Landslide displacement prediction method and device based on convolutional neural network
Patent Information
- Application Number
- CN202510613845.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-13
- Publication Date
- 2025-09-26
AI Technical Summary
Traditional landslide monitoring methods have the disadvantages of high equipment cost, insufficient sensitivity, and limited coverage, making it difficult to achieve large-scale deployment and continuous monitoring. In addition, existing prediction methods have the problems of slow model convergence, easy to fall into local optimality, and insufficient ability to fuse multi-source spatiotemporal data when dealing with the complex nonlinear characteristics of landslide dynamic evolution.
A convolutional neural network combined with the Mantis optimization algorithm was used to construct a sample data set by obtaining engineering geological and climatic conditions data of the landslide. The weights and thresholds of the neural network model were optimized using the training set, and a landslide displacement prediction model was established. Logistic chaos mapping and Gaussian mutation were introduced to improve the global search capability of the algorithm.
It achieves accurate landslide displacement prediction, saves time and labor costs, improves system safety, enhances adaptability to complex geological conditions, reduces safety hazards, and increases operating speed and optimizes performance.
Smart Images
Figure CN120706609A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of landslide safety technology, and in particular to a landslide displacement prediction method and device based on a convolutional neural network. Background Art
[0002] Traditional landslide monitoring relies primarily on manually buried sensors to obtain real-time data. However, these sensors are subject to high equipment costs, insufficient sensitivity, and limited coverage, making large-scale deployment and continuous monitoring difficult. With the development of intelligent algorithms, prediction methods based on monitoring data have become a research hotspot. However, current prediction methods still have limitations when dealing with the complex nonlinear characteristics of landslide dynamics, such as slow model convergence, susceptibility to local optimality, and insufficient ability to fuse multi-source spatiotemporal data. Summary of the Invention
[0003] The purpose of the present invention is to provide a landslide displacement prediction method and device based on convolutional neural network to improve the above problems. To achieve the above object, the technical solution adopted by the present invention is as follows:
[0004] In a first aspect, the present application provides a landslide displacement prediction method based on a convolutional neural network, comprising:
[0005] Obtaining engineering geological conditions, climatic conditions, and displacement of landslides in different regions to construct a sample data set, and dividing the sample data set into a training set and a test set;
[0006] Build a neural network model, initialize the parameters of the neural network model, and randomly assign the initial weights and initial thresholds of the parameters;
[0007] The neural network model is trained using the training set, and weights and thresholds are updated during the training process by back propagation of the Mantis optimization algorithm to obtain optimal weights and optimal thresholds;
[0008] Assign the optimal weights and optimal thresholds to the parameters of the neural network model to obtain the optimized neural network model;
[0009] The optimized neural network model is tested using the test set to obtain a trained landslide displacement prediction model.
[0010] In a second aspect, the present application also provides a landslide displacement prediction device based on a convolutional neural network, comprising:
[0011] Acquisition module: Acquiring engineering geological conditions, climatic conditions, and displacement of landslides in different regions to construct a sample data set, and dividing the sample data set into a training set and a test set;
[0012] Model building module: builds a neural network model, initializes the parameters of the neural network model, and randomly assigns the initial weights and initial thresholds of the parameters;
[0013] Training module: using the training set to train the neural network model, and updating the weights and thresholds through back propagation of the Mantis optimization algorithm during the training process to obtain the optimal weights and thresholds;
[0014] Assignment module: assign the optimal weights and optimal thresholds to the parameters of the neural network model to obtain the optimized neural network model;
[0015] Testing module: using the test set to test the optimized neural network model to obtain a trained landslide displacement prediction model.
[0016] In a third aspect, the present application also provides a landslide displacement prediction device based on a convolutional neural network, comprising:
[0017] memory for storing computer programs;
[0018] A processor is configured to implement the steps of the landslide displacement prediction method based on a convolutional neural network when executing the computer program.
[0019] In a fourth aspect, the present application also provides a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned landslide displacement prediction method based on convolutional neural network.
[0020] The beneficial effects of the present invention are:
[0021] 1. The present invention uses the engineering geological conditions, climatic conditions, and displacement of landslides in recent years as input data for a convolutional neural network (CNN) model. The optimal weights and thresholds of the CNN model are found through the Dung Beetle Optimization (DBO) algorithm to obtain a comprehensive landslide displacement prediction model. This model can accurately predict landslide deformation data, saving time and labor costs, making more effective use of the economy, thereby minimizing safety hazards caused by factors such as complex regional geological conditions and improving system safety.
