Tidal flat terrain deformation prediction method, system, electronic equipment and medium

Abstract

The present invention discloses a tidal flat terrain deformation prediction method, system, electronic device, and medium, relating to the technical field of terrain prediction. The method comprises calculating the gray correlation between the dataset of each influencing factor and the terrain dataset based on a terrain dataset and datasets of various influencing factors; obtaining a set of relevant influencing factors based on the gray correlation; eliminating relevant influencing factors from the set of relevant influencing factors based on the correlation between each two relevant influencing factors; determining the weight of each relevant influencing factor in the eliminated set of relevant influencing factors using an entropy weight method; decomposing the weighted values of each relevant influencing factor using a Fourier transform to obtain a frequency set corresponding to each relevant influencing factor; training a neural network using the frequency as input and the terrain change amount as output to obtain a prediction network; and using the prediction network to predict terrain deformation. The present invention can improve the prediction accuracy of terrain deformation.

Description

Technical Field

[0001] The present invention relates to the technical field of terrain prediction, and in particular to a method, system, electronic equipment and medium for predicting tidal flat terrain deformation. Background Art

[0002] Existing prediction technologies are all based on data obtained from non-contact remote sensing information. Remote sensing cannot conduct all-weather and all-angle photography, and cannot take environmental factors and soil properties into account, so the prediction accuracy is relatively low. Summary of the Invention

[0003] The purpose of the present invention is to provide a method, system, electronic equipment and medium for predicting tidal flat landform deformation, which can improve the prediction accuracy of landform deformation.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] A method for predicting tidal flat landform deformation, comprising:

[0006] The terrain changes and the values of various influencing factors of the target area at multiple moments in a historical period are obtained at set time intervals to obtain a terrain dataset and a dataset of various influencing factors; the influencing factors include: overlying water pressure, temperature, wind speed, precipitation, sea level pressure, soil evaporation, tide level, waves, soil friction angle, soil cohesion, soil dry density, soil elastic modulus, soil Poisson's ratio, and soil moisture content;

[0007] Calculating the grey correlation degree between the dataset of each influencing factor and the terrain dataset based on the terrain dataset and the dataset of each influencing factor;

[0008] Determine the influencing factors corresponding to the target grey correlation degree as relevant influencing factors to obtain a set of relevant influencing factors; the target grey correlation degree is a grey correlation degree whose value is greater than a set threshold;

[0009] Eliminating relevant influencing factors in the relevant influencing factor set according to the correlation between every two relevant influencing factors in the relevant influencing factor set to obtain a set of relevant influencing factors after elimination;

[0010] The entropy weight method is used to determine the weight of each relevant influencing factor in the set of relevant influencing factors after elimination, and the weighted value of each relevant influencing factor is obtained according to the weight of each relevant influencing factor;

[0011] Using Fourier transform to decompose the weighted value of each relevant influencing factor to obtain a frequency set corresponding to each relevant influencing factor; each frequency set includes multiple frequencies;

[0012] Decomposing the terrain variation amounts of each region in the terrain data set in the frequency set respectively to obtain terrain variation amounts corresponding to all relevant influencing factors;

[0013] For any frequency in the frequency set corresponding to all relevant influencing factors, the neural network is trained with the frequency as input and the terrain change amounts corresponding to all relevant influencing factors corresponding to the frequency as output to obtain a prediction network corresponding to the frequency;

[0014] The prediction network corresponding to each frequency is used to predict the terrain deformation of the prediction area.

[0015] Optionally, the calculating, based on the terrain dataset and the datasets of each influencing factor, the grey correlation degree between the dataset of each influencing factor and the terrain dataset specifically includes:

[0016] Performing dimensionless processing on the terrain dataset and the datasets of each influencing factor to obtain a processed terrain dataset and a processed dataset of each influencing factor;

[0017] The grey correlation degree between the dataset of each influencing factor and the terrain dataset is calculated based on the processed terrain dataset and the dataset of each influencing factor.

[0018] Optionally, the calculating of the grey correlation between the dataset of each influencing factor and the terrain dataset based on the processed terrain dataset and the datasets processed by each influencing factor specifically includes:

[0019] For any moment, the grey correlation degree between the i-th influencing factor corresponding to the moment and the terrain change amount in the processed terrain dataset is obtained according to the terrain change amount at the moment in the processed terrain dataset and the value of the influencing factor at the moment in the dataset after the i-th influencing factor is processed;

[0020] The grey correlation degree between the dataset of the i-th influencing factor and the terrain dataset is obtained according to the grey correlation degree between the i-th influencing factor and the terrain change amount in the processed terrain dataset corresponding to all moments.

