A reservoir geomorphology data acquisition method and system based on multi-source data fusion

Abstract

The application relates to the field of water conservancy projects and geographic information technology, and discloses a reservoir landform data acquisition method and system based on multi-source data fusion, which comprises the following steps: acquiring multi-source remote sensing data and sensor data, and performing pretreatment and time sequence alignment adjustment to obtain a corrected time sequence consistent feature sequence; performing fusion processing according to the feature sequence to obtain an optimized landform model input; performing landform elevation calculation according to the optimized landform model input to determine a high-precision reservoir landform DEM; and performing landform space analysis according to the DEM to obtain a high-precision reservoir landform acquisition result. The method can solve the technical problems of low landform data acquisition precision and model distortion.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of water conservancy and geographic information technology, and in particular to a reservoir geomorphology data acquisition method and system based on multi-source data fusion. BACKGROUND

[0002] At present, reservoir geomorphology data acquisition is a key link in the field of water conservancy and geographic information, and its accuracy is directly related to the decision-making effect of water resources regulation, flood control and disaster reduction, and ecological protection.

[0003] In one prior art, the geomorphology acquisition method often relies on isolated or static data processing strategies. Such strategies are difficult to handle the format heterogeneity and noise interference of multi-source data when facing the dynamic changes of the reservoir environment; at the same time, they ignore the matching deviation between historical water level and real-time geomorphology, leading to fragmentation of the feature extraction process, and finally causing distortion of the geomorphology model in different water level stages.

[0004] Therefore, there is a technical problem of low accuracy and model distortion in the prior art. SUMMARY

[0005] The present application provides a reservoir geomorphology data acquisition method and system based on multi-source data fusion to solve the technical problem of low accuracy and model distortion in the prior art.

[0006] In a first aspect, to solve the above technical problem, the present application provides a reservoir geomorphology data acquisition method based on multi-source data fusion, comprising:

[0007] Obtain multi-source remote sensing data and sensor data and preprocess to obtain a denoised remote sensing sensor unified feature set;

[0008] According to the denoised remote sensing sensor unified feature set, perform historical water level feature correlation analysis to obtain a matching deviation index;

[0009] Perform time sequence alignment adjustment for the matching deviation index to determine a corrected time sequence consistent feature sequence;

[0010] According to the corrected time sequence consistent feature sequence, perform remote sensing resolution fusion processing to obtain an optimized geomorphology model input;

[0011] According to the optimized geomorphology model input, perform geomorphology elevation calculation to determine a high-precision reservoir geomorphology DEM;

[0012] According to the high-precision reservoir geomorphology DEM, perform geomorphology spatial analysis and classification to obtain a high-precision reservoir geomorphology acquisition result.

[0013] Preferably, the multi-source remote sensing data and sensor data are acquired and preprocessed to obtain a denoised remote sensing sensor unified feature set, including:

[0014] The multi-source remote sensing data and sensor data are acquired and noise filtering and format heterogeneous conversion are performed to obtain multi-source fusion initial data.

[0015] According to the multi-source fusion initial data, remote sensing image band selection and sensor reading timestamp are extracted, and timestamp alignment and clustering grouping denoising processing are performed to determine the denoised remote sensing sensor unified feature set.

[0016] Preferably, the denoised remote sensing sensor unified feature set is used for historical water level feature correlation analysis to obtain a matching deviation index, including:

[0017] A dynamic feature description sequence is obtained from the denoised remote sensing sensor unified feature set.

[0018] According to the dynamic feature description sequence, feature correlation analysis is performed to obtain a first matching preparation data set.

[0019] For the first matching preparation data set, historical water level sequence comparison is performed in combination with the dynamic feature description sequence to determine the matching deviation index.

[0020] Preferably, the matching deviation index is subjected to timing alignment adjustment to determine a corrected timing consistent feature sequence, including:

[0021] For the matching deviation index, bias cumulative value timing alignment adjustment is performed to obtain a bias adjustment feature set.

[0022] According to the bias adjustment feature set, a sliding window translation correction is used to obtain a timing consistent sequence.

[0023] The timing consistent sequence is subjected to timing continuity statistical analysis to determine the corrected timing consistent feature sequence.

[0024] Preferably, the corrected timing consistent feature sequence is subjected to remote sensing resolution fusion processing to obtain an optimized topographic model input, including:

[0025] For the corrected timing consistent feature sequence, remote sensing image resolution multi-dimensional description integration is performed to obtain a fusion resolution data set.

[0026] According to the fusion resolution data set, a resolution consistent description set is obtained through spatial grid alignment.

[0027] The resolution consistent description set is subjected to spatial smoothness analysis to determine the optimized topographic model input.

[0028] Preferably, the topographic height calculation is performed according to the optimized topographic model input to determine a high-precision reservoir topographic DEM, comprising:

[0029] The coherent topographic description is extracted from the optimized topographic model input, and the deviation influence data set is obtained by fusing the deviation influence of the water level change scenario simulation;

[0030] According to the deviation influence data set, the deviation correction processing is performed to obtain the deviation optimization description set;

[0031] According to the deviation optimization description set, the topographic height calculation is performed to obtain the high-precision reservoir topographic DEM.

[0032] Preferably, the topographic space analysis and classification are performed according to the high-precision reservoir topographic DEM to obtain high-precision reservoir topographic collection results, comprising:

[0033] The slope gradient is calculated according to the high-precision reservoir topographic DEM to obtain a slope distribution matrix;

[0034] The flow path is simulated through the slope distribution matrix, and the path interpolation processing is performed to determine a continuous path chain;

[0035] According to the continuous path chain and the slope distribution matrix, the region block is divided by using a watershed algorithm;

[0036] The region block is classified to obtain a topographic category group;

[0037] The collection data is integrated according to the topographic category group to obtain the high-precision reservoir topographic collection results.

