CN118013290B - Ionosphere TEC forecasting method, ionosphere TEC forecasting system, computer equipment and ionosphere TEC forecasting medium - Google Patents
Ionosphere TEC forecasting method, ionosphere TEC forecasting system, computer equipment and ionosphere TEC forecasting medium Download PDFInfo
- Publication number
- CN118013290B CN118013290B CN202410424074.1A CN202410424074A CN118013290B CN 118013290 B CN118013290 B CN 118013290B CN 202410424074 A CN202410424074 A CN 202410424074A CN 118013290 B CN118013290 B CN 118013290B
- Authority
- CN
- China
- Prior art keywords
- tec
- data
- ionosphere
- seagull
- forecasting
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000005433 ionosphere Substances 0.000 title claims abstract description 103
- 238000013277 forecasting method Methods 0.000 title claims description 27
- 238000013528 artificial neural network Methods 0.000 claims abstract description 63
- 238000000034 method Methods 0.000 claims abstract description 34
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 28
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000005457 optimization Methods 0.000 claims abstract description 26
- 230000000694 effects Effects 0.000 claims abstract description 19
- 238000004364 calculation method Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000001932 seasonal effect Effects 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 5
- 230000006399 behavior Effects 0.000 claims description 4
- 230000033001 locomotion Effects 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000011161 development Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims description 2
- 230000007423 decrease Effects 0.000 claims 1
- 239000010410 layer Substances 0.000 description 55
- 238000003062 neural network model Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 239000003607 modifier Substances 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000004907 flux Effects 0.000 description 1
- 239000003574 free electron Substances 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 239000011229 interlayer Substances 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01W—METEOROLOGY
- G01W1/00—Meteorology
- G01W1/10—Devices for predicting weather conditions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Environmental & Geological Engineering (AREA)
- Ecology (AREA)
- Environmental Sciences (AREA)
- Atmospheric Sciences (AREA)
- Biodiversity & Conservation Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention belongs to the technical field of ionosphere electron content prediction, and discloses an ionosphere TEC prediction method, an ionosphere TEC prediction system, computer equipment and a medium. The invention comprises the following steps: acquiring global ionosphere TEC data generated by a standard IONEX file format provided by the CODE and parameter data such as solar activity indexes F10.7 and Dst in a corresponding time period; extracting TEC data in an IONEX file format, and forming parameters such as TEC, F10.7, dst and the like into time sequence data to construct a data set as input data of model training; and optimizing the BP neural network based on an enhanced seagull optimization algorithm ESOA, building an ionosphere TEC forecasting model, and training the ionosphere TEC forecasting model by using the built dataset. According to the invention, the ESOA and BP neural network are combined to carry out ionosphere TEC forecast modeling, so that the accuracy of ionosphere TEC forecast is improved.
Description
Technical Field
The invention belongs to the technical field of ionosphere electron content prediction, and particularly relates to an ionosphere TEC prediction method, an ionosphere TEC prediction system, computer equipment and a medium.
Background
The total electron content (Total Electronic Content, TEC) of the ionosphere is the total electron content of the ionosphere per unit area, and represents the number of free electrons of the ionosphere, and is one of the important parameters describing the structure, state and change of the ionosphere. The time delay and the phase delay caused by refraction during radio wave propagation are closely related to the ionized layer TEC, and the radio wave propagation cannot be separated in the application fields of satellite communication, navigation positioning and the like, so that the method has important significance for analysis and research on the ionized layer TEC.
In order to evaluate the effect of using different methods on ionospheric parameter predictions, a great deal of research has been conducted. From the perspective of the prediction method, the ionospheric prediction model can be classified into a conventional prediction model and a neural network prediction model. Traditional predictive models include empirical models (international reference ionosphere models) and mathematical models (time series analysis models, autoregressive moving average models, multiple linear regression methods, autocorrelation analysis methods, interpolation methods of regional ionosphere predictions). With the rapid development of artificial neural networks, a new idea is provided for ionosphere forecasting. Many scholars use neural networks to effectively forecast TEC, however, the neural networks are complex in terms of optimization of parameter selection networks and the like, and are easy to sink into local minima.
Disclosure of Invention
The invention aims to provide an ionosphere TEC forecasting method, which is based on TEC data provided by a CODE, and utilizes an enhanced seagull optimization algorithm to improve a traditional BP neural network, so that the problems that the BP neural network is slow in searching speed and easy to fall into local optimum are solved, and the accuracy of the neural network model in forecasting the ionosphere TEC is improved.
