Dam structure safety monitoring accurate prediction method and system based on RBF neural network
Technical Field
The invention relates to the technical field of dam operation safety monitoring, in particular to a dam structure safety monitoring accurate prediction method and system based on a Radial Basis Function (RBF) neural network.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
The dam construction is a systematic project which has huge workload and needs to carry out precise analysis on complex structures and foundation conditions, contains a lot of uncertain and random factors, and even the conditions encountered after normal operation are varied, so that all the factors influencing the structural form cannot be considered in the prior design and the best and most precise calculation and judgment can be made. Meanwhile, the quality of the dam cannot be perfect due to various reasons in the engineering construction process, the dam can be influenced by factors such as aging of used materials, geological and topographic changes, climate changes and the like in the construction and application processes, so that the safety of the hydroelectric engineering is guaranteed, and the safety monitoring of the dam is vital to measure and observe the main body structure, foundation, side slope, related facilities and the surrounding environment of the dam through electronic instrument observation and manual inspection tour and diagnosis, analysis, evaluation and monitoring of the safety of the dam through detection data.
The inventor finds that the safety prediction of the existing dam structure mainly comprises the following aspects: stepwise regression statistical model, gray model. However, these methods cannot predict dam displacement conveniently and accurately due to various defects of the methods. For example, a stepwise regression statistical model is a mathematical expression between a main influence factor of dam displacement and dam displacement according to years of research of a plurality of experts, and although data can be processed in the aspect of model establishment, the prediction effect is poor when abnormal points exist in the predicted data; the grey model has strong dependence on historical data, does not consider the relation between influence factors, and generates larger deviation in medium-long prediction.
Disclosure of Invention
The invention aims to solve the defects of the prior art, and provides a dam structure safety monitoring and accurate prediction method and system based on a Radial Basis Function (RBF) neural network.
In some embodiments, the following technical scheme is adopted:
a dam structure safety monitoring accurate prediction method based on an RBF neural network comprises the following steps:
acquiring deformation observation data of a dam to be detected within a set time at a certain measuring point;
carrying out data preprocessing on the obtained data;
establishing a neural network prediction model, and merging the deformation observation data vector and the dam horizontal displacement value vector to obtain a learning sample of the neural network prediction model;
dividing the learning sample into a training set and a verification set, and training the established neural network prediction model; and finally outputting a dam displacement predicted value.
As a further improvement, the method for acquiring deformation observation data of the dam to be measured in a set time at a certain measuring point specifically comprises the following steps: the dam horizontal displacement of the measuring point is obtained by dividing the accumulated days from the observation day to the measurement day by 100.
As a further improvement scheme, before the learning samples are divided into a training set and a verification set, data normalization processing is carried out on the learning samples of the obtained neural network prediction model.
As a further improvement scheme, the normalized water pressure component, temperature component and aging component are used as input layers of a neural network prediction model, and the dam horizontal displacement is used as an output layer of the neural network prediction model.
As a further improvement scheme, the result output by the network prediction model is subjected to inverse normalization processing to obtain the predicted value of dam displacement.
As a further improvement, the established neural network prediction model specifically comprises: and (4) an RBF neural network prediction model.
In other embodiments, the following technical solutions are adopted:
a dam structure safety monitoring accurate prediction system based on an RBF neural network comprises:
the module is used for collecting deformation observation data of the dam to be measured within a set time at a certain measuring point;
a module for preprocessing the data obtained;
a module for establishing a neural network prediction model, and merging the deformation observation data vector and the dam horizontal displacement value vector to obtain a learning sample of the neural network prediction model;
the neural network prediction model is used for dividing the learning samples into a training set and a verification set and training the established neural network prediction model; and finally outputting the dam displacement predicted value.
In other embodiments, the following technical solutions are adopted:
a terminal device comprising a processor and a computer-readable storage medium, the processor being configured to implement instructions; the computer readable storage medium is used for storing a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the RBF neural network-based dam structure safety monitoring accurate prediction method.
In other embodiments, the following technical solutions are adopted:
a computer readable storage medium, wherein a plurality of instructions are stored, and the instructions are suitable for being loaded by a processor of a terminal device and executing the RBF neural network-based dam structure safety monitoring accurate prediction method.
Compared with the prior art, the invention has the beneficial effects that:
(1) the data utilized by the RBF prediction model is obtained from actual measurement deformation observation data of the dam for 6 continuous years, and the RBF prediction model has the characteristics of large data volume, long data coverage time, real data, high reliability and the like.
(2) The MATLAB neural network toolbox almost summarizes all types and learning algorithms of the existing neural network, and the RBF prediction model established by using MATLAB is intelligent and convenient to operate and high in calculation speed.