[0038] In a second aspect, the present application provides a reservoir topographic data collection system based on multi-source data fusion, comprising:

[0039] The data preprocessing module is used for acquiring multi-source remote sensing data and sensor data, and performing preprocessing to obtain a denoised remote sensing sensor unified feature set;

[0040] The correlation analysis module is used for performing historical water level feature correlation analysis according to the denoised remote sensing sensor unified feature set to obtain a matching deviation index;

[0041] The time sequence adjustment module is used for performing time sequence alignment adjustment on the matching deviation index to determine a corrected time sequence consistent feature sequence;

[0042] The fusion processing module is used for performing remote sensing resolution fusion processing according to the corrected time sequence consistent feature sequence to obtain an optimized topographic model input;

[0043] a geomorphology modeling module, configured to perform geomorphology elevation calculation according to the optimized geomorphology model input, and determine a high-precision reservoir geomorphology DEM;

[0044] a spatial analysis module, configured to perform geomorphology spatial analysis and classification according to the high-precision reservoir geomorphology DEM, and obtain a high-precision reservoir geomorphology collection result.

[0045] In a third aspect, the present application further provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the reservoir geomorphology data collection method based on multi-source data fusion according to any one of the above.

[0046] In a fourth aspect, the present application further provides a computer readable storage medium, comprising a stored computer program, wherein the computer program controls the device where the computer readable storage medium is located to execute the reservoir geomorphology data collection method based on multi-source data fusion according to any one of the above when the computer program is running.

[0047] Compared with the prior art, the present application has the following beneficial effects:

[0048] (1) The present application acquires multi-source data such as remote sensing images and sensor readings, and performs noise filtering, format unification and timestamp alignment to form a unified feature set; then extracts dynamic features and performs relevance analysis with historical water level sequences to quantify matching deviation indicators; this method solves the problem of feature extraction fragmentation and neglecting historical matching deviation caused by data format heterogeneity and noise interference in the prior art, and provides a reliable data foundation for subsequent high-precision modeling.

[0049] (2) The present application performs time sequence alignment for matching deviation indicators to correct cumulative errors, and fuses multi-dimensional descriptions of the resolution of remote sensing images to generate a spatio-temporally consistent optimized geomorphology model input; this method solves the problem of geomorphology model distortion at different water level stages caused by inconsistent data time sequence and inability to accurately correspond to historical water levels in the prior art, and improves the coherence and reliability of geomorphology description.

[0050] (3) The present application further extracts geomorphology boundaries, calculates slopes and simulates water flow paths by using the high-precision reservoir geomorphology DEM, to realize a complete closed loop from data fusion, deviation correction to specific geomorphology spatial analysis; this method solves the problem of low accuracy of the final collection result caused by model distortion and lack of coherent description of geomorphology dynamic features in the prior art, and finally obtains a high-precision reservoir geomorphology collection result. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1is a process schematic diagram of a reservoir geomorphology data acquisition method based on multi-source data fusion provided by the first embodiment of the present application.

[0052] Figure 2 is a structural schematic diagram of a reservoir geomorphology data acquisition system based on multi-source data fusion provided by the second embodiment of the present application. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0054] With reference to Figure 1 The first embodiment of the present application provides a reservoir geomorphology data acquisition method based on multi-source data fusion, including the following steps:

[0055] S11, multi-source remote sensing data and sensor data are acquired and preprocessed to obtain a denoised remote sensing sensor unified feature set;

[0056] S12, historical water level feature correlation analysis is performed according to the denoised remote sensing sensor unified feature set to obtain a matching deviation index;

[0057] S13, time sequence alignment adjustment is performed for the matching deviation index to determine a corrected time sequence consistent feature sequence;

[0058] S14, remote sensing resolution fusion processing is performed according to the corrected time sequence consistent feature sequence to obtain an optimized geomorphology model input;

[0059] S15, geomorphology elevation calculation is performed according to the optimized geomorphology model input to determine a high-precision reservoir geomorphology DEM;

[0060] S16, geomorphology spatial analysis and classification are performed according to the high-precision reservoir geomorphology DEM to obtain a high-precision reservoir geomorphology acquisition result.

[0061] In step S11, multi-source remote sensing data and sensor data are acquired and preprocessed to obtain a denoised remote sensing sensor unified feature set, including:

[0062] The multi-source remote sensing data and sensor data are acquired and noise filtering and format heterogeneous conversion are performed to obtain multi-source fusion initial data;

[0063] According to the multi-source fusion initial data, remote sensing image band selection and sensor reading timestamp are extracted, timestamp alignment and clustering grouping denoising processing are performed, and the denoised remote sensing sensor unified feature set is determined.

[0064] In an implementation manner, the embodiment obtains multi-source remote sensing data (for example, satellite images containing remote sensing image resolution values) and sensor data (for example, ground monitoring station data containing sensor reading values). According to a preset measurement value threshold, the embodiment performs noise filtering on the sensor reading values.

[0065] It should be noted that the preset measurement value threshold is determined by statistical analysis of historical data. The embodiment collects reading values (for example, water level data) of the sensor in a stable running state for at least 24 hours, calculates the average value and standard deviation of the data set, and determines the value corresponding to "average value + 3 times standard deviation" as the preset measurement value threshold according to the 3-sigma principle. Any instantaneous reading value exceeding the preset measurement value threshold is identified as a noise point and is filtered out.

