoring and prediction of geological disasters based on dual-satellite fusion according to claim 1 is characterized in that: In S2, a "master-slave" node is constructed to build a "point-line-surface" three-dimensional monitoring network, which specifically includes the following steps: A 1+N monitoring network is arranged in each monitoring area. Data is transmitted between each monitoring site and the reference site in the local area network and then transmitted back to the server through the reference site. Communication links are designed to complete the reception and transmission of satellite observation data, analyze the power in the data sending and receiving states, fuse the data in sleep mode, and process redundant data in the network to minimize the required data volume while meeting application requirements.

4. The method for comprehensive monitoring and prediction of geological disasters based on dual-satellite fusion according to claim 1 is characterized in that: Data denoising and fusion processing is performed in S3, including the following steps: The adaptive attenuated memory Kalman filter algorithm AFMKF is used to introduce the attenuation factor to dynamically track and monitor the deformation data and reduce the noise error affecting the system. A data fusion algorithm based on the AFMKF algorithm is designed for the CKF algorithm, and a variety of auxiliary sensors are used to carry out auxiliary monitoring of disaster points. The Beidou high-precision positioning data and multi-source sensor data are fused and processed.

5. The method for comprehensive monitoring and prediction of geological disasters based on dual-satellite fusion according to claim 1 is characterized in that: The nonlinear precipitation estimation model in S4 is characterized by: Using the three-dimensional data of "sky-air-ground" as input factors, the time dimension is introduced, and the time window is the data of the past 7-8 frames. A nonlinear precipitation estimation model is established from the feature extraction of low-middle-high layers to the actual precipitation.

6. The method for comprehensive monitoring and prediction of geological disasters based on dual-satellite fusion according to claim 1 is characterized in that: In the modeling sample Sample constructed in S5: S_t+S+ S_+t, S_t is the radar echo sequence 1 to 2 hours before the precipitation process; S is the radar echo time series of the entire precipitation process; S_+t is the radar echo sequence 1 to 2 hours after the rain.

7. The method for comprehensive monitoring and prediction of geological disasters based on dual-satellite fusion according to claim 1 is characterized in that: In S6, the displacement prediction factors of landslides and the multi-factor influencing factor dataset in complex environments are constructed, which specifically includes the following steps: The surface displacement, grid rainfall data, ground fissures, unstable slopes, and hydrogeological condition data obtained by Beidou satellites and Fengyun meteorological satellites are fused with multi-dimensional features. The specific process is as follows: first, the multi-sensor monitoring data of the initial state are centrally stored and managed to generate an initial set; a nonlinear function is applied to each member of the initial set to generate a prediction set; the predicted state is used to estimate the mean value and error covariance matrix of the state; the disturbance value is used to generate an observation set; based on the prediction set and the observation set, the Kalman filter gain is calculated to obtain the analysis set and extract the overall deformation.

8. The method for comprehensive monitoring and prediction of geological disasters based on dual-satellite fusion according to claim 1 is characterized in that: In S7, a comprehensive qualitative and quantitative analysis of the deformation stage of the disaster body is carried out. According to the time scale of the deformation and destruction stage, it is divided into medium- and long-term forecasts, short-term forecasts and impending landslide forecasts. The specific steps are as follows: Taking the surface displacement and rainfall obtained by S3 as warning factors and combining with the deformation monitoring curve, a complex model of long and short-term memory artificial neural network based on various types of data is established according to the characteristics of the time series of data; the neural network method is used to predict the cumulative displacement in the medium and long-term forecast stage; the kinematic method is used to predict the time when the deformation enters the pre-slip stage in the short-term forecast stage; the Verhulst model is used to make a more accurate forecast of the time of disaster occurrence in the pre-slip stage.

Images