[0022] 2. The present invention optimizes the parameters in the convolutional neural network through the dung beetle optimization algorithm (DBO), which can eliminate the uncertainty of manual operation. At the same time, the standard DBO is improved by introducing logistic chaos mapping, Gaussian mutation, and Gaussian perturbation, thereby improving the global search and local optimality of the DBO algorithm, enhancing its optimization performance, and improving the running speed, overcoming the defects of long iteration time, low convergence accuracy, and easy search stagnation.
[0023] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the embodiments of the present invention. The purposes and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.
[0025] Figure 1 Schematic diagram of the flow of a landslide displacement prediction method based on a convolutional neural network according to an embodiment of the present invention;
[0026] Figure 2 Parameter mapping diagram of Logistic chaos described in an embodiment of the present invention;
[0027] Figure 3 A moving path diagram of the ball-rolling dung beetle described in an embodiment of the present invention;
[0028] Figure 4 A conceptual model diagram of the boundary selection strategy described in an embodiment of the present invention;
[0029] Figure 5 Schematic diagram of the structure of a landslide displacement prediction device based on a convolutional neural network according to an embodiment of the present invention;
[0030] Figure 6 Schematic diagram of the structure of the landslide displacement prediction device based on convolutional neural network described in an embodiment of the present invention.
[0031] Markings in the figure:
[0032] 800. Landslide displacement prediction device based on convolutional neural network; 801. Processor; 802. Memory; 803. Multimedia component; 804. I / O interface; 805. Communication component. DETAILED DESCRIPTION
[0033] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0034] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are used only to distinguish the description and should not be understood as indicating or implying relative importance.
[0035] Example 1:
[0036] This embodiment provides a landslide displacement prediction method based on convolutional neural network.
[0037] See also Figure 1 , the figure shows that the method includes:
[0038] S1. Obtaining engineering geological conditions, climatic conditions, and displacement of landslides in different regions to construct a sample data set, and dividing the sample data set into a training set and a test set;
[0039] Specifically, step S1 includes:
[0040] S11. Collecting vector data of engineering geological conditions of the landslide, the engineering geological conditions including stratum lithology, weathering grade and groundwater conditions;
[0041] Specifically, the stratum lithology includes sandstone and shale. In this embodiment, let sandstone = 1 and shale = 2;
[0042] Specifically, the weathering level (e.g., unweathered, weakly weathered, or strongly weathered) is converted into categorical labels or continuous values (e.g., 0, 1, 2);
[0043] Specifically, the groundwater level depth is converted into a floating point value;
[0044] S12. Convert the vector data into raster data of uniform spatial resolution (e.g., 100×100 grid) to obtain a plurality of raster cells;
[0045] S13. Assigning corresponding stratum lithology, weathering grade, and groundwater conditions to each grid cell;
[0046] S14. Collect the rainfall from the second to the eighth day before the landslide, and obtain the rainfall in the rainfall time series data;
[0047] S15. Collect the historical displacement increments of the landslide within the past 20 years and the displacement increments of the landslide within the past 24 hours;
[0048] S16. Using the engineering geological conditions, climate conditions, and historical displacement increments of the landslide as input data and the displacement increments within the predicted period as output data to construct a sample data set, and dividing the sample data set into a training set and a test set;
[0049] Specifically, step S16 includes:
[0050] S161. Fusing the raster data of landslide stratum lithology, weathering grade, groundwater status, rainfall time series data, and historical displacement increments to form multi-channel input data, namely, stratum lithology (1 channel), weathering grade (1 channel), groundwater status (1 channel), rainfall time series data (7 channels), and historical displacement increments (1 channel), for a total of 11 channels of input data;
[0051] S162. Take the displacement increment of the landslide in the past 24 hours as output data;
[0052] S163. A sample data set is formed by multi-channel input data and output data of several landslides;
[0053] S164. Divide the sample dataset into a training set and a test set in a ratio of 7:3.