[0066] Exemplarily, it is assumed that the preset measurement value threshold is 1000 mm determined by historical data analysis. In the obtained sensor data, a group of continuous water level sequence is [800, 850, 1200, 750] (unit: mm), and the reading of 1200 mm is determined as a noise point and is filtered out, and the sequence becomes [800, 850, 750].

[0067] It should be noted that after noise filtering, the embodiment performs format heterogeneous conversion on the multi-source remote sensing data and sensor data. The embodiment converts all data in different formats (for example, remote sensing images in TIFF format, vector data in GeoJSON format, and sensor time series data in CSV format) into a standardized and self-describing geospatial data format (for example, NetCDF format), and unifies the coordinate reference system (for example, WGS 84), thereby obtaining the multi-source fusion initial data.

[0068] In another implementation manner, according to the multi-source fusion initial data, remote sensing image band selection and sensor reading timestamp are extracted, and timestamp alignment and clustering grouping denoising processing are performed.

[0069] It should be noted that the specific implementation of the timestamp alignment is that the embodiment extracts the timestamp sequence of each data source from the multi-source fusion initial data. The embodiment adopts time series resampling processing to unify all data to a fixed time interval. The determination of the time interval is based on the slowest updating frequency data source (for example, the minute-level updating of remote sensing images) in the multi-source fusion initial data. For example, if the updating interval of the slowest data source (remote sensing image) is 1 minute, the unified and fixed time interval is set to 1 minute.

[0070] It should be noted that for data with faster updating frequency than the unified time interval (for example, second-level sensor reading value), the embodiment performs aggregation processing (for example, calculates the average or median of all second-level reading values in the window) within the time interval (i.e. 1 minute), and takes the aggregation result as the representative value of the 1-minute time point, ensuring the synchronization of all data in the time dimension.

[0071] It should be noted that the specific implementation of the clustering grouping denoising processing is that the embodiment adopts k-means clustering algorithm for the timestamp-aligned data features (for example, specific remote sensing image band selection values, sensor reading values). The algorithm divides the data points in the time window into k clusters. The determination of the k value can be achieved by offline analysis, and the elbow method is used to cluster the historical data set. The historical feature data set is constructed by collecting and storing representative historical timestamp-aligned data features (including remote sensing image band selection values and sensor reading values). The embodiment selects the elbow point corresponding to the cluster number at which the total sum of squares tends to be flat as the k value through the elbow method analysis.

[0072] In this embodiment, clusters satisfying any of the following conditions are identified as noise clusters: (1) the total number of data points in the cluster is lower than a preset cluster size threshold; (2) the Euclidean distance between the cluster center (centroid) and the total centroid of all data points in the time window exceeds a preset outlier threshold. The preset cluster size threshold can be determined by analyzing the cluster size distribution of the historical normal data set after clustering, and selecting the value corresponding to the 5th percentile of the distribution. This method is a robust statistical method for non-parametric distribution, which can effectively define the boundary of "small clusters" in a statistical sense. The preset outlier threshold can be determined by analyzing the cluster center distance distribution of the historical normal data set after clustering, and according to the 3-sigma principle, the value corresponding to "average distance + 3 times standard deviation". In this embodiment, all data points in these identified noise clusters are marked as noise and removed. The final output of this processing is the denoised remote sensor unified feature set.

[0073] In step S12, according to the denoised remote sensor unified feature set, historical water level feature correlation analysis is performed to obtain a matching deviation index, including:

[0074] Obtain a dynamic feature description sequence from the denoised remote sensor unified feature set;

[0075] According to the dynamic feature description sequence, perform feature correlation analysis to obtain a first matching preparation data set;

[0076] For the first matching preparation data set, and in combination with the dynamic feature description sequence, perform historical water level sequence comparison to determine the matching deviation index.

[0077] In one implementation, in this embodiment, from the denoised remote sensor unified feature set obtained in S11, specific feature columns related to reservoir topographic changes (such as water level, flooded area) (for example, reflectivity of specific wave bands, reading values of specific sensors) are extracted, and sorted according to the aligned time stamps in S11 to form a feature vector that can reflect the change of reservoir topography over time. The feature vector is the dynamic feature description sequence.

[0078] It should be noted that the specific implementation of the feature correlation analysis is that this embodiment uses a pre-trained convolutional neural network (Convolutional Neural Network, CNN) model. The structure of the CNN model can include a one-dimensional convolutional layer (for example, containing 32 convolutional kernels with a size of 3) for receiving the dynamic feature description sequence, a ReLU activation layer, a one-dimensional maximum pooling layer (for example, the pooling size is 2), a flattening layer, and a fully connected layer for outputting correlation weights.

[0079] It should be noted that the CNN model is determined by offline training. The training process is that the embodiment uses a historical data set containing historical dynamic feature description sequences and their corresponding real historical water level sequences as labels. The embodiment uses mean squared error (MSE) as the loss function, and iteratively trains the CNN model through the Adam optimizer until the loss value of the model converges on the independent validation set. The hyperparameters (such as the number of convolution kernels 32 and the size 3) in the model are determined by using the grid search (GridSearch) strategy in the offline training stage, optimizing in the preset parameter space (for example, the convolution kernel number range [16, 32, 64]), and selecting the parameter combination with the smallest loss value on the validation set.

[0080] In another implementation, the embodiment inputs the dynamic feature description sequence into the trained convolutional neural network model, which captures local time sequence patterns in the dynamic feature description sequence through its one-dimensional convolutional layer, analyzes the nonlinear dependence between these patterns and the water level change indicators, and finally outputs an association weight matrix that quantifies the dependence strength. The association weight matrix is the first matching preparation data set.