In order to achieve the above purpose, the invention adopts the following technical scheme:
an ionosphere TEC forecasting method comprises the following steps:
step 1, acquiring global ionosphere TEC data generated by a standard IONEX file format provided by a CODE and solar activity index F10.7 and Dst parameter data of a corresponding time period;
Step 2, extracting TEC data in an IONEX file format, and forming the TEC data, the solar activity index F10.7 and Dst parameter data into time sequence data to construct a TEC data set as input data of model training;
Step 3, optimizing the weight and the threshold value of the BP neural network based on an enhanced seagull optimization algorithm ESOA, and building an ionosphere TEC forecast model based on the optimized BP neural network;
training the ionosphere TEC forecasting model constructed in the step 3 by using the TEC data set obtained in the step 2, and forecasting the ionosphere TEC by using the trained ionosphere TEC forecasting model.
In addition, on the basis of the ionosphere TEC forecasting method, the invention also provides an ionosphere TEC forecasting system corresponding to the ionosphere TEC forecasting method, which adopts the following technical scheme:
An ionospheric TEC forecasting system, comprising:
The data acquisition module is used for acquiring global ionosphere TEC data generated by a standard IONEX file format provided by the CODE, and solar activity index F10.7 and Dst parameter data of a corresponding time period;
The data preprocessing module is used for extracting TEC data in an IONEX file format, forming time series data from the TEC data, the solar activity index F10.7 and Dst parameter data, and constructing a TEC data set as input data of model training;
the ionized layer TEC forecasting module is used for optimizing the weight and the threshold value of the BP neural network based on an enhanced seagull optimization algorithm ESOA, and building an ionized layer TEC forecasting model based on the optimized BP neural network;
The method comprises the steps of training a constructed ionized layer TEC forecasting model by using a TEC data set obtained by a data preprocessing module, and forecasting the ionized layer TEC by using the trained ionized layer TEC forecasting model.
In addition, on the basis of the ionosphere TEC forecasting method, the invention also provides computer equipment for realizing the ionosphere TEC forecasting method.
The computer device comprises a memory and a processor, wherein executable codes are stored in the memory, and the processor is used for realizing the steps of the ionosphere TEC forecasting method when executing the executable codes.
In addition, on the basis of the ionosphere TEC forecasting method, the invention further provides a computer readable storage medium for realizing the ionosphere TEC forecasting method. The computer readable storage medium has stored thereon a program for implementing the steps of the ionosphere TEC forecasting method described above when the program is executed by a processor.
The invention has the following advantages:
As described above, the invention relates to an ionosphere TEC forecasting method, which is based on TEC data provided by a CODE, and utilizes an enhanced seagull optimization algorithm to improve a traditional BP neural network so as to solve the problems that the BP neural network is slow in searching speed and easy to fall into local optimum. The BP neural network learns the ability of adapting to data through nonlinear conversion among multiple layers of neurons, and is very effective in processing nonlinear data. The network structure can be adjusted according to the data set, the number of hidden layer nodes can be increased or decreased, or the topology structure can be changed. Furthermore, the network performance can be optimized by training and adjusting parameters. The enhanced seagull optimization algorithm (Enhanced Seagull Optimization Algorithm, ESOA) has certain advantages in search precision, convergence speed and stability, so that the invention provides the method for predicting the ionosphere by utilizing the model of ESOA optimized BP neural network, solves the problems that the BP neural network is easy to fall into a local optimal solution and the convergence speed is low when nonlinear fitting is performed, improves the global optimizing capability of the BP neural network algorithm, and improves the accuracy of model prediction by optimizing the weight and the threshold value of the BP neural network algorithm. The method of the invention optimizes by introducing the enhanced seagull optimization algorithm on the basis of not increasing the complexity of the traditional BP neural network model, thereby effectively solving the problem of reduced network searching speed when the BP neural network processes a large amount of data, and the enhanced seagull optimization algorithm with strong stability and high convergence speed is combined with the neural network, so that the optimal weight and threshold can be found, the problem of local optimization can be effectively avoided, and the performance and generalization capability of the model are improved.
Drawings
FIG. 1 is a flowchart of the ionosphere TEC forecasting method in embodiment 1 of the present invention.
Fig. 2 is a training schematic diagram of a BP neural network model optimized by ESOA in example 1 of the present invention.
Detailed Description
The invention is described in further detail below with reference to the attached drawings and detailed description:
Example 1
The embodiment 1 describes an ionosphere TEC forecasting method, so as to solve the problem that in the traditional BP neural network process, network searching speed is low and local optimization is easy to fall into, so that the accuracy of an ionosphere TEC forecasting model is low.