[0054] Based on the above embodiment, the method further includes:
[0055] S2. Build a neural network model and determine the number of nodes and topology of the input layer, hidden layer, and output layer of the convolutional neural network (CNN); initialize the parameters of the neural network model and randomly assign initial weights and thresholds to the parameters;
[0056] Based on the above embodiment, the method further includes:
[0057] S3. The neural network model is trained using the training set, and the weights and thresholds are updated by back propagation of the Mantis Optimization Algorithm (DBO) during the training process to obtain the optimal weights and thresholds;
[0058] Specifically, step S3 includes:
[0059] S31. Generate chaotic sequences by chaotic mapping and map them to the search space to form the initial population and determine the maximum number of iterations T max ;
[0060] In this embodiment, the Logistic chaotic mapping can evenly distribute dung beetles in the search space, which not only increases the diversity of the initial population of dung beetles, but also avoids falling into local optimality during the search process of the dung beetle algorithm. Specifically, the mathematical expression of the Logistic mapping is as follows:
[0061]
[0062] Where, is the current iteration state value of the i-th variable, and its value range is [0,1]; is the next iteration state value; μ is the branch parameter that determines whether the Logistic map is in a chaotic state, and its value range is [0,4]. The parameter mapping diagram after 1000 iterations of the Logistic chaotic map is shown in Figure 2 As shown;
[0063] S32. Calculate the initial fitness of each individual in the population in turn, and obtain the initial fitness F(X) of the population based on the initial fitness of each individual:
[0064]
[0065] Where d is the dimension of the solution space, x N,d is the position of the Nth dung beetle in the population in the dth dimension, f([x N,1 ,x N，2 …x N,d ]) is the fitness value of the Nth dung beetle;
[0066] S33. Update the positions of individuals in the population according to the initial fitness of the population, and perform Gaussian mutation or Gaussian perturbation on the individuals;
[0067] Specifically, step S33 includes:
[0068] S331. Determine the proportion of dung beetles in the population based on the characteristics of the sample data set. r , the proportion of egg-laying dung beetles P e , the proportion of foraging dung beetles P f and the proportion of stealing dung beetles P s ;
[0069] S332. Update the positions of the rolling dung beetle, the egg-laying dung beetle, the foraging dung beetle, and the stealing mantis in space in sequence;
[0070] Specifically, the position update process of the rolling ball dung beetle is:
[0071] The dung beetle uses celestial clues to navigate and ensure that the ball rolling path is straight. To simulate the ball rolling process, the dung beetle is set to move in a given direction within the search area. The movement path of the dung beetle is as follows: Figure 3 As shown, the dung beetle uses the sun for navigation. The arrow indicates the direction of movement. At the same time, the light intensity also affects the movement path of the dung beetle. The position update of the rolling dung beetle can be expressed as:
[0072]
[0073] Where x i t+1 is the position of the i-th dung beetle at the t+1th iteration, is the position of the i-th dung beetle at the t-th iteration, t is the number of iterations, k is the perturbation coefficient which is a fixed value (k∈(0,0.2]), b∈(0,1) is a fixed value, a is the natural coefficient, which is assigned to 1 or -1, Δx is the light intensity change, X w is the global worst position.
[0074] When the ball-rolling dung beetle encounters an obstacle, it needs to adjust itself by dancing. It changes direction by dancing to avoid the obstacle and continue to move forward to obtain a new route. The new ball direction and new position update after the dance is completed are obtained through the tangent function as follows:
[0075] x i t+1 =x i t +tanψ|x i t -x phest |;
[0076] Where: ψ is the disturbance angle and ψ∈[0,π], x pbest It is the best position in the history of dung beetles.
[0077] Specifically, since the female dung beetles in the population will choose the egg-laying location and perform egg-laying behavior, the position update process of the egg-laying mantis can be expressed as:
[0078]
[0079] Where: Lb′ and Ub′ are the upper and lower limits of the spawning area, X′ is the local optimal position, T max is the maximum number of iterations, Lb and Ub are the upper and lower levels of the optimization problem, and B i t is the position information of the egg ball at the tth iteration, B i t+1The position information of the egg ball at the t+1th iteration, b1 and b2 are two random independent vectors of size 1×D, D is the dimension of the optimization problem, and R represents the scale factor; the conceptual model of the boundary selection strategy is as follows Figure 4 shown.