[0081] It should be noted that the embodiment compares the first matching preparation data set with a preset historical water level sequence in combination with the dynamic feature description sequence. The embodiment compares the association weight value in the first matching preparation data set (i.e., the association weight matrix) with a preset weight threshold. If the association weight value exceeds the preset weight threshold at a certain time point, the embodiment determines that the current dynamic feature deviates significantly from the historical rule. The embodiment extracts the current feature value (for example, the current average water level 0.80 meters) in the dynamic feature description sequence at the same time point, and subtracts it from the reference value (for example, the historical same period average 0.75 meters) in the historical water level sequence. The calculated difference is determined as the matching deviation indicator.

[0082] It should be noted that the preset historical water level sequence is constructed as follows: the embodiment collects historical water level measurement data of a reservoir (for example, in the past five years), and aggregates them according to the time dimension (for example, according to the calendar day or according to the week), calculates the statistical reference value (for example, the historical same period average) of each time unit, and the set of reference values constitutes the preset historical water level sequence.

[0083] It should be noted that the preset weight threshold is determined by statistical analysis of historical data. The determination process is that the embodiment collects a large number of correlation weight values under historical normal working conditions (for example, non-flood season, non-extreme dry period), and constructs a baseline data set; the embodiment calculates the probability density distribution of the baseline data set, and selects the value corresponding to the 95th percentile of the distribution as the preset weight threshold. Selecting the 95th percentile is a robust statistical method commonly used to define the statistical upper limit of the normal operating state in non-normal distribution data, and to identify high-confidence abnormal events.

[0084] Exemplarily, it is assumed that the preset weight threshold is 0.85 determined by historical data analysis. In S11, the correlation weight value (i.e. the first matching preparation data set) of a certain dynamic feature obtained by CNN calculation is 0.92. Since 0.92 exceeds 0.85, the embodiment determines that deviation occurs. At this time, the embodiment obtains the dynamic feature value (current water level 800 mm) corresponding to the time point and the historical water level sequence value (750 mm), and records the difference (50 mm) as the matching deviation indicator.

[0085] In step S13, the matching deviation indicator is adjusted in time sequence alignment to determine the corrected time sequence consistent feature sequence, including:

[0086] The matching deviation indicator is adjusted in deviation cumulative value time sequence alignment to obtain a deviation adjustment feature set;

[0087] According to the deviation adjustment feature set, a sliding window translation correction is adopted to obtain a time sequence consistent sequence;

[0088] The time sequence consistent sequence is statistically analyzed in time sequence continuity to determine the corrected time sequence consistent feature sequence.

[0089] In one implementation, the embodiment adjusts the matching deviation indicator obtained in S12 in deviation cumulative value time sequence alignment. The specific implementation of the adjustment is that the embodiment first calculates the cumulative sum of the matching deviation indicator in a time window to obtain a deviation cumulative value. The embodiment compares the deviation cumulative value with a preset cumulative threshold. If the deviation cumulative value exceeds the preset cumulative threshold, the embodiment adopts Min-Max normalization processing on the deviation indicator sequence in the time window, maps the sequence value linearly to a new interval of [0, the preset cumulative threshold] by multiplying the difference between the original value and the sequence minimum value by the preset cumulative threshold, and then dividing the product by the difference between the sequence maximum value and the sequence minimum value, to obtain the deviation adjustment feature set.

[0090] It should be noted that the preset cumulative threshold is determined by statistical analysis of historical data. The determination process is that the embodiment collects the matching deviation index under historical normal working conditions, and calculates the deviation cumulative value. The embodiment constructs a statistical distribution of all historical deviation cumulative values, and selects the value corresponding to the 95th percentile of the distribution as the preset cumulative threshold. The percentile is a robust statistical method widely used in system performance engineering, which is based on the fact that this value can effectively cover most (95%) of the normal running fluctuations while excluding the top 5% of extreme abnormal values, thereby balancing the alert sensitivity and system stability.

[0091] In another implementation, the embodiment adopts sliding window translation correction according to the deviation adjustment feature set. The specific implementation of the correction is that the embodiment uses a sliding average filter with a preset window size (for example, 3 time units) to traverse the deviation adjustment feature set point by point. The embodiment replaces the arithmetic mean value of all data points in the window with the new value of the center point of the window, thereby smoothing the local noise peaks in the sequence and obtaining the time sequence consistent sequence.

[0092] It is worth noting that the embodiment performs time sequence consistency statistical analysis on the time sequence consistent sequence. The specific implementation of the analysis is that the embodiment calculates the first-order difference (i.e., the change rate between adjacent time points) of the time sequence consistent sequence and obtains the variance of the first-order difference sequence. The variance value quantifies the local jitter degree of the sequence.

[0093] It should be noted that the embodiment compares the calculated variance value with the preset consistency threshold. The preset consistency threshold is determined by offline analysis, and the determination process is that the embodiment collects a large number of historical terrain sequences known to be "highly consistent" and labeled, and calculates the first-order difference variance. Select the 95th percentile of the variance distribution as the preset consistency threshold. If the calculated variance value is lower than the preset consistency threshold (indicating that the sequence is smooth and consistent), the embodiment determines the time sequence consistent sequence as the corrected time sequence consistent feature sequence.

[0094] In step S14, remote sensing resolution fusion processing is performed according to the corrected time sequence consistent feature sequence to obtain an optimized terrain model input, including:

[0095] For the corrected time sequence consistent feature sequence, remote sensing image resolution multi-dimensional description integration is performed to obtain a fused resolution data set.

[0096] According to the fused resolution data set, a resolution consistent description set is obtained through spatial grid alignment.

[0097] performing spatial smoothness analysis on the resolution consistent description set to determine the optimized geomorphic model input.

[0098] In an implementation, the embodiment is directed to the post-correction time-consistent feature sequence obtained in S13, and performs integration of remote sensing image resolution multi-dimensional description.