As shown in fig. 1, the ionosphere TEC forecasting method in this embodiment includes the following steps:
Step 1, global ionosphere TEC data generated in a standard IONEX (IONosphere map EXchange format) file format provided by the CODE are obtained, and solar activity index F10.7 and Dst parameter data of a corresponding time period are obtained.
Step 1.1. Download the TEC data in IONEX format with time resolution of 1h provided by the CODE.
Wherein the downloaded TEC data has a spatial longitude ranging from 180 ° western longitude to 180 ° eastern longitude and a resolution of 5 ° and a latitude ranging from 87.5 ° north latitude to 87.5 ° south latitude and a resolution of 2.5 °.
Step 1.2, selecting a geomagnetic index Dst as an index of overall geomagnetic activity and magnetic storm, and taking the geomagnetic index Dst as an index of solar activity by means of F10.7 index; dst and F10.7 indexes of the corresponding times are obtained.
Wherein the F10.7 index is 10.7 cm, solar radio flux at 2800 megahertz.
And 1.3, calculating LTS and LTC daily variation factors and DOYS and DOYC seasonal variation factors as input data of an ionosphere TEC forecasting model. The step 1.3 specifically comprises the following steps:
Step 1.3.1. The ionosphere TEC has obvious 24-hour daily change characteristic, and considering that jump occurs at night in local time, the LT in local time is split into two orthogonal inputs LTS and LTC, and the calculation formula of the daily change factor is as follows:
；
。
Step 1.3.2. Different seasons cause significant seasonal variation of TEC due to different zenith angles of the sun. Therefore, the annual date DOY is also converted into two orthogonal components, namely DOYS and DOYC, and the calculation formula of the seasonal variation factor is as follows:
；
。
and 2, extracting TEC data in an IONEX file format, and forming the TEC data, the solar activity index F10.7 and Dst parameter data into time sequence data to construct a TEC data set as input data of model training.
Specifically, MATLAB is used for processing TEC data files in the IONEX format, TEC values in the range of longitude and latitude are screened, parameters such as longitude and latitude, TEC, F10.7 and Dst are formed into time series data, and the TEC is labeled.
The step 2 specifically comprises the following steps:
And 2.1, opening a TEC data file in the IONEX format, reading file header information, checking whether a default value exists, ensuring data integrity, and preventing algorithm operation abnormality caused by data loss.
And 2.2. Analyzing TEC data, traversing each row of the global ionosphere TEC data file generated by the IONEX file format, and analyzing the content of each row. And extracting a numerical value in a required longitude and latitude range according to a data format, calculating the data to obtain TEC data corresponding to the longitude and latitude, storing the analyzed TEC data into an inx file, wherein the time resolution is 1 hour.
And 2.3, extracting longitude and latitude of the inx file and the corresponding TEC value, and then storing the time, TEC, dst and F10.7 indexes, the daily variation factor and the seasonal variation factor into an array to form time series data, so as to construct a TEC data set.
Specifically, dst and F10.7 parameters in corresponding time are downloaded, the time, longitude and latitude, TEC, dst, F10.7.7 and year and day parameters are correspondingly changed into time series one by one, and a data set is constructed, so that the data set is convenient to use in subsequent steps.
The input data for this model is a dataset containing 7 ionospheric data features.
And 3, optimizing the weight and the threshold value of the BP neural network based on an enhanced seagull optimization algorithm ESOA, and building an ionosphere TEC forecasting model based on the optimized BP neural network.
Training the ionosphere TEC forecasting model constructed in the step 3 by using the TEC data set obtained in the step 2, and forecasting the ionosphere TEC by using the trained ionosphere TEC forecasting model.
According to the invention, on the basis of not increasing the complexity of the traditional BP neural network model, the BP neural network is optimized by introducing the enhanced seagull optimization algorithm ESOA, the problem that the network searching speed is reduced when the neural network processes a large amount of data is effectively solved, the enhanced seagull optimization algorithm with strong stability and high convergence speed is combined with the BP neural network to find the optimal weight and the threshold, the problem of local optimization is effectively avoided, and the performance and the generalization capability of the model are improved.
The step 3 specifically comprises the following steps:
And 3.1, importing the TEC data set obtained in the step 2 as an input data set of the ionosphere TEC forecasting model, and dividing the input data set of the ionosphere TEC forecasting model into a training set and a testing set.