[0021] Optionally, the prediction network corresponding to each frequency is used to predict the terrain deformation of the prediction area, specifically including:

[0022] Obtain the frequency set corresponding to all relevant influencing factors of the area to be predicted;

[0023] For any frequency in the frequency set corresponding to all relevant influencing factors of the area to be predicted, the frequency is input into the target prediction network to obtain the terrain deformation components corresponding to all relevant influencing factors corresponding to the frequency; the target prediction network is the prediction network corresponding to the frequency;

[0024] The terrain deformation of the area to be predicted is obtained by superimposing the terrain deformation components corresponding to all relevant influencing factors corresponding to all frequencies in the frequency set corresponding to all relevant influencing factors of the area to be predicted.

[0025] A tidal flat terrain deformation prediction system, comprising:

[0026] An acquisition module is used to obtain the terrain changes and the values of various influencing factors of the target area at multiple times at set time intervals within a historical time period to obtain a terrain dataset and a dataset of various influencing factors; the influencing factors include: overlying water pressure, temperature, wind speed, precipitation, sea level pressure, soil evaporation, tide level, waves, soil friction angle, soil cohesion, soil dry density, soil elastic modulus, soil Poisson's ratio, and soil moisture content;

[0027] A grey correlation calculation module is used to calculate the grey correlation between the dataset of each influencing factor and the terrain dataset based on the terrain dataset and the dataset of each influencing factor;

[0028] A relevant influencing factor set determination module is used to determine the influencing factors corresponding to the target grey correlation degree as relevant influencing factors, and obtain a relevant influencing factor set; the target grey correlation degree is a grey correlation degree whose value is greater than a set threshold;

[0029] a removal module, configured to remove relevant influencing factors from the relevant influencing factor set according to the correlation between every two relevant influencing factors in the relevant influencing factor set, to obtain a set of relevant influencing factors after removal;

[0030] A weight calculation module is used to determine the weight of each relevant influencing factor in the set of relevant influencing factors after elimination by using an entropy weight method, and obtain a weighted value of each relevant influencing factor according to the weight of each relevant influencing factor;

[0031] A decomposition module is used to decompose the weighted value of each relevant influencing factor using Fourier transform to obtain a frequency set corresponding to each relevant influencing factor; each frequency set includes multiple frequencies;

[0032] A terrain change decomposition module is used to decompose the terrain change amounts of each region in the terrain data set in the frequency set to obtain terrain change amounts corresponding to all relevant influencing factors;

[0033] A network training module is used to train a neural network for any frequency in the frequency set corresponding to all relevant influencing factors, using the frequency as input and the terrain change corresponding to all relevant influencing factors corresponding to the frequency as output to obtain a prediction network corresponding to the frequency;

[0034] The prediction module is used for the prediction network corresponding to each frequency to predict the terrain deformation of the prediction area.

[0035] Optionally, the grey relational degree calculation module specifically includes:

[0036] A dimensionless unit, configured to perform dimensionless processing on the terrain dataset and the datasets of the influencing factors to obtain processed terrain datasets and processed datasets of the influencing factors;

[0037] The grey relational degree calculation unit is used to calculate the grey relational degree between the dataset of each influencing factor and the terrain dataset based on the processed terrain dataset and the dataset processed by each influencing factor.

[0038] Optionally, the grey relational degree calculation unit specifically includes:

[0039] The grey relational degree calculation subunit at any moment is used to obtain, for any moment, the grey relational degree between the i-th influencing factor corresponding to the moment and the terrain change amount in the processed terrain dataset according to the terrain change amount at the moment in the processed terrain dataset and the value of the influencing factor at the moment in the dataset after processing the i-th influencing factor;

[0040] The total grey correlation calculation subunit is used to obtain the grey correlation between the dataset of the i-th influencing factor and the terrain dataset according to the grey correlation between the i-th influencing factor and the terrain change in the processed terrain dataset at all times.