(4) The RBF neural network technology has strong self-learning capability, can be directly input and calculated when new data is updated, is simple to operate and does not need repeated modeling. And the more data, the higher the prediction accuracy of the model.
(5) The RBF neural network model has simple representation form, and does not increase too much complexity even for multivariable input; the radial symmetry is realized, the smoothness is good, and derivatives of any order exist; the basic function representation is simple, the analytic performance is good, and the theoretical analysis is convenient to carry out.
Drawings
FIG. 1 is a schematic flow chart of a dam displacement prediction method according to the present invention;
FIG. 2 is a diagram of an RBF neuron model according to the prediction model of the present invention;
FIG. 3 is a block diagram of an RBF neural network according to the prediction model of the present invention;
FIG. 4 is a graph of RBF neural network training results according to the test model of the present invention;
FIG. 5 is a diagram of prediction of RBF neural network according to the prediction model of the present invention.
Detailed Description
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
The embodiments and features of the embodiments of the present invention may be combined with each other without conflict.
Example one
In one or more embodiments, an RBF neural network technology-based dam structure security monitoring accurate prediction method is disclosed, and referring to fig. 1, in order to more clearly illustrate the implementation example, a row-level security access control implementation process may be specifically described as follows:
(1) collecting actual measurement deformation data of the dam: selecting a dam, and collecting measured deformation data of a certain measuring point of the dam for 6 continuous years, wherein each measured data comprises the upstream water level elevation (H) of the dam, the dam body temperature (T) of the measuring point, the accumulated days from the current day to the measuring day divided by 100 (theta), and the horizontal displacement (y) of the dam of the measuring point. The applicant collects 93 groups of field measured data by carrying out field measurement (the observation time is 2013-2018) on the 13 th point of the DZ dam 6# dam body, consulting documents and the like.
Example of implementation, field data are shown in table 1.
TABLE 1 actual measurement data table for dam
Numbering
|
H/m
|
T/℃
|
θ
|
y/mm
|
Numbering
|
H/m
|
T/℃
|
θ
|
y/mm
|
1
|
1745.12
|
15.7
|
1.09
|
0.76
|
51
|
1775.84
|
17.9
|
9.44
|
45.54
|
2
|
1748.17
|
12.3
|
1.24
|
1.91
|
52
|
1768.91
|
17.9
|
9.74
|
37.48
|
3
|
1752.84
|
16.4
|
1.55
|
3.91
|
53
|
1764.35
|
17.9
|
10.05
|
57.38
|
4
|
1752.73
|
19.8
|
1.68
|
5.35
|
54
|
1771.85
|
17.8
|
10.35
|
58.2
|
5
|
1753.64
|
18.4
|
1.83
|
9.95
|
55
|
1786
|
17.8
|
10.66
|
58.38
|
6
|
1755.66
|
16.6
|
2.14
|
6.06
|
56
|
1789.28
|
17.8
|
10.97
|
65.28
|
7
|
1751.09
|
22.7
|
2.44
|
0
|
57
|
1790.13
|
17.8
|
11.27
|
58.38
|
8
|
1743.74
|
29.2
|
2.75
|
4.05
|
58
|
1790.24
|
17.7
|
11.58
|
72.64
|
9
|
1752.73
|
27.3
|
3.05
|
3.82
|
59
|
1789.07
|
17.7
|
11.88
|
73.54
|
…
|
|
|
|
|
…
|
|
|
|
|
50
|
1789.59
|
18.1
|
8.54
|
53.4
|
93
|
1785.81
|
17.1
|
21.99
|
68.9 |
(2) Data drying: performing data preprocessing on each measured deformation observation data, wherein the number of water pressure components is 3, and the number of the water pressure components is H, H2、H3Are respectively marked as x1、x2、x3(ii) a Temperature component 1, i.e. T, denoted x4(ii) a The aging components are 2, are respectively theta and ln theta and are respectively marked as x5、x6。
After the data is preprocessed, the resulting data is arranged in a prescribed format, see table 2.