[0080] Specifically, the foraging dung beetle simulates the foraging behavior of the dung beetle and updates its position to find food. The position update process of the foraging mantis is:
[0081]
[0082] Where x i t+1 The position of the foraging dung beetle at the t+1th iteration, x i t is the location of the foraging dung beetle at the tth iteration, Lb and Ub are the upper and lower bounds of the feeding area, X b is the optimal feeding position of the dung beetle, C1 is a random number, C2 is a random vector belonging to (0,1), R represents the proportional coefficient, Lb b With Ub b Represents the lower and upper bounds of the expanded search space.
[0083] Specifically, the stealing dung beetle simulates the stealing behavior of dung beetles and updates its position to steal food from other dung beetles. The position update process of the stealing dung beetle can be expressed as:
[0084] x i t+1 =X b +S·g·(|x i t -X′|+|x i t -X b |);
[0085] Where: x i t+1 The position of the stolen dung beetle at the t+1th iteration, x i t is the location information of the i-th stealing dung beetle at the t-th iteration, X b is the location of the optimal food source, g is a random vector of size 1×D, S is a constant value, and X′ is the historical optimal location;
[0086] S333. Calculate the average fitness of the population based on the updated population position:
[0087]
[0088] Where, f represents the fitness of the i-th mantis, f av represents the average fitness, and N represents the population size.
[0089] S334. Calculate the fitness of each individual in turn and determine whether the fitness of each individual is greater than the average fitness f av :
[0090] If so, perform Gaussian perturbation on the individual;
[0091] Otherwise, Gaussian mutation is performed on the individuals.
[0092] In this example, Gaussian mutation and Gaussian perturbation are introduced to improve the algorithm's optimization capabilities and escape local optima. Gaussian mutation involves extracting a random number from a normal distribution with mean μ and σ², and permuting the parameters in the DBO algorithm to achieve the optimization goal. Gaussian perturbation improves the ability to escape local optima by applying a small Gaussian perturbation to the optimal position of individual dung beetles after each iteration, helping them escape local optima.
[0093] When the individual fitness is greater than the average fitness f av When , Gaussian mutation is performed on the individual, the specific formula is as follows:
[0094] T g =Lo(1+N(0,1));
[0095] Where: N(0,1) is a normally distributed random number with an expected value of 0 and a standard deviation of 1; T g is the value calculated by Gaussian mutation after initializing the parameters;
[0096] When the individual fitness is less than the average fitness f av When , Gaussian perturbation is performed on the individual, the specific formula is as follows:
[0097]
[0098] Where: G b is the optimal position of the dung beetle in each iteration; G is the fitness of the dung beetle before Gaussian perturbation, Gaussian(μ,σ 2 ) has a mean of μ and a variance of σ 2 Gaussian function of is the optimal fitness of the disturbed dung beetle individual; G t is the optimal fitness of the dung beetle individual in the tth iteration; G t+1 is the optimal fitness of the dung beetle individual in the t+1th iteration.
[0099] S34. Repeatedly calculate the fitness of the population after the updated position and determine whether the number of position updates reaches the maximum number of iterations;
[0100] If not, the individual's position is repeatedly updated according to the calculated fitness;
[0101] If so, the position of the current population individual is taken as the optimal position to generate the optimal weights and optimal thresholds of the convolutional neural network model.
[0102] Based on the above embodiment, the method further includes:
[0103] S4. Assigning the optimal weights and optimal thresholds to the parameters of the neural network model to obtain an optimized neural network model (DBO-CNN model);
[0104] Based on the above embodiment, the method further includes:
[0105] S5. Testing the optimized neural network model using the test set to obtain a trained landslide displacement prediction model, including:
[0106] S51. Input the test set into the optimized neural network model;
[0107] S52. After extracting the data features of the multi-channel input data by the neural network model, the predicted displacement increment is output;
[0108] S53. Calculate the error value based on the predicted displacement increment and the actual displacement increment in the test set;
[0109] S54. When the error value is less than a preset threshold, the neural network model training is completed and a landslide displacement prediction model is obtained.
[0110] Based on the above embodiment, the method further includes:
[0111] S6. Collect the lithology, weathering grade, groundwater conditions, rainfall in the past seven days, and historical displacement increments within the past 20 years of the predicted landslide as input data;
[0112] S7. Input the output data into a landslide displacement prediction model for prediction, and obtain the displacement increment of the landslide to be predicted within the next 24 hours.