[0041] Optionally, the prediction module specifically includes:

[0042] An acquisition unit, configured to acquire a frequency set corresponding to all relevant influencing factors of the area to be predicted;

[0043] A prediction unit is configured to input any frequency in a set of frequencies corresponding to all relevant influencing factors of the area to be predicted into a target prediction network to obtain terrain deformation components corresponding to all relevant influencing factors corresponding to the frequency; the target prediction network is a prediction network corresponding to the frequency;

[0044] The superposition unit is configured to superpose the terrain deformation components corresponding to all relevant influencing factors corresponding to all frequencies in the frequency set corresponding to all relevant influencing factors of the to-be-predicted area to obtain the terrain deformation of the to-be-predicted area.

[0045] An electronic device, comprising:

[0046] A memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the tidal flat landform deformation prediction method described above.

[0047] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the above-mentioned method for predicting tidal flat landform deformation.

[0048] According to the specific embodiment provided by the present invention, the present invention discloses the following technical effects: the present invention calculates the gray correlation degree between the data set of each influencing factor and the terrain data set based on the terrain data set and the data set of each influencing factor; determines the influencing factor corresponding to the target gray correlation degree as the relevant influencing factor, and obtains a set of relevant influencing factors; the target gray correlation degree is the gray correlation degree whose value is greater than the set threshold; the relevant influencing factors in the relevant influencing factor set are eliminated according to the correlation between every two relevant influencing factors in the relevant influencing factor set, and obtain a set of relevant influencing factors after elimination; the entropy weight method is used to determine the weight of each relevant influencing factor in the set of relevant influencing factors after elimination; the Fourier transform is used to decompose the weighted value of each relevant influencing factor to obtain a frequency set corresponding to each relevant influencing factor; each frequency set includes multiple frequencies; for any frequency, the frequency is used as input and the terrain change amount corresponding to the frequency is used as output to train the neural network to obtain a prediction network corresponding to the frequency; the prediction network corresponding to each frequency is used to predict the terrain deformation of the predicted area, which can improve the prediction accuracy of the terrain deformation. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0050] Figure 1 A flowchart of a method for predicting tidal flat landform deformation provided by an embodiment of the present invention;

[0051] Figure 2 This is a diagram of the neural network structure. DETAILED DESCRIPTION

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0053] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0054] The embodiment of the present invention provides a method for predicting tidal flat terrain deformation. The general steps are: 1. Select influencing factors for terrain deformation prediction based on the correlation between different influencing factors and terrain deformation; 2. Perform correlation analysis on the factors selected in step 1, perform dimensionality reduction processing on the influencing factors, and finally determine the influencing factors; 3. Calculate the weight coefficient of the influencing factor determined in step 2, and multiply the numerical value of the influencing factor by the weight coefficient to obtain the weighted value of the influencing factor; 4. Decompose the terrain change and the weighted value of the influencing factor determined in step 3 into different frequency components; 5. Use the weighted value of the influencing factor and the terrain change at the same frequency to train a nonlinear autoregressive neural network with external input; 6. Use the neural network trained in step 5 to predict tidal flat terrain deformation. Figure 1 As shown, the specific steps include:

[0055] Step 101: Obtain the terrain change amount and the values of each influencing factor of the target area at multiple times at set time intervals within the historical time period to obtain a terrain dataset and a dataset of each influencing factor. The influencing factors include: environmental factors and soil property factors. Environmental factors include: overlying water pressure, temperature, wind speed, precipitation, sea level pressure, soil evaporation, tide level, and waves. Soil property factors include: friction angle, cohesion, dry density, elastic modulus, Poisson's ratio, and moisture content. The above influencing factors are respectively denoted as f 1 ,f 2 ,...,f 14 , and the terrain change is recorded as h .

[0056] Step 102: Calculate the grey correlation degree between the dataset of each influencing factor and the terrain dataset based on the terrain dataset and the dataset of each influencing factor.

[0057] Step 103: determining the influencing factors corresponding to the target grey correlation degree as relevant influencing factors to obtain a set of relevant influencing factors; the target grey correlation degree is a grey correlation degree whose value is greater than a set threshold.

[0058] Step 104: Eliminate the relevant influencing factors in the relevant influencing factor set according to the correlation between every two relevant influencing factors in the relevant influencing factor set to obtain a set of relevant influencing factors after elimination.

[0059] Step 105: using an entropy weight method to determine the weight of each relevant influencing factor in the eliminated relevant influencing factor set, and obtaining a weighted value of each relevant influencing factor according to the weight of each relevant influencing factor.

