Regional geological disaster risk assessment system and method based on big data

Abstract

The invention discloses a regional geological disaster risk assessment system and method based on big data, and the method comprises the following steps: S1, obtaining geological data and related data, and carrying out the preprocessing of the geological data and related data, and generating a geological disaster data set; s2, training Bi-LSTM through the preprocessed geological disaster data set, and constructing a geological disaster risk assessment model; and S3, collecting a new geological disaster data set, inputting the new geological disaster data set into the geological disaster risk assessment model, and generating a geological disaster risk prediction result. According to the geological disaster risk assessment method, the convolutional neural network, the Bi-LSTM and the attention mechanism layer are combined to construct the geological disaster risk assessment model, geological disaster assessment is realized, and compared with a current geological disaster risk assessment method, the PSO algorithm is utilized to perform optimal matching on hyper-parameters of the Bi-LSTM model, so that diversity characteristics of geological data and related data are captured more comprehensively, and the geological disaster risk assessment method is more efficient. Therefore, the geological disaster risk prediction precision and reliability are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of geological disaster risk assessment, and specifically relates to a regional geological disaster risk assessment system and method based on big data. Background Art

[0002] Geological disasters such as landslides, mudslides, and earthquakes pose a great threat to densely populated and built-up areas. It is urgent to accurately and effectively assess the risks of geological disasters. Through risk assessment, high-risk areas can be identified, preventive measures can be taken in advance, and the damage that may be caused when disasters occur can be reduced. Geological hazard risk assessment is a multidimensional process that involves comprehensive analysis and assessment of the potential hazards of geological hazards, human exposure, and vulnerability. The current methods of geological hazard risk assessment are mainly divided into qualitative and quantitative evaluation. Qualitative evaluation is based on expert experience and judgment, while quantitative evaluation is based on mathematical models and statistical analysis. Its evaluation methods are relatively limited and not comprehensive enough, resulting in low reliability of the evaluation results, which makes it difficult to cope with complex and changing geological environments and diverse evaluation needs. Summary of the invention

[0003] In view of the above-mentioned deficiencies in the prior art, the present invention provides a regional geological hazard risk assessment system and method based on big data, which solves the problem that the reliability of the assessment results calculated by the current geological hazard risk assessment method is low.

[0004] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a regional geological disaster risk assessment method based on big data, comprising the following steps:

[0005] S1. Acquire geological data and related data, and pre-process them to generate a geological disaster data set;

[0006] S2, train Bi-LSTM through the preprocessed geological disaster data set to build a geological disaster risk assessment model;

[0007] S3. Collect new geological disaster data sets, input them into the geological disaster risk assessment model, and generate geological disaster risk prediction results.

[0008] Further: In S1, geological data include surface cover type, stratigraphic structure, stratigraphic lithology, topography, vegetation coverage, elevation and slope, and related data include rainfall, surface temperature and soil moisture content.

[0009] Further: S2 comprises the following steps:

[0010] S21, inputting the geological disaster data set into the convolutional neural network to extract the features of the geological disaster data;

[0011] S22, input the characteristics of geological disaster data into Bi-LSTM, train the Bi-LSTM through the PSO algorithm, and build a geological disaster risk assessment model based on the trained Bi-LSTM;

[0012] In S21, the convolutional neural network includes an input layer, a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a fully connected layer and an output layer which are connected in sequence.

[0013] Further: In S22, the method for training Bi-LSTM by PSO algorithm is specifically as follows:

[0014] S221. Set the size of the population according to the model parameters of Bi-LSTM, and initialize the population and the number of iterations;

[0015] S222, calculating the fitness of particles and determining the optimal particle individual;

[0016] S223, updating the position and speed of particles in the population according to the position of the optimal individual particle;

[0017] S224, determine whether the number of iterations reaches the iteration threshold. If so, obtain the optimal learning rate and number of hidden units of Bi-LSTM according to the position of the current particle population, and build a geological disaster risk assessment model. If not, add 1 to the number of iterations and return to S222.