80% Of ionosphere data in the TEC dataset is used as a training set, and 20% of ionosphere data is used as a test set.
And carrying out normalization processing on the data in the TEC data set.
And 3.2, determining the topological structure of the ionosphere TEC forecast model built by the BP neural network, initializing the weight and the threshold of the BP neural network, and inputting the TEC data set obtained in the step 2 into the model for training.
And selecting the average absolute error of the training set and the testing set as an adaptive value function of optimizing.
Initializing parameters of an enhanced seagull optimization algorithm ESOA, wherein the number of dimensions of ESOA is equal to the total number of parameters of the BP neural network; and performing network global optimization by ESOA, and outputting optimal parameters after iteration is finished.
And constructing an enhanced seagull optimizer, and setting the initial weight and the paranoid matrix of the BP neural network as the seagull position.
The BP neural network global optimization is performed by using ESOA through a migration behavior and an attack behavior, and the process is as follows:
Step 3.3.1. Initialization parameters including upper and lower bounds of search space, seagull space dimension, seagull size and 。
And 3.3.2, randomly generating a seagull initial population, calculating the fitness value of all seagulls, and finding out the optimal seagull.
And 3.3.3. For each seagull, updating the position information of the seagull according to the formula (1) -the formula (11), then checking whether the updated position crosses the boundary or not and making corresponding adjustment, and finally calculating the fitness value.
The calculation formula for updating the position of the seagull is as follows:
（1）
Wherein the method comprises the steps of Representing the position where no collision occurs between seagulls,Storing the optimal solution and updating the positions of other seagulls, wherein C represents the movement behavior of the seagulls; the calculation of C is shown in formula (2).
（2）
Where n represents the number of iterations of ESOA algorithm, n=1, 2 …，The maximum number of iterations of the ESOA algorithm is represented,For controlling the frequency of C, C is fromLinearly decrementing to 0.
（3）
Wherein the method comprises the steps ofIndicating the current position of the seagull,Indicating the optimal seagull.
（4）
Wherein I represents a parameter that balances between seagull exploration and development, and ran is a random number in [0,1 ].
（5）
Wherein the method comprises the steps ofRepresenting the distance between the seagull and the optimal seagull,Indicating the direction of the optimal seagull.
（6）
（7）
（8）
Wherein R is the radius of each revolution of the spiral, k is a random number in the range [0,2 pi ] representing the attack angle; 、 、 respectively representing the motion trail of the seagull in the X, Y, Z direction.
（9）
Where A is a dynamic convergence factor and u and v are constants related to the shape of the helical flight trajectory.
（10）
Wherein the method comprises the steps ofRepresenting a random number in [ -1,1],Representing the current iteration number;
（11）
Wherein, Representing a random number vector based on a Levy distribution;
step 3.3.4, updating the optimal seagull through the step 3.3.3; repeating step 3.3.3 until the maximum number of iterations is reached And outputting the position information and the fitness value of the optimal seagull.
And 3.4. Assigning the optimal parameters ESOA, namely the optimal seagull positions, to the initial weight and the initial threshold of the BP neural network, and then performing network training to update the weight and the threshold of the neural network.
The formula for assigning the optimal seagull position information of ESOA to the initial weight and initial threshold of the BP neural network is as follows:
（12）
（13）
（14）
（15）
Wherein, For the initial input layer to hidden layer weight matrix,For the initial hidden layer threshold matrix,A weight matrix from the initial hidden layer to the output layer,A threshold matrix for the output layer; The optimal seagull position information obtained by ESOA algorithm in the ionosphere TEC prediction model is represented, Representing the number of input layer nodes of the BP neural network in the ionosphere TEC predictive model,Represents the hidden layer number of the BP neural network in the ionosphere TEC prediction model,And (5) representing the number of nodes of an output layer of the BP neural network in the ionosphere TEC prediction model.
The specific process of the step 3.4 is as follows:
step 3.4.1. Determining the input sample of the BP neural network And desired output data。
，For n number of input data,，For n output data.
Step 3.4.2, adding weights and calculating the ionosphere data of the input layer to determine the input of each node of the hidden layerAnd output of，AndThe calculation formula of (2) is as follows:
， ; wherein the method comprises the steps of Representing the connection weights of the nodes of the input layer and the nodes of the hidden layer,Representing the computation of the input data of the hidden layer in the hidden layer.