[0113] Example 2:
[0114] like Figure 5 As shown, this embodiment provides a landslide displacement prediction device based on a convolutional neural network, the device comprising:
[0115] Acquisition module: Acquiring engineering geological conditions, climatic conditions, and displacement of landslides in different regions to construct a sample data set, and dividing the sample data set into a training set and a test set;
[0116] Model building module: builds a neural network model, initializes the parameters of the neural network model, and randomly assigns the initial weights and initial thresholds of the parameters;
[0117] Training module: using the training set to train the neural network model, and updating the weights and thresholds through back propagation of the Mantis optimization algorithm during the training process to obtain the optimal weights and thresholds;
[0118] Assignment module: assign the optimal weights and optimal thresholds to the parameters of the neural network model to obtain the optimized neural network model;
[0119] Testing module: using the test set to test the optimized neural network model to obtain a trained landslide displacement prediction model.
[0120] Based on the above embodiment, the acquisition module includes:
[0121] The first acquisition unit is used to collect vector data of engineering geological conditions of the landslide, including stratum lithology, weathering grade, and groundwater conditions;
[0122] Conversion unit: convert vector data into raster data of uniform spatial resolution to obtain several raster units;
[0123] Assignment unit: Assign corresponding stratum lithology, weathering grade and groundwater conditions to each grid cell;
[0124] The second collection unit: collects the rainfall from the second to the eighth day in the past of the landslide to obtain rainfall time series data;
[0125] The third collection unit: collects the historical displacement increments of the landslide within the past 20 years and the displacement increments of the landslide within the past 24 hours.
[0126] Based on the above embodiment, the acquisition module further includes:
[0127] The first selection unit: integrates the raster data of landslide stratigraphic lithology, weathering grade, groundwater status, rainfall time series data, and historical displacement increments to form multi-channel input data;
[0128] The second selection unit: takes the displacement increment of the landslide in the past 24 hours as output data;
[0129] Sample construction unit: a sample data set is constructed by multi-channel input and output data of several landslides;
[0130] Division unit: Divide the sample data set into training set and test set in a ratio of 7:3.
[0131] Based on the above embodiment, the training module includes:
[0132] Mapping unit: generates chaotic sequences through chaotic mapping and maps them to the search space to form the initial population and determine the maximum number of iterations;
[0133] The first calculation unit calculates the initial fitness of each individual in the population in sequence, and obtains the initial fitness of the population based on the initial fitness of each individual;
[0134] The first updating unit: updates the positions of individuals in the population according to the initial fitness of the population, and performs Gaussian mutation or Gaussian perturbation on the individuals;
[0135] The first judgment unit: repeatedly calculates the fitness of the population after the updated position, and determines whether the number of position updates reaches the maximum number of iterations;
[0136] If not, the individual's position is repeatedly updated according to the calculated fitness;
[0137] If so, the position of the current population individual is taken as the optimal position to generate the optimal weights and optimal thresholds of the convolutional neural network model.
[0138] Based on the above embodiment, the updating unit includes:
[0139] Determine the unit: Determine the proportion of ball-rolling dung beetles, egg-laying dung beetles, foraging dung beetles, and stealing dung beetles in the population based on the characteristics of the sample data set;
[0140] The second updating unit: updates the positions of the rolling dung beetle, egg-laying dung beetle, foraging dung beetle and stealing mantis in space in sequence;
[0141] The second calculation unit calculates the average fitness of the population based on the updated population position;
[0142] Second judgment unit: calculate the fitness of individuals in turn and judge whether the fitness of individuals is greater than the average fitness:
[0143] If so, perform Gaussian perturbation on the individual;
[0144] Otherwise, Gaussian mutation is performed on the individuals.
[0145] Based on the above embodiment, the test module includes:
[0146] Input unit: input the test set into the optimized neural network model;
[0147] Output unit: After extracting the data features of multi-channel input data through the neural network model, it outputs the predicted displacement increment;
[0148] The third calculation unit calculates the error value based on the predicted displacement increment and the actual displacement increment in the test set;
[0149] The third judgment unit: when the error value is less than a preset threshold, the training of the neural network model is completed, and a landslide displacement prediction model is obtained.
[0150] It should be noted that, regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated on here.
[0151] Example 3:
[0152] Corresponding to the above method embodiment, this embodiment also provides a landslide displacement prediction device based on a convolutional neural network. The landslide displacement prediction device based on a convolutional neural network described below and the landslide displacement prediction method based on a convolutional neural network described above can refer to each other.