TABLE 2 Pre-processing data
Numbering
|
H
|
H2 |
H3 |
T
|
θ
|
lnθ
|
y
|
1
|
1745.12
|
3045443.814
|
5314664909
|
15.7
|
1.09
|
0.086177696
|
0.76
|
2
|
1748.17
|
3056098.349
|
5342579451
|
12.3
|
1.24
|
0.21511138
|
1.91
|
3
|
1752.84
|
3072448.066
|
5385509867
|
16.4
|
1.55
|
0.438254931
|
3.91
|
4
|
1752.73
|
3072062.453
|
5384496023
|
19.8
|
1.68
|
0.518793793
|
5.35
|
5
|
1753.64
|
3075253.25
|
5392887109
|
18.4
|
1.83
|
0.604315967
|
9.95
|
6
|
1755.66
|
3082342.036
|
5411544618
|
16.6
|
2.14
|
0.760805829
|
6.06
|
7
|
1751.09
|
3066316.188
|
5369395614
|
22.7
|
2.44
|
0.891998039
|
0
|
8
|
1743.74
|
3040629.188
|
5302066740
|
29.2
|
2.75
|
1.011600912
|
4.05
|
9
|
1752.73
|
3072062.453
|
5384496023
|
27.3
|
3.05
|
1.115141591
|
3.82
|
…
|
|
|
|
|
|
|
|
93
|
1785.81
|
3189117.356
|
5695157666
|
17.1
|
21.99
|
3.090587805
|
68.9 |
(3) Learning sample normalization processing: although the learning samples in table 2 have the same dimension, the data sizes are different, and the amplitude has a great influence on the training precision of the neural network. Normalization processing is carried out according to the formula 1, training data are normalized to [ -1,1], and the influence of different types of data dimension and amplitude on training precision can be avoided. After the neural network training is finished, inverse normalization processing is required according to a multiple output result of the formula 2.
In the formula: x (i) is the raw data, x (i)' is the normalized data, xmaxIs the maximum value, x, in the raw dataminIs the minimum value in the raw data, y (i) is the raw output value of the neural network, y (i)' is the data after inverse normalization, ymaxIs the maximum value in the output data, yminIs the minimum value in the output data.
(4) Training a neural network: and dividing the normalized learning sample into a training data set and a verification data set according to the proportion of 5: 1. The training data and the verification data are brought into a neural network model to complete the neural network training and verification work, the neuron number of an input layer of the neural network model is 6, and the neuron number is respectively a water pressure component H, H2、H3Temperature component T, aging component theta, ln theta; the number of neurons in the output layer is 1, namely the horizontal displacement y of the dam; the number of hidden layer neurons is 5, as contemplated by the present applicationThe RBF neuron model involved in the measurement method is shown in fig. 2, and it can be seen from the figure that the radial basis network transfer function radbas is obtained by taking the distance | dist | between the weight vector and the threshold vector as an argument, wherein | dist | is obtained by multiplying the input vector by the row vector of the weighting matrix; the RBF neural network structure framework is shown in fig. 3, which clearly shows all the information of the RBF neural network model, such as the number of distance between the weight vector and the threshold vector | dist |, as the number of input layer neurons, the number of hidden layer neurons, the number of output layer neurons, etc.
An RBF neural network model is established by using an MATLAB neural network tool box, the target value of the mean square error of the network is 0.1, the distribution coefficient of the radial basis function is 0.1, the maximum number of neurons is 6, and the number of added neurons between two displays is 25.
And (3) bringing the sorted training set data into an RBF neural network model, establishing a neural network, bringing the sorted verification data into the established RBF neural network model, and carrying out reverse normalization processing on the output data to obtain a predicted value of dam displacement. The training result graph and the prediction are shown in fig. 4 and 5, and the prediction graph shows that the change of the prediction value point is consistent with the verification data point, so that the actual requirement is completely met. The model is written into an MATLAB program, the program contains a cross validation subprogram, field measured data can be directly input into the model, the model can quickly call the program to perform data screening analysis, and finally dam body displacement of the dam is predicted.
Example two
In one or more embodiments, disclosed is an RBF neural network-based dam structure safety monitoring accurate prediction system, including:
the module is used for collecting deformation observation data of the dam to be measured within a set time at a certain measuring point;
a module for preprocessing the data obtained;
a module for establishing a neural network prediction model, and merging the deformation observation data vector and the dam horizontal displacement value vector to obtain a learning sample of the neural network prediction model;
the neural network prediction model is used for dividing the learning samples into a training set and a verification set and training the established neural network prediction model; and finally outputting the dam displacement predicted value.
In some embodiments, the model of this embodiment can be written as a MATLAB program to automatically perform the steps of embodiment one.
In other embodiments, the steps of the first embodiment may be implemented in a terminal device, where the terminal device includes a processor and a computer-readable storage medium, and the processor is configured to implement the instructions; the computer readable storage medium is used for storing a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the concrete panel rock-fill dam horizontal displacement prediction accurate method in the first embodiment.
In other embodiments, the steps of the first embodiment may be implemented in a computer-readable storage medium, where a plurality of instructions are stored, and the instructions are adapted to be loaded by a processor of a terminal device and to execute the RBF neural network-based dam structure safety monitoring and accurate prediction method.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, it is not intended to limit the scope of the present invention, and it should be understood by those skilled in the art that various modifications and variations can be made without inventive efforts by those skilled in the art based on the technical solution of the present invention.