Step 3.4.3. After implicit layer computation, the data will be passed to the output layer, thus requiring determination of the input to the output layerAnd output of，AndThe calculation formula of (2) is as follows:
， ; wherein the method comprises the steps of Representing the connection weight of each node of the output layer and each node of the hidden layer,Representing the computation of the input data at the output layer.
The forward transmission stage of the BP neural network standard algorithm is completed in the steps 3.4.1-3.4.3.
Then, a reverse feedback stage is performed, i.e. steps 3.4.4 to 3.4.8, in which a gradient descent method is adopted, the output error meets the requirement by adjusting the weight each time, and the adjustment of the weight is to correct the bias of the hidden layer to the output layer and the input layer to the hidden layer by using an error function. The specific process is as follows:
step 3.4.4. Firstly, the weight value from the hidden layer to the output layer is adjusted, namely, the weight value of the error function pair is calculated Is a partial guide of (c).
；
Wherein,Representing the output of the output layer(s),Representing the error term of the output layer node.
Step 3.4.5. The derivation result of the input layer to implicit interlayer error function on the weight is:
；
Wherein, The input value is represented by a value of the input,An error term representing an hidden layer node.
And 3.4.6, after the weights corresponding to the input layer to the hidden layer and the hidden layer to the output layer are subjected to partial guide, correcting the connection weights of all the layers by using a partial guide result, and correcting the connection weights of all the nodes of the output layer and all the nodes of the hidden layer.
Firstly, obtaining a weight modifier：
。
Wherein,Representing the connection weight of each node of the output layer and each node of the hidden layer,And the super parameter representing the update step length of the control weight. Thus updated weightsTo update the sum of the pre-weight and the weight modifier.
。
And 3.4.7, correcting the connection weight of each node of the hidden layer and each node of the input layer.
Firstly, the weight correction quantity is obtained：
。
Thus, the updated weightsTo update the sum of the previous weight and the weight correction.
。
And 3.4.8, after each weight in the BP neural network is corrected, calculating the network error again.
And 3.5, model training is carried out on the ionosphere TEC forecast model built based on the optimized BP neural network.
And outputting a result after the training reaches the maximum iteration number, and then carrying out inverse normalization on the result to obtain a final prediction result, and evaluating the performance of the model through three different statistical indexes of RMSE, MAE and correlation coefficient rho.
The evaluation index formula is as follows:
（16）
（17）
（18）
Wherein, The value of ionosphere TEC issued for CODE, m represents the number of ionosphere TEC values,An ionospheric TEC value predicted for an ionospheric TEC prediction model; And The average value of the ionosphere TEC values and the average value of the model predicted values issued by the CODE are respectively.
According to the method, the ionosphere TEC forecast modeling is carried out by combining an enhanced seagull optimization algorithm with the BP neural network, so that parameters such as a given date and a solar activity index are realized, and the ionosphere TEC is accurately forecasted. And comparing the ionosphere TEC prediction result constructed by the traditional BP neural network with three evaluation indexes. The results are shown in Table 1.
Table 1 comparison of two forecast model indexes
As can be seen from the above Table 1, the method for forecasting the ionosphere TEC by combining ESOA provided by the invention with the BP neural network has superior forecasting performance under various indexes. Specifically, compared with the traditional BP neural network, the method adopting ESOA and BP neural network combination has smaller RMES and MAE values of 6.35 and 5.65 respectively, and the correlation coefficient value is higher and reaches 0.93, which shows that the method adopting ESOA and BP neural network combination to carry out ionosphere TEC prediction is more accurate in prediction precision and has a tighter linear relation with an actual value.
Example 2
Embodiment 2 describes an ionospheric TEC forecasting system based on the same inventive concept as the ionospheric TEC forecasting method described in embodiment 1.
Specifically, the ionosphere TEC forecasting system comprises:
The data acquisition module is used for acquiring global ionosphere TEC data generated by a standard IONEX file format provided by the CODE, and solar activity index F10.7 and Dst parameter data of a corresponding time period;
The data preprocessing module is used for extracting TEC data in an IONEX file format, forming time series data from the TEC data, the solar activity index F10.7 and Dst parameter data, and constructing a TEC data set as input data of model training;
the ionized layer TEC forecasting module is used for optimizing the weight and the threshold value of the BP neural network based on an enhanced seagull optimization algorithm ESOA, and building an ionized layer TEC forecasting model based on the optimized BP neural network;
The method comprises the steps of training a constructed ionized layer TEC forecasting model by using a TEC data set obtained by a data preprocessing module, and forecasting the ionized layer TEC by using the trained ionized layer TEC forecasting model.