[0018] The beneficial effect of the above further scheme is: the present invention uses the PSO algorithm to perform global optimal search for the model parameters of Bi-LSTM, effectively improving the accuracy and operation speed of Bi-LSTM. The PSO algorithm can find the optimal learning rate and number of hidden units, showing the ability of global optimization. It can not only improve the prediction accuracy of Bi-LSTM, but also avoid blind trial calculation of parameters, so that the prediction effect of geological disaster risks can be optimized.

[0019] Further: In S223, the expression for calculating the position of the i-th particle at the t+1-th iteration is specifically:

[0020]

[0021] In the formula, are the parameters of the perturbation mechanism, is the initial updated position of the ith particle, T is the iteration threshold, and c1 is the first learning factor;

[0022]

[0023] In the formula, is the position of the i-th particle at the t-th iteration, is the velocity of the i-th particle at the t+1-th iteration;

[0024]

[0025] In the formula, is the velocity of the ith particle at the tth iteration, is the inertia weight of the ith particle at the tth iteration, c2 is the second learning factor, is a random number, is the optimal position of the ith particle, is the historical optimal position of the population.

[0026] The beneficial effect of the above further scheme is: as the iteration proceeds, the particles in the population may stagnate at the local extreme value. The particles falling into the local optimum will affect the final optimization efficiency. In order to prevent the algorithm from falling into the local optimum, the present invention introduces a perturbation mechanism to enable individuals to jump out of the local optimum and expand the search space.

[0027] Further: Parameters of the perturbation mechanism The specific expression is:

[0028]

[0029] In the formula, is the fitness function, and N is the total number of particles.

[0030] Further: The inertia weight of the i-th particle at the t+1th iteration The specific expression is:

[0031]

[0032] In the formula, is the fitness threshold.

[0033] The beneficial effect of the above further solution is: the present invention proposes to dynamically adjust the inertia weight according to the iteration result , which aims to optimize the performance of the particle swarm algorithm. During the fitness function optimization process, if the fitness function value changes less than the fitness threshold, it is considered that the iteration is trapped in the local optimum, and the weight is updated according to the first expression of the above formula. If the fitness function value changes not less than the fitness threshold, it is considered that it is not trapped in the local optimum, and the weight is updated according to the second expression of the above formula. By adjusting the weight in time during the iteration process and continuously correcting the search range, the algorithm's optimization ability can be flexibly adjusted, thereby enhancing the efficiency of searching for the global optimal solution.

[0034] Further: S3 is specifically:

[0035] Collect new geological hazard data sets, which are geological data and related data within a set time period, input the new geological hazard data sets into the geological hazard risk assessment model, and predict the geological hazard risks generated by the geological data and related data.

[0036] Further: In S3, the geological hazard risk assessment model includes a convolutional neural network, a Bi-LSTM and an attention mechanism layer connected in sequence, the weights are calculated by the attention mechanism layer, and the output results of the Bi-LSTM are weighted according to the weights.

[0037] A regional geological disaster risk assessment system based on big data, the system includes:

[0038] Data acquisition module, used to obtain geological data and related data, and pre-process them to generate geological disaster data sets;

[0039] Model training module, used to train Bi-LSTM with preprocessed geological disaster data set to build a geological disaster risk assessment model;

[0040] The prediction module is used to collect new geological disaster data sets, input them into the geological disaster risk assessment model, and generate geological disaster risk prediction results.

[0041] The beneficial effects of the present invention are:

[0042] (1) The present invention provides a regional geological hazard risk assessment system and method based on big data, which collects geological data and related data, combines convolutional neural network, Bi-LSTM and attention mechanism layer to construct a geological hazard risk assessment model to achieve geological hazard assessment. Compared with the current geological hazard risk assessment method, the PSO algorithm is used to optimally match the hyperparameters of the Bi-LSTM model so as to more comprehensively capture the diverse characteristics of geological data and related data, thereby improving the accuracy and reliability of geological hazard risk prediction.

[0043] (2) The present invention proposes an innovative method for constructing a geological hazard risk assessment model, namely, setting the inertia weight through the fitness function value in the iterative process to improve the PSO algorithm, aiming to improve the accuracy of parameter search and the convergence speed of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flow chart of a regional geological disaster risk assessment method based on big data of the present invention.