[0153] Figure 6 FIG. 8 is a block diagram of a landslide displacement prediction device 800 based on a convolutional neural network according to an exemplary embodiment. Figure 6 As shown, the landslide displacement prediction device 800 based on convolutional neural network may include: a processor 801, a memory 802. The landslide displacement prediction device 800 based on convolutional neural network may also include one or more of a multimedia component 803, an I / O interface 804, and a communication component 805.
[0154] The processor 801 is used to control the overall operation of the convolutional neural network-based landslide displacement prediction device 800 to complete all or part of the steps in the convolutional neural network-based landslide displacement prediction method described above. The memory 802 is used to store various types of data to support the operation of the convolutional neural network-based landslide displacement prediction device 800. This data may include, for example, instructions for any application or method operating on the convolutional neural network-based landslide displacement prediction device 800, as well as application-related data such as contact information, sent and received messages, images, audio, video, and the like. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 802 or transmitted via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, which may include a keyboard, mouse, buttons, etc. These buttons may be virtual or physical. The communication component 805 is used for wired or wireless communication between the convolutional neural network-based landslide displacement prediction device 800 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof, may include a Wi-Fi module, a Bluetooth module, or an NFC module.
[0155] In an exemplary embodiment, the landslide displacement prediction device 800 based on a convolutional neural network can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-mentioned landslide displacement prediction method based on a convolutional neural network.
[0156] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided. When executed by a processor, the program instructions implement the steps of the above-described convolutional neural network-based landslide displacement prediction method. For example, the computer-readable storage medium may be the aforementioned memory 802 including the program instructions. The program instructions may be executed by the processor 801 of the convolutional neural network-based landslide displacement prediction device 800 to implement the above-described convolutional neural network-based landslide displacement prediction method.
[0157] Example 4:
[0158] Corresponding to the above method embodiment, this embodiment further provides a readable storage medium. The readable storage medium described below and the landslide displacement prediction method based on convolutional neural network described above can refer to each other.
[0159] A readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the landslide displacement prediction method based on a convolutional neural network of the above-mentioned method embodiment.
[0160] The readable storage medium may specifically be any readable storage medium that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.
[0161] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.
[0162] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.
Claims
1. A landslide displacement prediction method based on convolutional neural network, characterized in that: include: Obtaining engineering geological conditions, climatic conditions, and displacement of landslides in different regions to construct a sample data set, and dividing the sample data set into a training set and a test set; Build a neural network model, initialize the parameters of the neural network model, and randomly assign the initial weights and initial thresholds of the parameters; The neural network model is trained using the training set, and weights and thresholds are updated during the training process by back propagation of the Mantis optimization algorithm to obtain optimal weights and optimal thresholds; Assign the optimal weights and optimal thresholds to the parameters of the neural network model to obtain the optimized neural network model; The optimized neural network model is tested using the test set to obtain a trained landslide displacement prediction model.
2. The landslide displacement prediction method based on convolutional neural network according to claim 1 is characterized in that: Obtain the engineering geological conditions, climatic conditions and displacement of landslides in different regions, including: Collecting vector data of engineering geological conditions of the landslide, including stratum lithology, weathering grade, and groundwater conditions; Convert the vector data into raster data of uniform spatial resolution respectively to obtain a number of raster units; Assign corresponding stratum lithology, weathering grade and groundwater conditions to each grid cell; Collect the rainfall from the 2nd to the 8th day in the past of the landslide to obtain rainfall time series data; Collect the historical displacement increments of the landslide within the past 20 years and the displacement increments of the landslide within the past 24 hours.
3. The landslide displacement prediction method based on convolutional neural network according to claim 2 is characterized in that: Constructing a sample data set and dividing the sample data set into a training set and a test set, including: The raster data of landslide stratigraphic lithology, weathering grade, groundwater status, rainfall time series data, and historical displacement increments are integrated to form multi-channel input data. The displacement increment of the landslide in the past 24 hours is used as output data; The sample dataset consists of multi-channel input and output data of several landslides; The sample data set is divided into training set and test set in a ratio of 7:
3.