[0099] It should be noted that the specific implementation of the integration includes spatial interpolation processing and raster fusion processing. Since the post-correction time-consistent feature sequence (for example, water level change vector) is one-dimensional data existing in discrete sensor points, and the resolution multi-dimensional description (for example, pixel density vector) is two-dimensional spatial data distributed in the entire remote sensing image, the embodiment first uses the inverse distance weighting interpolation (Inverse Distance Weighting, IDW) method to estimate the feature value of each grid cell in the entire study area according to the known value and spatial coordinates of each sensor point, and generates a "time feature raster map". Subsequently, the embodiment performs weighted summation processing to perform weighted summation of the "time feature raster map" and the resolution multi-dimensional description (rasterized) on a grid-by-grid basis to obtain the fused resolution data set.

[0100] It should be noted that the weight coefficient in the weighted summation processing is determined through offline regression analysis. The determination process is that the embodiment uses the historical data set, takes the final prediction accuracy as the dependent variable, and takes the time-consistent feature and the resolution multi-dimensional description as the independent variables, calculates the contribution of each to the prediction accuracy through regression analysis, and normalizes it as the weight coefficient.

[0101] It should be noted that the embodiment aligns the spatial grid according to the fused resolution data set. The specific implementation of the alignment is that the embodiment first defines a unified reference spatial grid (for example, a 10m x 10m geographic grid); subsequently, the embodiment uses bilinear interpolation processing. The specific implementation of the processing is that for each new node in the reference spatial grid, the embodiment finds the four closest original data points in the fused resolution data set, and performs weighted averaging according to the values of the four points and their distances from the new node to calculate the value of the new node, and maps the value to the corresponding node of the reference spatial grid, thereby obtaining the resolution consistent description set with consistent spatial coordinates.

[0102] It is worth noting that the embodiment performs spatial smoothness analysis on the resolution-consistent description set (as a raster image). The specific implementation of the analysis is that the embodiment convolves the resolution-consistent description set with a Laplacian operator, calculates its second derivative, and counts the variance of the second derivative result. The variance value quantifies the noise level or "roughness" of the spatial data.

[0103] It is worth noting that the embodiment compares the calculated variance value with the preset accuracy threshold. The preset accuracy threshold is determined by offline analysis, and the determination process is that the embodiment collects a large amount of historical geomorphology raster data known to be "high-precision", and calculates the Laplacian variance, and selects the 95th percentile of the variance distribution as the preset accuracy threshold (representing the upper limit of acceptable noise). If the currently calculated variance value is lower than the preset accuracy threshold (indicating that the data is smooth and the noise is low), the embodiment determines the resolution-consistent description set as the optimized geomorphology model input.

[0104] In step S15, according to the optimized geomorphology model input, geomorphology elevation calculation is performed to determine a high-precision reservoir geomorphology DEM, including:

[0105] Extracting a continuous geomorphology description from the optimized geomorphology model input, and fusing with the water level change scenario simulation bias effect to obtain a bias effect data set;

[0106] According to the bias effect data set, bias correction processing is performed to obtain a bias optimization description set;

[0107] According to the bias optimization description set, geomorphology elevation calculation is performed to obtain the high-precision reservoir geomorphology DEM.

[0108] In one implementation, the embodiment extracts a series of elevation values of grid cells along the same preset path representing the key profile of the reservoir (for example, the center line of the main river channel) from the optimized geomorphology model input (as two-dimensional raster data) obtained in S14 to form the continuous geomorphology description (for example, the elevation sequence [120 meters, 125 meters, 122 meters]), and a series of bias values after spatial interpolation in S14 to form the water level change scenario simulation bias effect (for example, the bias sequence [0.05 meters, 0.07 meters, 0.06 meters]). The embodiment performs weighted sum processing to fuse the continuous geomorphology description (elevation sequence) and the water level change scenario simulation bias effect (bias sequence) point by point according to the spatial position to obtain the bias effect data set.

[0109] It should be noted that the weight coefficients (for example, the geomorphic description weight 0.6, the bias influence weight 0.4) in the weighted sum processing are determined by offline multivariate regression analysis. The determination process is that the embodiment constructs a historical data set containing historical geomorphic description, bias influence data (as independent variables) and corresponding known final flooding prediction accuracy (as dependent variables); the embodiment uses a multiple linear regression (Multiple Linear Regression) model and fits the model by ordinary least squares (Ordinary Least Squares, OLS) to solve the standard regression coefficients of each independent variable; the embodiment normalizes the absolute values of the standard regression coefficients (for example, divides the absolute value of each coefficient by the sum of the absolute values of all coefficients) to obtain the final weight coefficients with a sum of 1.

[0110] In another implementation, the embodiment determines a bias correction indicator set according to the bias influence data set. The specific implementation of the determination is that the embodiment first calculates the bias fluctuation amplitude of the bias influence data set (for example, calculates the standard deviation of the sequence). The embodiment compares the bias fluctuation amplitude with a preset fluctuation threshold.

[0111] It should be noted that the preset fluctuation threshold is determined by statistical analysis of the bias fluctuation amplitude of the historical stable geomorphic data set. The determination process is that the embodiment calculates the probability density distribution of the fluctuation amplitude of the historical data set, and selects the value corresponding to the 95th percentile of the distribution as the preset fluctuation threshold. The percentile is a robust statistical method widely used in system performance engineering, which is based on the fact that this value can effectively cover most (95%) of the normal operation fluctuations while excluding the top 5% of extreme abnormal values, thereby balancing the alarm sensitivity and system stability.