[0045] Figure 2 Schematic diagram of a convolutional neural network.

[0046] Figure 3 This is a schematic diagram of a regional geological disaster risk assessment system based on big data. DETAILED DESCRIPTION

[0047] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0048] like Figure 1 As shown, in one embodiment of the present invention, a regional geological disaster risk assessment method based on big data includes the following steps:

[0049] S1. Acquire geological data and related data, and pre-process them to generate a geological disaster data set;

[0050] S2, train Bi-LSTM through the preprocessed geological disaster data set to build a geological disaster risk assessment model;

[0051] S3. Collect new geological disaster data sets, input them into the geological disaster risk assessment model, and generate geological disaster risk prediction results.

[0052] In S1, geological data include surface cover type, stratigraphic structure, stratigraphic lithology, topography, vegetation coverage, elevation and slope, and related data include rainfall, surface temperature and soil moisture content.

[0053] In S1, the method for preprocessing geological data and related data is as follows:

[0054] In this embodiment, the collected geological data and related data are cleaned, and the data are unified through deletion, interpolation, scaling, and manual filling to establish a geological disaster data set to ensure the quality and integrity of the data.

[0055] For geological data, stratigraphic lithology, geological structure and topography are represented by text information. The present invention converts the text information into numerical information that can be used for calculation. Vegetation coverage and elevation are stored in image format. The present invention parses the collected images and converts them into numerical formats that can be analyzed, thereby completing the sampling of geological data.

[0056] As for the relevant data, the collected original meteorological data cannot be directly parsed. The present invention uses Python to parse the original meteorological data and extract rainfall, surface temperature and soil moisture content.

[0057] S2 includes the following steps:

[0058] S21, inputting the geological disaster data set into the convolutional neural network to extract the features of the geological disaster data;

[0059] S22. Input the characteristics of geological disaster data into Bi-LSTM, train Bi-LSTM through PSO algorithm, and build a geological disaster risk assessment model based on the trained Bi-LSTM.

[0060] like Figure 2 As shown, in S21, the convolutional neural network includes an input layer, a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a fully connected layer and an output layer which are connected in sequence.

[0061] In this embodiment, by performing convolution operation on the input data through a convolutional neural network, the correlation between multidimensional information in the input data can be deeply explored. Considering that the convolutional neural network lacks consideration of time dependence when processing time series data that requires prediction tasks, the present invention combines the convolutional neural network with Bi-LSTM to improve the prediction performance.

[0062] In S22, the method for training Bi-LSTM by PSO algorithm is as follows:

[0063] S221. Set the size of the population according to the model parameters of Bi-LSTM, and initialize the population and the number of iterations;

[0064] S222, calculating the fitness of particles and determining the optimal particle individual;

[0065] S223, updating the position and speed of particles in the population according to the position of the optimal individual particle;

[0066] S224, determine whether the number of iterations reaches the iteration threshold. If so, obtain the optimal learning rate and number of hidden units of Bi-LSTM according to the position of the current particle population, and build a geological disaster risk assessment model. If not, add 1 to the number of iterations and return to S222.

[0067] In this embodiment, the present invention uses the PSO algorithm to perform a global optimal search for the model parameters of the Bi-LSTM, which effectively improves the accuracy and operation speed of the Bi-LSTM. The PSO algorithm can find the optimal learning rate and number of hidden units, showing the ability of global optimization. It can not only improve the prediction accuracy of the Bi-LSTM, but also avoid blind trial calculation of parameters, so that the prediction effect of geological disaster risks can be optimized.

[0068] In S223, the expression for calculating the position of the i-th particle at the t+1-th iteration is specifically:

[0069]

[0070] In the formula, are the parameters of the perturbation mechanism, is the initial updated position of the ith particle, T is the iteration threshold, and c1 is the first learning factor;

[0071]

[0072] In the formula, is the position of the i-th particle at the t-th iteration, is the velocity of the i-th particle at the t+1-th iteration;

[0073]

[0074] In the formula, is the velocity of the ith particle at the tth iteration, is the inertia weight of the ith particle at the tth iteration, c2 is the second learning factor, is a random number, is the optimal position of the ith particle, is the historical optimal position of the population.