[0060] Step 106: Decompose the weighted value of each relevant influencing factor using Fourier transform to obtain a frequency set corresponding to each relevant influencing factor; each frequency set includes multiple frequencies.

[0061] Step 107: Decompose the terrain variation amounts of each region in the terrain data set in the frequency set respectively to obtain terrain variation amounts corresponding to all relevant influencing factors.

[0062] Step 108: For any frequency in the frequency set corresponding to all relevant influencing factors, the frequency is used as input and the terrain change corresponding to all relevant influencing factors corresponding to the frequency is used as output to train the neural network to obtain the prediction network corresponding to the frequency. One neural network is trained for one frequency data. The neural network structure is as follows: Figure 2 As shown in the figure, this neural network is a feedback-based neural network consisting of one input layer, five hidden layers, and one output layer. Only one hidden layer is shown for illustration. The activation function of each neuron in the hidden layer is a sigmoid function, while the activation function of each neuron in the output layer is a linear function. Neural networks trained on data at different frequencies predict the terrain deformation component at that frequency.

[0063] Step 109: The prediction network corresponding to each frequency is used to predict the terrain deformation of the prediction area.

[0064] In actual applications, the influencing factors in step 101 are obtained by using multiple sensors to synchronously obtain environmental factors and taking soil samples for experimental analysis at regular intervals to obtain soil property factors. Assuming that the prediction time step is set to ΔT, the historical data of the tidal flat topography of the target area, i.e., the terrain dataset, and the historical data of the factors affecting the tidal flat topography, i.e., the datasets of each influencing factor, are divided into discrete sequences with a time interval of ΔT. The terrain dataset is represented as The dataset of the i-th influencing factor is expressed as The sequence superscript i=1,2,...,14 is the influencing factor number, and the sequence subscript t n The collection time interval of two adjacent data in the same data set should satisfy t n -t n-1 =ΔT.

[0065] In practical applications, step 102 specifically includes:

[0066] Step 201: performing dimensionless processing on the terrain dataset and the datasets of each influencing factor to obtain a processed terrain dataset and a processed dataset of each influencing factor.

[0067] Step 202: Calculate the grey correlation degree between the dataset of each influencing factor and the terrain dataset based on the processed terrain dataset and the dataset of each influencing factor.

[0068] In practical applications, step 201 is specifically as follows:

[0069] The data were dimensionless processed using the mean method. The reference sequence and comparison sequence after processing were and

[0070] In practical applications, step 202 specifically includes:

[0071] Step 220: For any moment, the grey correlation degree between the i-th influencing factor corresponding to the moment and the terrain change amount in the processed terrain dataset is obtained according to the terrain change amount at the moment in the processed terrain dataset and the value of the influencing factor at the moment in the dataset processed by the i-th influencing factor.

[0072] Step 221: obtaining the grey correlation degree between the dataset of the i-th influencing factor and the terrain dataset according to the grey correlation degree between the i-th influencing factor and the terrain change amount in the processed terrain dataset corresponding to all moments.

[0073] In practical applications, step 220 is specifically as follows:

[0074] For the i-th influencing factor, the difference sequence, maximum difference and minimum difference are calculated according to the following three formulas.

[0075] Difference sequence:

[0076] Maximum difference:

[0077] Minimum difference:

[0078] According to the formula

[0079]

[0080] Determine t k The influencing factor at time i and t kThe grey correlation degree between the terrain changes at each moment, λ∈(0,1) is the resolution coefficient, which can be taken as 0.5.

[0081] In practical applications, step 221 is specifically as follows:

[0082] According to the formula Calculate the dataset F of the i-th influencing factor i Grey relational degree with H.

[0083] In practical applications, the correlation between every two relevant influencing factors in the relevant influencing factor set is calculated, and the relevant influencing factors in the relevant influencing factor set are eliminated according to the correlation between every two relevant influencing factors in the relevant influencing factor set. The relevant influencing factor set after elimination is specifically:

[0084] Assume that after screening, there are m influencing factors whose grey correlation degree is greater than the set threshold, that is, there are m influencing factors in the relevant influencing factor set, then calculate the correlation between these m influencing factors. The correlation calculation method is as follows:

[0085] There are m influencing factors in total, each of which is a time series containing n data, which are combined to form a data matrix Column a in the matrix and column b The calculation formula of the correlation coefficient rho(a,b) is as follows:

[0086]

[0087] in and are the averages of columns a and b, respectively.