[0112] It is worth noting that if the bias fluctuation amplitude is greater than the preset fluctuation threshold, the embodiment starts bias correction according to a bias correction indicator set, which includes a window size parameter (for example, the window size is 5) for median filtering processing. According to the parameter, the bias influence data set is smoothed to reduce noise interference and obtain the bias optimization description set. If the bias fluctuation amplitude does not exceed the preset fluctuation threshold, the bias influence data set is directly determined as the bias optimization description set.

[0113] In an implementation, the embodiment calculates the geomorphic elevation according to the bias optimization description set. The specific implementation of the calculation is that the embodiment uses a pre-trained geomorphic elevation regression model.

[0114] It should be noted that the geomorphic elevation regression model (for example, a Gradient Boosting Decision Tree (GBDT) model) is determined by offline training. The training process is that the embodiment uses a historical data set containing historical deviation optimization description set (as input features) and corresponding known high-precision geomorphic elevation values (for example, grid cell elevation obtained by satellite radar measurement or LiDAR scanning) as labels. The embodiment uses the data set to supervise the learning of the GBDT model, and the training target is to minimize the root mean squared error (RMSE).

[0115] It should be noted that a key hyperparameter of the GBDT model, such as learning rate (learning_rate) or number of trees (n_estimators), is determined by using a grid search (Grid Search) strategy in the offline training stage, optimizing in the preset parameter space (for example, learning rate range [0.1, 0.01, 0.001]), and selecting the parameter combination with the minimum RMSE value on the independent validation set.

[0116] It is worth noting that after the trained GBDT model receives the deviation optimization description set (which is itself a feature set corresponding to the two-dimensional grid data of S14), the embodiment traverses each grid cell in the optimization geomorphic model input and inputs the feature vector in the deviation optimization description set corresponding to the cell to the GBDT model; the model outputs a predicted elevation value for the grid cell (for example, predicting the elevation of the cell to be 121.5 meters). The collection of predicted elevation values for all grid cells constitutes an elevation grid map (i.e., the DEM generated by the present application) containing (X, Y, Z) spatial position information, which is the high-precision reservoir geomorphic DEM.

[0117] In step S16, according to the high-precision reservoir geomorphic DEM, geomorphic spatial analysis and classification are performed to obtain high-precision reservoir geomorphic collection results, including:

[0118] Calculate the slope gradient according to the high-precision reservoir geomorphic DEM to obtain a slope distribution matrix;

[0119] Simulate the water flow path through the slope distribution matrix and perform path interpolation processing to determine a continuous path chain;

[0120] According to the continuous path chain and the slope distribution matrix, a watershed algorithm is used to divide the area block;

[0121] Classify the region blocks to obtain a landform category group;

[0122] Integrate the collected data according to the landform category group to obtain the high-precision reservoir landform collection result.

[0123] In an implementation manner, the embodiment calculates the slope gradient according to the high-precision reservoir landform DEM obtained in S15. The specific implementation of the calculation is that the embodiment analyzes the grid data in the high-precision DEM, quantifies the terrain inclination of each grid cell by calculating the ratio of the height difference between each grid cell and its adjacent grid cell to the horizontal distance, and finally obtains the slope distribution matrix.

[0124] It is worth noting that the embodiment simulates the water flow path through the slope distribution matrix. The specific implementation of the simulation is that the embodiment uses the D8 (Deterministic 8) flow direction algorithm, which traverses each grid cell in the slope distribution matrix and determines the direction of the water flow (or potential water flow) to the eight adjacent grid cells with the largest slope gradient, thereby generating a water flow path sequence. If there is a path break (for example, encountering a local depression or flat area) in the water flow path sequence, the embodiment uses a linear interpolation algorithm to fill the break point. The specific implementation of the interpolation is that the embodiment identifies the two endpoints (P1 and P2) of the broken path, and generates a series of new intermediate coordinate points between the two points at a fixed step, inserts these new coordinate points into the water flow path sequence, and ensures that the continuous path chain is obtained.

[0125] In an implementation manner, the embodiment divides the region blocks according to the continuous path chain and the slope distribution matrix using a watershed algorithm. The specific implementation of the division is that the embodiment takes the continuous path chain as the "river network" or "pour points" in hydrological analysis, and takes the slope distribution matrix as the basis for flow direction; the watershed algorithm starts from the position of the continuous path chain, reverses the tracking of all grid cells flowing to the path chain, and aggregates all grid cells flowing into the same path chain segment into an independent polygon region, which is a region block (i.e., a sub-basin).

[0126] In an implementation manner, the embodiment classifies the region blocks and uses a pre-trained random forest (Random Forest) classifier.

[0127] It is worth noting that the random forest classifier is determined through offline training. The training process is that the embodiment uses a historical data set containing input features extracted from the region blocks and labeled landform categories (for example, "valley" and "slope") as labels.

[0128] It should be noted that the extraction process of the input features is that, first, according to the high-precision reservoir geomorphology DEM (elevation grid map) obtained in S15, the corresponding slope value, slope direction value and curvature value of each grid cell in the DEM are calculated. The slope value and the slope direction value are obtained by calculating the elevation change rate (first derivative) of each grid cell relative to its adjacent cells. The calculation of the curvature value is achieved by fitting a 3x3 neighborhood local polynomial surface on each grid cell and calculating the second derivative of the surface. Subsequently, for each of the region blocks, the average values of the slope value, the slope direction value and the curvature value of all grid cells in the block are calculated, and these average values are taken as the input features.