It should be noted that, in the ionosphere TEC forecast system described in embodiment 2, the implementation process of the functions and roles of each functional module is specifically shown in the implementation process of the corresponding steps in the method in embodiment 1, and will not be described herein.
Example 3
Embodiment 3 describes a computer device for implementing the ionosphere TEC forecasting method described in embodiment 1 above.
In particular, the computer device includes a memory and one or more processors. Executable code is stored in the memory for implementing the steps of the ionosphere TEC forecasting method described above when executed by the processor.
In this embodiment, the computer device is any device or apparatus having data processing capability, which is not described herein.
Example 4
Embodiment 4 describes a computer readable storage medium for implementing the ionosphere TEC forecasting method described in embodiment 1 above.
Specifically, the computer readable storage medium in this embodiment 4 stores a program thereon, which when executed by a processor, is configured to implement the steps of the ionosphere TEC forecasting method described above.
The computer readable storage medium may be any internal storage unit of a device or apparatus having data processing capability, such as a hard disk or a memory, or may be any external storage device of a device having data processing capability, such as a plug-in hard disk, a smart memory card (SMART MEDIA CARD, SMC), an SD card, a flash memory card (FLASH CARD), or the like, provided on the device.
The foregoing description is, of course, merely illustrative of preferred embodiments of the present invention, and it should be understood that the present invention is not limited to the above-described embodiments, but is intended to cover all modifications, equivalents and alternatives falling within the spirit and scope of the present invention as defined by the appended claims.
Claims (6)
1. The ionosphere TEC forecasting method is characterized by comprising the following steps of:
step 1, acquiring global ionosphere TEC data generated by a standard IONEX file format provided by a CODE and solar activity index F10.7 and Dst parameter data of a corresponding time period;
the step 1 specifically comprises the following steps:
Step 1.1, downloading TEC data with IONEX format and time resolution of 1h provided by the CODE;
Wherein, the downloaded TEC data has a spatial longitude ranging from 180 degrees in western longitude to 180 degrees in eastern longitude and a resolution of 5 degrees, and a latitude ranging from 87.5 degrees in northern latitude to 87.5 degrees in southern latitude and a resolution of 2.5 degrees;
Step 1.2, selecting a geomagnetic index Dst as an index of overall geomagnetic activity and magnetic storm, and taking the geomagnetic index Dst as an index of solar activity by means of F10.7 index; obtaining Dst and F10.7 indexes of corresponding time;
Step 1.3, calculating LTS and LTC daily variation factors and DOYS and DOYC seasonal variation factors; wherein LTS, LTC are two orthogonal inputs split by LT when local, DOYS and DOYC are two orthogonal components of the yearday Doy transform;
Step 2, extracting TEC data in an IONEX file format, and forming the TEC data, the solar activity index F10.7 and Dst parameter data into time sequence data to construct a TEC data set as input data of model training;
the step 2 specifically comprises the following steps:
Step 2.1, opening a TEC data file in an IONEX format, reading file header information, and checking whether a default value exists;
Analyzing TEC data, traversing each row of the global ionosphere TEC data file generated by the IONEX file format, and analyzing the content of each row; according to the data format, extracting the numerical value in the required longitude and latitude range, calculating the data to obtain TEC data corresponding to the longitude and latitude, and finally storing the analyzed TEC data into an inx file;
Step 2.3, extracting longitude and latitude of an inx file and a corresponding TEC value, and then storing the time, TEC, dst and F10.7 indexes, a daily variation factor and a seasonal variation factor into an array to form time series data, so as to construct a TEC data set;
step 3, optimizing the weight and the threshold value of the BP neural network based on an enhanced seagull optimization algorithm ESOA, and building an ionosphere TEC forecast model based on the optimized BP neural network;
Training the ionosphere TEC forecasting model constructed in the step 3 by using the TEC data set obtained in the step 2, and forecasting the ionosphere TEC by using the trained ionosphere TEC forecasting model;
the step 3 specifically comprises the following steps:
step 3.1, 80% of ionosphere data in TEC data sets are used as training sets, and 20% are used as test sets;
carrying out normalization processing on the data in the TEC data set;
step 3.2, determining the topological structure of an ionosphere TEC forecast model built by the BP neural network, initializing the weight and the threshold of the BP neural network, and inputting the TEC data set obtained in the step 2 into the model for training;