[0088] If the correlation coefficient is greater than 0.85, it indicates that the linear correlation between the two factors is extremely high. One of the two factors is eliminated. The role of this step is to reduce the dimension of the influencing factors selected in step 101 to increase the calculation speed.

[0089] The entropy weight method is an objective weighting method that can make the most of the data of the index system to calculate the weight value of each index. For a certain influencing factor i, if the element in the influencing factor to be evaluated is The greater the gap between them, the smaller the entropy value of the influencing factor. According to the information entropy theory, the greater the amount of information that the influencing factor can provide, the greater the impact on the predicted terrain change, and the higher the weight should be.

[0090] In practical applications, step 105 is specifically as follows:

[0091] Step 1: Calculate the t under the i-th influencing factork The characteristic weight of the influencing factor values obtained at each moment

[0092]

[0093] Step 2: Calculation of entropy. Information entropy is a measure of the degree of disorder in a system, and its expression is:

[0094]

[0095] In the formula, when η ik =0, η ik ln(η ik )=0.

[0096] Step 3: Calculation of weights

[0097] The entropy weight of the i-th influencing factor is

[0098] The entropy weight of each influencing factor is the weight of the corresponding influencing factor. The weight of the influencing factor is multiplied by the value in the corresponding influencing factor data set to obtain the weighted value of the relevant influencing factor, that is, the weighted value of the i-th influencing factor: X i =F i ·ω i .

[0099] In practical applications, step 106 is specifically as follows:

[0100] Use Fourier transform to determine the frequency domain information of tidal flat topography deformation:

[0101]

[0102] where X s (f s ) indicates the frequency value is f s Spectrum value of X i (n) is the time series data of the weighted value of the influencing factors, which is composed of X at n moments i Composition; j is an imaginary unit, that is t n is the total length of the sequence, ΔT is the time step of the prediction; since the frequency of terrain changes is low and the cycle is long, the prediction time step is divided by 3600 to compress the change cycle. s (f s The frequency values that account for more than 20% of the total spectrum are determined to be the frequency values where the tidal flat topography changes are concentrated. Based on these frequency values, the tidal flat topography changes and the weighted values of the influencing factors are decomposed into components of different frequencies. The specific calculation method is as follows:

[0103]

[0104] H(n) is the change in tidal flat topography before transformation; H s (n) is the change of tidal flat topography at frequency f w The component under; j is the imaginary unit, that is t n is the total length of the sequence, and ΔT is the time step of the prediction.

[0105]

[0106] X i (n) is the weighted value of the influencing factors before transformation; is the weighted value of the influencing factor at frequency f w The component under; j is the imaginary unit, that is t n is the total length of the sequence, and ΔT is the time step of the prediction.

[0107] In practical applications, step 108 is specifically as follows:

[0108] In addition to analyzing the time series of terrain deformation, predicting tidal flat terrain deformation also requires considering the influence of environmental factors and soil properties. The nonlinear autoregressive neural network model with external input is suitable for this situation.

[0109] At a given frequency, select the external input sequence of the time series neural network of the weighted values of the influencing factors calculated in step 3, the time series of terrain changes as the output sequence, the number of hidden layers is p, and the time lag is q. The network structure is as follows:

[0110] The data obtained in step 106 and the corresponding terrain changes are used to train the neural network according to the 70% training set, 15% validation set and 15% test set, and the R-squared error R is calculated according to the following formula: 2 :

[0111]

[0112] In the above formula, y represents the true value of terrain deformation, Represents the terrain deformation prediction value, output by the prediction network, This represents the mean of the true values of terrain deformation. Specifically, the numerator represents the sum of the squared differences between the true and predicted values, while the denominator represents the sum of the squared differences between the true and mean values. Values closer to 1 indicate greater accuracy. If the R-squared error is less than 0.75, consider increasing the number of hidden layers p or the time lag q to improve prediction accuracy.

[0113] In practical applications, step 109 specifically includes:

[0114] Get the frequency set corresponding to all relevant influencing factors of the area to be predicted.

[0115] For any frequency in the frequency set corresponding to all relevant influencing factors of the area to be predicted, the frequency is input into the target prediction network to obtain the terrain deformation components corresponding to all relevant influencing factors corresponding to the frequency; the target prediction network is the prediction network corresponding to the frequency.