[0129] The embodiment adopts the historical data set to supervise the learning of the random forest classifier until the classification accuracy of the model on the independent validation set converges. The judgment criterion for the convergence is that if the improvement amplitude of the validation set accuracy is less than a preset convergence threshold (for example, 1e-4) in a plurality of consecutive training periods (Patience, for example, 10 periods), it is determined that the model has converged. A key hyperparameter of the classifier, for example, the number of decision trees in the forest (n_estimators), is determined by adopting a grid search (Grid Search) strategy in the offline training stage to optimize in a preset parameter space (for example, [50, 100, 200] trees) and selecting the parameter value with the highest F1 score or classification accuracy on the independent validation set. The result output by the random forest classifier is the geomorphology class group.

[0130] Illustratively, the embodiment integrates the collected data according to the geomorphology class group. For example, for a geomorphology class group classified as a valley, the embodiment correlates and integrates the sensor readings of the region collected in S11, the predicted elevation value (for example, 121.5 meters) of the region obtained in S15 and other data, thereby obtaining the high-precision reservoir geomorphology collection result containing rich spatial semantics and physical quantitative indicators.

[0131] In summary, the present application solves the technical problems of low geomorphology data collection accuracy and model distortion in the prior art caused by data format heterogeneity, inconsistent time sequence and model distortion by filtering noise, unifying format and aligning time sequence of multi-source data, introducing historical water level sequences for bias correction, generating a spatio-temporally consistent optimized geomorphology model input, and then generating a high-precision DEM through geomorphology elevation calculation and performing spatial analysis and classification on it.

[0132] Reference Figure 2The second embodiment of the present application provides a reservoir geomorphology data acquisition system based on multi-source data fusion, comprising:

[0133] A data preprocessing module is configured to acquire multi-source remote sensing data and sensor data, and perform preprocessing to obtain a denoised remote sensing sensor unified feature set.

[0134] An association analysis module is configured to perform historical water level feature association analysis according to the denoised remote sensing sensor unified feature set, and obtain a matching deviation index.

[0135] A time sequence adjustment module is configured to perform time sequence alignment adjustment on the matching deviation index, and determine a corrected time sequence consistent feature sequence.

[0136] A fusion processing module is configured to perform remote sensing resolution fusion processing according to the corrected time sequence consistent feature sequence, and obtain an optimized geomorphology model input.

[0137] A geomorphology modeling module is configured to perform geomorphology elevation calculation according to the optimized geomorphology model input, and determine a high-precision reservoir geomorphology DEM.

[0138] A spatial analysis module is configured to perform geomorphology spatial analysis and classification according to the high-precision reservoir geomorphology DEM, and obtain a high-precision reservoir geomorphology acquisition result.

[0139] It should be noted that the reservoir geomorphology data acquisition system based on multi-source data fusion provided by the embodiments of the present application is used to execute all process steps of the reservoir geomorphology data acquisition method based on multi-source data fusion of the above-mentioned embodiments, and the working principles and beneficial effects of the two are one-to-one correspondence, so they will not be repeated here.

[0140] The embodiments of the present application also provide an electronic device. The electronic device comprises a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a reservoir geomorphology data acquisition program based on multi-source data fusion. The processor executes the computer program to implement the steps in each of the above reservoir geomorphology data acquisition methods based on multi-source data fusion, such as the step S11 shown in the figure. Figure 1 Alternatively, the processor executes the computer program to implement the functions of each module / unit in each of the above system embodiments, such as the data preprocessing module.

[0141] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present application. The one or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the electronic device.

[0142] The electronic device can be a computing device such as a desktop computer, a notebook computer, a palm computer, a smart tablet, etc. The electronic device can include, but is not limited to, a processor, a memory. Those skilled in the art can understand that the above components are only examples of the electronic device and do not constitute a limitation on the electronic device, and the electronic device can include more or fewer components than the above, or combine certain components, or different components, for example, the electronic device can also include an input / output device, a network access device, a bus, etc.

[0143] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The processor is the control center of the electronic device, and connects various parts of the electronic device through various interfaces and lines.

[0144] The memory can be used to store the computer program and / or modules, and the processor realizes various functions of the electronic device by running or executing the computer program and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function (such as a sound playing function, an image playing function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, for example, a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state memory device.

[0145] The modules / units integrated in the electronic device, if realized in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. The computer program can implement the steps of each method embodiment when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer readable medium can include any entity or device, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. that can carry the computer program code. It should be noted that the contents included in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0146] It should be noted that the above-described system embodiments are only illustrative, and the units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e., they can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment according to actual needs. In addition, the connection relationship between the modules in the system embodiments provided by the present application indicates that there is a communication connection between them, which can be realized as one or more communication buses or signal lines. Those skilled in the art can understand and implement it without creative labor.