the average absolute error of the training set and the testing set is selected as an adaptive value function of optimizing;
Initializing parameters of an enhanced seagull optimization algorithm ESOA, wherein the number of dimensions of ESOA is equal to the total number of parameters of the BP neural network; performing network global optimization by ESOA, and outputting optimal parameters after iteration is finished;
ESOA the global optimization process is as follows:
Initializing parameters including upper and lower boundaries of a search space, a seagull space dimension, a seagull scale and max i;
Step 3.3.2, randomly generating initial sea-gull population, calculating the fitness value of all sea-gulls, and finding out the optimal sea-gulls;
Step 3.3.3. For each seagull, updating the position information of the seagull according to the formulas (1) to (11), then checking whether the updated position crosses the boundary or not and making corresponding adjustment, and finally calculating a fitness value;
the calculation formula for updating the position of the seagull is as follows:
Wherein the method comprises the steps of Representing the position of no collision between seagulls,Storing the optimal solution and updating the positions of other seagulls, wherein C represents the movement behavior of the seagulls; the calculation of C is shown in formula (2);
where n represents the number of iterations of the ESOA algorithm, n=1, 2 … max i,maxi represents the maximum number of iterations of the ESOA algorithm, f c is used to control the frequency of C, which decreases linearly from f c =2 to 0;
Wherein the method comprises the steps of Representing the current position of seagull,Representing the optimal seagull;
I＝2×C2×ran (4)
Wherein I represents a parameter that balances between seagull exploration and development, and ran is a random number in [0,1 ];
Wherein the method comprises the steps of Representing the distance between the seagull and the optimal seagull,The direction of the optimal seagull is indicated;
X′＝Rcos(k) (6)
Y′＝Rsin(k) (7)
Z′＝Rk (8)
wherein R is the radius of each revolution of the spiral, k is a random number in the range [0,2 pi ] representing the attack angle; x ', Y ', Z ' respectively represent the motion trail of the seagull in the X, Y, Z direction;
R＝Auekν (9)
wherein A is a dynamic convergence factor, u and v are constants related to the shape of the helical flight trajectory;
A＝a(1-(iter(1/maxi))) (10)
wherein a represents a random number in [ -1,1], and iter represents the current iteration number;
wherein L represents a random number vector based on Levy distribution;
Step 3.3.4, updating the optimal seagull through the step 3.3.3; repeating the step 3.3.3 until the maximum iteration number max i is reached, and outputting the position information and the fitness value of the optimal seagull;
Step 3.4, assigning the optimal parameters of ESOA to weights and thresholds of the BP neural network;
step 3.5, model training is carried out on an ionosphere TEC forecast model built based on the optimized BP neural network;
And outputting a result after the training reaches the maximum iteration number, and then carrying out inverse normalization on the result to obtain a final prediction result, and evaluating a model through three different statistical indexes of RMSE, MAE and correlation coefficient rho.
2. The ionospheric TEC forecasting method according to claim 1, characterized in that,
The step 3.4 specifically comprises the following steps:
the calculation formula for assigning the optimal seagull position information of ESOA to the weight and the threshold value of the BP neural network is as follows:
W1＝x(1:inputnum hiddennum) (12)
B1＝x(inputnum×hiddennum+1:inputnum×hiddennum+hiddennum) (13)
Wherein, W 1 is the weight matrix from the initial input layer to the hidden layer, B 1 is the threshold matrix of the initial hidden layer, W 2 is the weight matrix from the initial hidden layer to the output layer, and B 2 is the threshold matrix of the output layer; x represents the optimal seagull position information obtained by ESOA algorithm in the ionosphere TEC prediction model, inputnum represents the number of input layer nodes of the BP neural network in the ionosphere TEC prediction model, hiddennum represents the hidden layer number of the BP neural network in the ionosphere TEC prediction model, and outputnum represents the number of output layer nodes of the BP neural network in the ionosphere TEC prediction model.
3. The ionospheric TEC forecasting method according to claim 1, characterized in that,
The step 3.5 specifically comprises the following steps:
The evaluation index formula is as follows:
Wherein, The ionosphere TEC value issued for CODE, m represents the number of ionosphere TEC values,An ionospheric TEC value predicted for an ionospheric TEC prediction model; /(I)AndThe average value of the ionosphere TEC values and the average value of the model predicted values issued by the CODE are respectively.