[0116] The terrain deformation of the area to be predicted is obtained by superimposing the terrain deformation components corresponding to all relevant influencing factors corresponding to all frequencies in the frequency set corresponding to all relevant influencing factors of the area to be predicted.

[0117] In view of the above method, the present invention provides a tidal flat landform deformation prediction system, comprising:

[0118] The acquisition module is used to obtain the terrain changes and the values of various influencing factors of the target area at multiple times at set time intervals within a historical time period to obtain a terrain dataset and a dataset of various influencing factors; the influencing factors include: overlying water pressure, temperature, wind speed, precipitation, sea level pressure, soil evaporation, tide level, waves, soil friction angle, soil cohesion, soil dry density, soil elastic modulus, soil Poisson's ratio and soil moisture content.

[0119] The grey relational degree calculation module is used to calculate the grey relational degree between the dataset of each influencing factor and the terrain dataset based on the terrain dataset and the datasets of each influencing factor.

[0120] The relevant influencing factor set determination module is used to determine the influencing factors corresponding to the target grey correlation degree as relevant influencing factors to obtain a relevant influencing factor set; the target grey correlation degree is a grey correlation degree whose value is greater than a set threshold.

[0121] The elimination module is configured to eliminate the relevant influencing factors in the relevant influencing factor set according to the correlation between every two relevant influencing factors in the relevant influencing factor set, so as to obtain a set of relevant influencing factors after elimination.

[0122] The weight calculation module is used to determine the weight of each relevant influencing factor in the set of relevant influencing factors after elimination by using the entropy weight method, and obtain the weighted value of each relevant influencing factor according to the weight of each relevant influencing factor.

[0123] The decomposition module is used to decompose the weighted value of each relevant influencing factor by using Fourier transform to obtain a frequency set corresponding to each relevant influencing factor; each frequency set includes multiple frequencies.

[0124] The terrain change decomposition module is used to decompose the terrain change amounts of each region in the terrain data set in the frequency set to obtain terrain change amounts corresponding to all relevant influencing factors.

[0125] The network training module is used to train a neural network for any frequency in the frequency set corresponding to all relevant influencing factors, using the frequency as input and the terrain change corresponding to all relevant influencing factors corresponding to the frequency as output to obtain a prediction network corresponding to the frequency.

[0126] The prediction module is used for the prediction network corresponding to each frequency to predict the terrain deformation of the prediction area.

[0127] In practical applications, the grey relational degree calculation module specifically includes:

[0128] The dimensionless unit is used to perform dimensionless processing on the terrain dataset and the datasets of each influencing factor to obtain a processed terrain dataset and a processed dataset of each influencing factor.

[0129] The grey relational degree calculation unit is used to calculate the grey relational degree between the dataset of each influencing factor and the terrain dataset based on the processed terrain dataset and the dataset processed by each influencing factor.

[0130] In practical applications, the grey relational degree calculation unit specifically includes:

[0131] The grey correlation degree calculation subunit at any moment is used to obtain, for any moment, the grey correlation degree between the i-th influencing factor corresponding to the moment and the terrain change amount in the processed terrain dataset according to the terrain change amount at the moment in the processed terrain dataset and the value of the influencing factor at the moment in the dataset processed by the i-th influencing factor.

[0132] The total grey correlation calculation subunit is used to obtain the grey correlation between the dataset of the i-th influencing factor and the terrain dataset according to the grey correlation between the i-th influencing factor and the terrain change in the processed terrain dataset at all times.

[0133] In practical applications, the prediction module specifically includes:

[0134] The acquisition unit is used to obtain the frequency set corresponding to all relevant influencing factors of the area to be predicted.

[0135] The prediction unit is used to input any frequency in the frequency set corresponding to all relevant influencing factors of the area to be predicted into the target prediction network to obtain the terrain deformation components corresponding to all relevant influencing factors corresponding to the frequency; the target prediction network is the prediction network corresponding to the frequency.

[0136] The superposition unit is configured to superpose the terrain deformation components corresponding to all relevant influencing factors corresponding to all frequencies in the frequency set corresponding to all relevant influencing factors of the to-be-predicted area to obtain the terrain deformation of the to-be-predicted area.

[0137] An embodiment of the present invention further provides an electronic device, including:

[0138] A memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute a tidal flat landform deformation prediction method according to the above embodiment.

[0139] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the method for predicting tidal flat landform deformation as described in the above embodiment is implemented.