[0075] As the iteration proceeds, particles in the population may stagnate at local extremes. Particles falling into local optimality will affect the final optimization efficiency. In order to prevent the algorithm from falling into local optimality, the present invention introduces a perturbation mechanism to enable individuals to jump out of local optimality and expand the search space.

[0076] Parameters of the perturbation mechanism The specific expression is:

[0077]

[0078] In the formula, is the fitness function, and N is the total number of particles.

[0079] Inertia weight of the i-th particle at the t+1th iteration The specific expression is:

[0080]

[0081] In the formula, is the fitness threshold.

[0082] In the initial iteration, a larger weight is used to expand the global search range of the algorithm. As the iteration proceeds, the inertia weight is gradually reduced to improve the convergence accuracy. When the inertia weight is small, the algorithm is easy to converge, but it is also easy to fall into the local optimum. Therefore, the present invention proposes to dynamically adjust the inertia weight according to the iteration results. , which aims to optimize the performance of the particle swarm algorithm. During the fitness function optimization process, if the fitness function value changes less than the fitness threshold, it is considered that the iteration is trapped in the local optimum, and the weight is updated according to the first expression of the above formula. If the fitness function value changes not less than the fitness threshold, it is considered that it is not trapped in the local optimum, and the weight is updated according to the second expression of the above formula. By adjusting the weight in time during the iteration process and continuously correcting the search range, the algorithm's optimization ability can be flexibly adjusted, thereby enhancing the efficiency of searching for the global optimal solution.

[0083] S3 is specifically:

[0084] Collect new geological hazard data sets, which are geological data and related data within a set time period, input the new geological hazard data sets into the geological hazard risk assessment model, and predict the geological hazard risks generated by the geological data and related data.

[0085] In this embodiment, the present invention divides the geological hazard risks generated by geological data and related data into three levels: low, medium and high. In order to further improve the prediction efficiency, the present invention connects an attention mechanism layer after the Bi-LSTM to assign different weights to the output of the Bi-LSTM. This mechanism can focus on the parts related to the geological hazard risks and significantly improve the performance and generalization ability of the model.

[0086] In S3, the geological hazard risk assessment model includes a convolutional neural network, a Bi-LSTM and an attention mechanism layer connected in sequence. The weights are calculated through the attention mechanism layer, and the output results of the Bi-LSTM are weighted according to the weights.

[0087] like Figure 3 As shown, a regional geological disaster risk assessment system based on big data, a regional geological disaster risk assessment method applied to big data, the system includes:

[0088] Data acquisition module, used to obtain geological data and related data, and pre-process them to generate geological disaster data sets;

[0089] Model training module, used to train Bi-LSTM with preprocessed geological disaster data set to build a geological disaster risk assessment model;

[0090] The prediction module is used to collect new geological disaster data sets, input them into the geological disaster risk assessment model, and generate geological disaster risk prediction results.

[0091] The beneficial effects of the present invention are as follows: the present invention provides a regional geological hazard risk assessment system and method based on big data, collects geological data and related data, combines convolutional neural networks, Bi-LSTM and attention mechanism layers to construct a geological hazard risk assessment model, and realizes geological hazard evaluation. Compared with the current geological hazard risk assessment method, the PSO algorithm is used to optimally match the hyperparameters of the Bi-LSTM model so as to more comprehensively capture the diverse characteristics of geological data and related data, thereby improving the accuracy and reliability of geological hazard risk prediction.

[0092] The present invention proposes an innovative method for constructing a geological hazard risk assessment model, which is to set the inertia weight through the fitness function value in the iterative process to improve the PSO algorithm, aiming to improve the accuracy of parameter search and the convergence speed of the algorithm.

[0093] In the description of the present invention, it is necessary to understand that the orientation or positional relationship indicated by the terms "center", "thickness", "upper", "lower", "horizontal", "top", "bottom", "inner", "outer", "radial", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, the terms "first", "second", and "third" are used only for descriptive purposes, and cannot be understood as indicating or implying the relative importance or the number of implicitly specified technical features. Therefore, the features defined by "first", "second", and "third" may explicitly or implicitly include one or more of the features.