4. The landslide displacement prediction method based on convolutional neural network according to claim 1 is characterized in that: During the training process, the weights and biases are updated through backpropagation of the Mantis optimization algorithm to obtain the optimal weights and thresholds, including: Generate chaotic sequences through chaotic mapping and map them to the search space to form the initial population and determine the maximum number of iterations; Calculate the initial fitness of each individual in the population in turn, and obtain the initial fitness of the population based on the initial fitness of each individual; Update the positions of individuals in the population according to the initial fitness of the population, and perform Gaussian mutation or Gaussian perturbation on the individuals; Repeatedly calculate the fitness of the population after the updated position, and determine whether the number of position updates reaches the maximum number of iterations; If not, the individual's position is repeatedly updated according to the calculated fitness; If so, the position of the current population individual is taken as the optimal position to generate the optimal weights and optimal thresholds of the convolutional neural network model.
5. The landslide displacement prediction method based on convolutional neural network according to claim 1 is characterized in that: The optimized neural network model is tested using the test set to obtain a trained landslide displacement prediction model, including: Input the test set into the optimized neural network model; After extracting the data features of multi-channel input data through the neural network model, the predicted displacement increment is output; The error value is calculated based on the predicted displacement increment and the actual displacement increment in the test set; When the error value is less than a preset threshold, the training of the neural network model is completed and a landslide displacement prediction model is obtained.
6. A landslide displacement prediction device based on convolutional neural network, characterized in that: include: Acquisition module: Acquiring engineering geological conditions, climatic conditions, and displacement of landslides in different regions to construct a sample data set, and dividing the sample data set into a training set and a test set; Model building module: builds a neural network model, initializes the parameters of the neural network model, and randomly assigns the initial weights and initial thresholds of the parameters; Training module: using the training set to train the neural network model, and updating the weights and thresholds through back propagation of the Mantis optimization algorithm during the training process to obtain the optimal weights and thresholds; Assignment module: assign the optimal weights and optimal thresholds to the parameters of the neural network model to obtain the optimized neural network model; Testing module: using the test set to test the optimized neural network model to obtain a trained landslide displacement prediction model.
7. The landslide displacement prediction device based on convolutional neural network according to claim 6, characterized in that: The acquisition module includes: The first acquisition unit is used to collect vector data of engineering geological conditions of the landslide, including stratum lithology, weathering grade, and groundwater conditions; Conversion unit: convert vector data into raster data of uniform spatial resolution to obtain several raster units; Assignment unit: Assign corresponding stratum lithology, weathering grade and groundwater conditions to each grid cell; The second collection unit: collects the rainfall from the second to the eighth day in the past of the landslide to obtain rainfall time series data; The third collection unit: collects the historical displacement increments of the landslide within the past 20 years and the displacement increments of the landslide within the past 24 hours.
8. The landslide displacement prediction device based on convolutional neural network according to claim 7, characterized in that: The acquisition module also includes: The first selection unit: integrates the raster data of landslide stratigraphic lithology, weathering grade, groundwater status, rainfall time series data, and historical displacement increments to form multi-channel input data; The second selection unit: takes the displacement increment of the landslide in the past 24 hours as output data; Sample construction unit: a sample data set is constructed by multi-channel input and output data of several landslides; Division unit: Divide the sample data set into training set and test set in a ratio of 7:
3.
9. The landslide displacement prediction device based on convolutional neural network according to claim 5, characterized in that: The training module includes: Mapping unit: generates chaotic sequences through chaotic mapping and maps them to the search space to form the initial population and determine the maximum number of iterations; The first calculation unit calculates the initial fitness of each individual in the population in sequence, and obtains the initial fitness of the population based on the initial fitness of each individual; The first updating unit: updates the positions of individuals in the population according to the initial fitness of the population, and performs Gaussian mutation or Gaussian perturbation on the individuals; The first judgment unit: repeatedly calculates the fitness of the population after the updated position, and determines whether the number of position updates reaches the maximum number of iterations; If not, the individual's position is repeatedly updated according to the calculated fitness; If so, the position of the current population individual is taken as the optimal position to generate the optimal weights and optimal thresholds of the convolutional neural network model.
10. The landslide displacement prediction device based on convolutional neural network according to claim 5, characterized in that: The test module includes: Input unit: input the test set into the optimized neural network model; Output unit: After extracting the data features of multi-channel input data through the neural network model, it outputs the predicted displacement increment; The third calculation unit calculates the error value based on the predicted displacement increment and the actual displacement increment in the test set; The third judgment unit: when the error value is less than a preset threshold, the training of the neural network model is completed, and a landslide displacement prediction model is obtained.