[0140] The present invention has the following technical effects:

[0141] 1. The factors affecting the topographic deformation of tidal flats in different locations and the degree of their influence are slightly different. The present invention can screen the factors affecting the topographic deformation of tidal flats according to local conditions and construct a prediction model suitable for the topographic deformation of tidal flats in different locations.

[0142] 2. According to the periodic change characteristics of tidal flat topography, component decomposition and prediction can be carried out to improve the prediction accuracy.

[0143] 3. Since the main factors of tidal flat topography changes in different regions are different, the present invention first determines the contribution degree of tidal flat topography changes and selects influencing factors using the grey correlation method. After the calculation is completed, if the grey correlation degree is greater than the set threshold, it is considered that the influencing factor has a large contribution to the topography change and is considered in the subsequent prediction steps. Otherwise, this factor is not considered in the subsequent prediction steps, thereby improving the running speed and prediction accuracy.

[0144] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0145] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for predicting tidal flat landform deformation, characterized in that: include: The terrain change amount and the values of each influencing factor of the target area at multiple moments are obtained at set time intervals within the historical time period to obtain a terrain dataset and a dataset of each influencing factor; The influencing factors include: overlying water pressure, temperature, wind speed, precipitation, sea level pressure, soil evaporation, tide level, waves, soil friction angle, soil cohesion, soil dry density, soil elastic modulus, soil Poisson's ratio and soil moisture content; Calculating the grey correlation degree between the dataset of each influencing factor and the terrain dataset based on the terrain dataset and the dataset of each influencing factor; Determine the influencing factors corresponding to the target grey correlation degree as relevant influencing factors to obtain a set of relevant influencing factors; the target grey correlation degree is a grey correlation degree whose value is greater than a set threshold; Eliminating relevant influencing factors in the relevant influencing factor set according to the correlation between every two relevant influencing factors in the relevant influencing factor set to obtain a set of relevant influencing factors after elimination; The entropy weight method is used to determine the weight of each relevant influencing factor in the set of relevant influencing factors after elimination, and the weighted value of each relevant influencing factor is obtained according to the weight of each relevant influencing factor; Using Fourier transform to decompose the weighted value of each relevant influencing factor to obtain a frequency set corresponding to each relevant influencing factor; each frequency set includes multiple frequencies; Decomposing the terrain variation amounts of each region in the terrain data set in the frequency set respectively to obtain terrain variation amounts corresponding to all relevant influencing factors; For any frequency in the frequency set corresponding to all relevant influencing factors, the neural network is trained with the frequency as input and the terrain change amounts corresponding to all relevant influencing factors corresponding to the frequency as output to obtain a prediction network corresponding to the frequency; The prediction network corresponding to each frequency is used to predict the terrain deformation of the prediction area.

2. A tidal flat landform deformation prediction method according to claim 1, characterized in that: The step of calculating the grey correlation between the dataset of each influencing factor and the terrain dataset based on the terrain dataset and the dataset of each influencing factor specifically includes: Performing dimensionless processing on the terrain dataset and the datasets of each influencing factor to obtain a processed terrain dataset and a processed dataset of each influencing factor; The grey correlation degree between the dataset of each influencing factor and the terrain dataset is calculated based on the processed terrain dataset and the dataset of each influencing factor.

3. A tidal flat landform deformation prediction method according to claim 2, characterized in that: The calculating of the grey correlation degree between the dataset of each influencing factor and the terrain dataset based on the processed terrain dataset and the datasets processed by each influencing factor specifically includes: For any moment, the grey correlation degree between the i-th influencing factor corresponding to the moment and the terrain change amount in the processed terrain dataset is obtained according to the terrain change amount at the moment in the processed terrain dataset and the value of the influencing factor at the moment in the dataset after the i-th influencing factor is processed; The grey correlation degree between the dataset of the i-th influencing factor and the terrain dataset is obtained according to the grey correlation degree between the i-th influencing factor and the terrain change amount in the processed terrain dataset corresponding to all moments.

4. The method for predicting tidal flat landform deformation according to claim 1, wherein: The prediction network corresponding to each frequency is used to predict the terrain deformation of the prediction area, specifically including: Obtain the frequency set corresponding to all relevant influencing factors of the area to be predicted; For any frequency in the frequency set corresponding to all relevant influencing factors of the area to be predicted, the frequency is input into the target prediction network to obtain the terrain deformation components corresponding to all relevant influencing factors corresponding to the frequency; the target prediction network is the prediction network corresponding to the frequency; The terrain deformation of the area to be predicted is obtained by superimposing the terrain deformation components corresponding to all relevant influencing factors corresponding to all frequencies in the frequency set corresponding to all relevant influencing factors of the area to be predicted.