[0147] The above-described specific embodiments further illustrate the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above-described specific embodiments are only examples of the present application and are not intended to limit the protection scope of the present application. It is particularly pointed out that any modification, equivalent replacement, improvement, etc. made by those skilled in the art within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A reservoir geomorphology data acquisition method based on multi-source data fusion, characterized in that, The method comprises the following steps: acquiring multi-source remote sensing data and sensor data, and preprocessing to obtain a denoised remote sensing sensor unified feature set; performing historical water level feature correlation analysis according to the denoised remote sensing sensor unified feature set to obtain a matching deviation index; performing time sequence alignment adjustment on the matching deviation index to determine a corrected time sequence consistent feature sequence; performing remote sensing resolution fusion processing according to the corrected time sequence consistent feature sequence to obtain an optimized topographic model input; performing topographic elevation calculation according to the optimized topographic model input to determine a high-precision reservoir topographic DEM; performing topographic spatial analysis and classification according to the high-precision reservoir topographic DEM to obtain a high-precision reservoir topographic collection result; wherein, according to the denoised remote sensing sensor unified feature set, the historical water level feature correlation analysis is performed to obtain the matching deviation index, which comprises: acquiring a dynamic feature description sequence from the denoised remote sensing sensor unified feature set; performing feature correlation analysis according to the dynamic feature description sequence to obtain a first matching preparation data set; performing historical water level sequence comparison on the first matching preparation data set in combination with the dynamic feature description sequence to determine the matching deviation index; wherein, the time sequence alignment adjustment on the matching deviation index to determine the corrected time sequence consistent feature sequence comprises: performing deviation cumulative value time sequence alignment adjustment on the matching deviation index to obtain a deviation adjustment feature set; using a sliding window translation correction on the deviation adjustment feature set to obtain a time sequence consistent sequence; performing time sequence coherence statistical analysis on the time sequence consistent sequence to determine the corrected time sequence consistent feature sequence; wherein, the time sequence alignment adjustment on the matching deviation index comprises: calculating the cumulative sum of the matching deviation index within a time window to obtain a deviation cumulative value; comparing the deviation cumulative value with a cumulative threshold value determined by statistical analysis of historical data, if the deviation cumulative value exceeds the preset cumulative threshold value, then performing minimum-maximum normalization processing on the deviation index sequence within the time window, mapping the sequence value linearly to a new interval of [0, the preset cumulative threshold value] by multiplying the difference between the original value and the sequence minimum value by the preset cumulative threshold value and then dividing the product by the difference between the sequence maximum value and the sequence minimum value, thereby obtaining the deviation adjustment feature set; wherein, using a sliding window translation correction on the deviation adjustment feature set to obtain a time sequence consistent sequence comprises: using a sliding average filter with a preset window size to traverse the deviation adjustment feature set point by point; replacing the arithmetic mean value of all data points in the window with a new value of the center point of the window to smooth the local noise peaks in the sequence and obtain the time sequence consistent sequence; performing time sequence coherence statistical analysis on the time sequence consistent sequence to calculate the first-order difference of the time sequence consistent sequence and obtain the variance of the first-order difference sequence; comparing the calculated variance value with a preset coherence threshold value.

2. The reservoir geomorphic data acquisition method based on multi-source data fusion according to claim 1, characterized in that, The acquisition multi-source remote sensing data and sensor data, and pre-processing, get denoising after remote sensing sensor uniform feature set, including: Acquire the multi-source remote sensing data and sensor data, and carry out noise filtering and format heterogeneous conversion, obtain multi-source fusion initial data; According to the multi-source fusion initial data, extract remote sensing image band selection and sensor reading time stamp, and carry out time stamp alignment and clustering grouping denoising processing, determine the denoising after remote sensing sensor uniform feature set.

3. The reservoir geomorphic data acquisition method based on multi-source data fusion according to claim 1, characterized in that, According to the corrected time sequence consistent feature sequence, the remote sensing resolution fusion processing is carried out, and the optimized geomorphology model input is obtained, including: For the corrected time sequence consistent feature sequence, the remote sensing image resolution multi-dimensional description integration is carried out, and the fusion resolution data set is obtained; According to the fusion resolution data set, the resolution consistent description set is obtained by spatial grid alignment; The spatial smoothness analysis is carried out on the resolution consistent description set, and the optimized geomorphology model input is determined.

4. The reservoir geomorphology data acquisition method based on multi-source data fusion according to claim 1, characterized in that, According to the optimized geomorphology model input, the geomorphology elevation calculation is carried out, and the high-precision reservoir geomorphology DEM is determined, including: From the optimized geomorphology model input, the continuous geomorphology description is extracted, and the deviation influence data set is obtained by fusion processing with water level change scene simulation deviation influence; According to the deviation influence data set, the deviation correction processing is carried out, and the deviation optimization description set is obtained; According to the deviation optimization description set, the geomorphology elevation calculation is carried out, and the high-precision reservoir geomorphology DEM is obtained.

5. The reservoir geomorphology data acquisition method based on multi-source data fusion according to claim 1, characterized in that, According to the high-precision reservoir geomorphology DEM, the geomorphology spatial analysis and classification are carried out, and the high-precision reservoir geomorphology acquisition result is obtained, including: According to the high-precision reservoir geomorphology DEM, the slope gradient is calculated, and the slope distribution matrix is obtained; The water flow path is simulated through the slope distribution matrix, and the path interpolation processing is carried out to determine the continuous path chain; According to the continuous path chain and the slope distribution matrix, the region block is divided by using the watershed algorithm; The region block is classified to obtain the geomorphology category group; According to the geomorphology category group, the acquisition data is integrated to obtain the high-precision reservoir geomorphology acquisition result.

6. A reservoir geomorphology data acquisition system based on multi-source data fusion, used to implement the reservoir geomorphology data acquisition method based on multi-source data fusion according to any one of claims 1-5. Including: Data preprocessing module, for acquiring multi-source remote sensing data and sensor data, and pre-processing, get denoising after remote sensing sensor uniform feature set; Correlation analysis module, for according to the denoising after remote sensing sensor uniform feature set, the historical water level feature correlation analysis is carried out, and the matching deviation index is obtained; Time sequence adjustment module, for adjusting the time sequence according to the matching deviation index, to determine the corrected time sequence consistent feature sequence; Fusion processing module, for according to the corrected time sequence consistent feature sequence, the remote sensing resolution fusion processing is carried out, and the optimized geomorphology model input is obtained; Geomorphology modeling module, for according to the optimized geomorphology model input, the geomorphology elevation calculation is carried out, and the high-precision reservoir geomorphology DEM is determined; Spatial analysis module, for according to the high-precision reservoir geomorphology DEM, the geomorphology spatial analysis and classification are carried out, and the high-precision reservoir geomorphology acquisition result is obtained.

Images