Claims

1. A regional geological disaster risk assessment method based on big data, characterized in that: The following steps are involved: S1. Acquire geological data and related data, and pre-process them to generate a geological disaster data set; S2, train Bi-LSTM through the preprocessed geological disaster data set to build a geological disaster risk assessment model; S3. Collect new geological disaster data sets, input them into the geological disaster risk assessment model, and generate geological disaster risk prediction results.

2. The regional geological disaster risk assessment method based on big data according to claim 1 is characterized in that: In S1, geological data include surface cover type, stratigraphic structure, stratigraphic lithology, topography, vegetation coverage, elevation and slope, and related data include rainfall, surface temperature and soil moisture content.

3. The regional geological disaster risk assessment method based on big data according to claim 1 is characterized in that: S2 includes the following steps: S21, inputting the geological disaster data set into the convolutional neural network to extract the features of the geological disaster data; S22, input the characteristics of geological disaster data into Bi-LSTM, train the Bi-LSTM through the PSO algorithm, and build a geological disaster risk assessment model based on the trained Bi-LSTM; In S21, the convolutional neural network includes an input layer, a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer, a fully connected layer and an output layer which are connected in sequence.

4. The regional geological disaster risk assessment method based on big data according to claim 3 is characterized in that: In S22, the method for training Bi-LSTM by PSO algorithm is as follows: S221. Set the size of the population according to the model parameters of Bi-LSTM, and initialize the population and the number of iterations; S222, calculating the fitness of particles and determining the optimal particle individual; S223, updating the position and speed of particles in the population according to the position of the optimal individual particle; S224, determine whether the number of iterations reaches the iteration threshold. If so, obtain the optimal learning rate and number of hidden units of Bi-LSTM according to the position of the current particle population, and build a geological disaster risk assessment model. If not, add 1 to the number of iterations and return to S222.

5. The regional geological disaster risk assessment method based on big data according to claim 4 is characterized in that: In S223, the expression for calculating the position of the i-th particle at the t+1-th iteration is specifically: In the formula, are the parameters of the perturbation mechanism, is the initial updated position of the ith particle, T is the iteration threshold, and c1 is the first learning factor; In the formula, is the position of the i-th particle at the t-th iteration, is the velocity of the i-th particle at the t+1-th iteration; In the formula, is the velocity of the ith particle at the tth iteration, is the inertia weight of the ith particle at the tth iteration, c2 is the second learning factor, is a random number, is the optimal position of the ith particle, is the historical optimal position of the population.

6. The regional geological disaster risk assessment method based on big data according to claim 5 is characterized in that: Parameters of the perturbation mechanism The specific expression is: In the formula, is the fitness function, and N is the total number of particles.

7. The regional geological disaster risk assessment method based on big data according to claim 6 is characterized in that: Inertia weight of the i-th particle at the t+1th iteration The specific expression is: In the formula, is the fitness threshold.

8. The regional geological disaster risk assessment method based on big data according to claim 1 is characterized in that: S3 is specifically: Collect new geological hazard data sets, which are geological data and related data within a set time period, input the new geological hazard data sets into the geological hazard risk assessment model, and predict the geological hazard risks generated by the geological data and related data.

9. The regional geological disaster risk assessment method based on big data according to claim 8 is characterized in that: In S3, the geological hazard risk assessment model includes a convolutional neural network, a Bi-LSTM and an attention mechanism layer connected in sequence. The weights are calculated through the attention mechanism layer, and the output results of the Bi-LSTM are weighted according to the weights.

10. A regional geological disaster risk assessment system based on big data, applied to the regional geological disaster risk assessment method based on big data as claimed in any one of claims 1 to 9, characterized in that: The system includes: Data acquisition module, used to obtain geological data and related data, and pre-process them to generate geological disaster data sets; Model training module, used to train Bi-LSTM with preprocessed geological disaster data set to build a geological disaster risk assessment model; The prediction module is used to collect new geological disaster data sets, input them into the geological disaster risk assessment model, and generate geological disaster risk prediction results.

Images