5. A tidal flat landform deformation prediction system, characterized in that: include: An acquisition module is used to obtain the terrain change amount and the value of each influencing factor of the target area at multiple moments according to a set time interval within a historical time period to obtain a terrain dataset and a dataset of each influencing factor; The influencing factors include: overlying water pressure, temperature, wind speed, precipitation, sea level pressure, soil evaporation, tide level, waves, soil friction angle, soil cohesion, soil dry density, soil elastic modulus, soil Poisson's ratio and soil moisture content; A grey correlation calculation module is used to calculate the grey correlation between the dataset of each influencing factor and the terrain dataset based on the terrain dataset and the dataset of each influencing factor; A relevant influencing factor set determination module is used to determine the influencing factors corresponding to the target grey correlation degree as relevant influencing factors, and obtain a relevant influencing factor set; the target grey correlation degree is a grey correlation degree whose value is greater than a set threshold; a removal module, configured to remove relevant influencing factors from the relevant influencing factor set according to the correlation between every two relevant influencing factors in the relevant influencing factor set, to obtain a set of relevant influencing factors after removal; A weight calculation module is used to determine the weight of each relevant influencing factor in the set of relevant influencing factors after elimination by using an entropy weight method, and obtain a weighted value of each relevant influencing factor according to the weight of each relevant influencing factor; A decomposition module is used to decompose the weighted value of each relevant influencing factor using Fourier transform to obtain a frequency set corresponding to each relevant influencing factor; each frequency set includes multiple frequencies; A terrain change decomposition module is used to decompose the terrain change amounts of each region in the terrain data set in the frequency set to obtain terrain change amounts corresponding to all relevant influencing factors; A network training module is used to train a neural network for any frequency in the frequency set corresponding to all relevant influencing factors, using the frequency as input and the terrain change corresponding to all relevant influencing factors corresponding to the frequency as output to obtain a prediction network corresponding to the frequency; The prediction module is used for the prediction network corresponding to each frequency to predict the terrain deformation of the prediction area.

6. A tidal flat landform deformation prediction system according to claim 5, characterized in that: The grey relational degree calculation module specifically includes: A dimensionless unit, configured to perform dimensionless processing on the terrain dataset and the datasets of the influencing factors to obtain processed terrain datasets and processed datasets of the influencing factors; The grey relational degree calculation unit is used to calculate the grey relational degree between the dataset of each influencing factor and the terrain dataset based on the processed terrain dataset and the dataset processed by each influencing factor.

7. A tidal flat landform deformation prediction system according to claim 6, characterized in that: The grey relational degree calculation unit specifically includes: The grey relational degree calculation subunit at any moment is used to obtain, for any moment, the grey relational degree between the i-th influencing factor corresponding to the moment and the terrain change amount in the processed terrain dataset according to the terrain change amount at the moment in the processed terrain dataset and the value of the influencing factor at the moment in the dataset after processing the i-th influencing factor; The total grey correlation calculation subunit is used to obtain the grey correlation between the dataset of the i-th influencing factor and the terrain dataset according to the grey correlation between the i-th influencing factor and the terrain change in the processed terrain dataset at all times.

8. The tidal flat landform deformation prediction system according to claim 5, characterized in that: The prediction module specifically includes: An acquisition unit, configured to acquire a frequency set corresponding to all relevant influencing factors of the area to be predicted; A prediction unit is configured to input any frequency in a set of frequencies corresponding to all relevant influencing factors of the area to be predicted into a target prediction network to obtain terrain deformation components corresponding to all relevant influencing factors corresponding to the frequency; the target prediction network is a prediction network corresponding to the frequency; The superposition unit is configured to superpose the terrain deformation components corresponding to all relevant influencing factors corresponding to all frequencies in the frequency set corresponding to all relevant influencing factors of the to-be-predicted area to obtain the terrain deformation of the to-be-predicted area.

9. An electronic device, characterized in that: include: A memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform a tidal flat landform deformation prediction method according to any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that The device stores a computer program, which, when executed by a processor, implements a tidal flat landform deformation prediction method according to any one of claims 1 to 4.

Images