4. An ionospheric TEC forecasting system for implementing the ionospheric TEC forecasting method of any one of claims 1 to 3, characterized in that the ionospheric TEC forecasting system comprises:
The data acquisition module is used for acquiring global ionosphere TEC data generated by a standard IONEX file format provided by the CODE, and solar activity index F10.7 and Dst parameter data of a corresponding time period;
The data preprocessing module is used for extracting TEC data in an IONEX file format, and forming time series data from the TEC data, the solar activity index F10.7 and Dst parameter data to construct a TEC data set as input data of model training;
the ionized layer TEC forecasting module is used for optimizing the weight and the threshold value of the BP neural network based on an enhanced seagull optimization algorithm ESOA, and building an ionized layer TEC forecasting model based on the optimized BP neural network;
The method comprises the steps of training a constructed ionized layer TEC forecasting model by using a TEC data set obtained by a data preprocessing module, and forecasting the ionized layer TEC by using the trained ionized layer TEC forecasting model.
5. A computer device comprising a memory and a processor, the memory having executable code stored therein, wherein the processor, when executing the executable code, is adapted to implement the steps of the ionosphere TEC forecasting method of any one of claims 1 to 3.
6. A computer readable storage medium having stored thereon a program, which when executed by a processor is adapted to carry out the steps of the ionosphere TEC forecasting method of any one of the preceding claims 1 to 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410424074.1A CN118013290B (en) | 2024-04-10 | 2024-04-10 | Ionosphere TEC forecasting method, ionosphere TEC forecasting system, computer equipment and ionosphere TEC forecasting medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410424074.1A CN118013290B (en) | 2024-04-10 | 2024-04-10 | Ionosphere TEC forecasting method, ionosphere TEC forecasting system, computer equipment and ionosphere TEC forecasting medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118013290A CN118013290A (en) | 2024-05-10 |
| CN118013290B true CN118013290B (en) | 2024-06-25 |
Family
ID=90947469
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410424074.1A Active CN118013290B (en) | 2024-04-10 | 2024-04-10 | Ionosphere TEC forecasting method, ionosphere TEC forecasting system, computer equipment and ionosphere TEC forecasting medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118013290B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116805516A (en) * | 2023-06-29 | 2023-09-26 | 武汉大学 | Ionosphere intelligent forecasting and monitoring method and system based on integrated model |
| CN117236746A (en) * | 2023-09-01 | 2023-12-15 | 安徽理工大学 | Air conditioner load adjustable potential evaluation method |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113139327B (en) * | 2020-01-19 | 2024-03-12 | 中国科学院国家空间科学中心 | Ionized layer TEC single-point prediction method and system based on GRU network model |
| CN111967679B (en) * | 2020-08-20 | 2023-12-08 | 江苏师范大学 | Ionosphere total electron content forecasting method based on TCN model |
| CN113379107B (en) * | 2021-05-26 | 2023-12-19 | 江苏师范大学 | Regional ionosphere TEC forecasting method based on LSTM and GCN |
| CN115374710A (en) * | 2022-09-19 | 2022-11-22 | 兰州理工大学 | Wind speed prediction method based on improved gull optimization algorithm and multi-core extreme learning machine |
| CN117217069A (en) * | 2023-06-13 | 2023-12-12 | 哈尔滨工业大学 | Radar equipment system deployment point location optimization method based on seagull optimization algorithm |
-
2024
- 2024-04-10 CN CN202410424074.1A patent/CN118013290B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116805516A (en) * | 2023-06-29 | 2023-09-26 | 武汉大学 | Ionosphere intelligent forecasting and monitoring method and system based on integrated model |
| CN117236746A (en) * | 2023-09-01 | 2023-12-15 | 安徽理工大学 | Air conditioner load adjustable potential evaluation method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118013290A (en) | 2024-05-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111784041B (en) | Wind power prediction method and system based on graph convolution neural network | |
| CN111967679B (en) | Ionosphere total electron content forecasting method based on TCN model | |
| Sikorska-Senoner et al. | A novel ensemble-based conceptual-data-driven approach for improved streamflow simulations | |
| CN111563706A (en) | Multivariable logistics freight volume prediction method based on LSTM network | |
| CN111045046B (en) | Short-term ionosphere forecasting method and device based on NARX | |
| CN113379107B (en) | Regional ionosphere TEC forecasting method based on LSTM and GCN | |
| CN110443417A (en) | Multi-model integrated load prediction method based on wavelet transformation | |
| CN115347571B (en) | Photovoltaic power generation short-term prediction method and device based on transfer learning | |
| CN112488396A (en) | Wavelet transform-based electric power load prediction method of Holt-Winters and LSTM combined model | |
| US11886967B1 (en) | Long-term streamflow forecast method and system based on process-data synergic drive | |
| CN114330935B (en) | New energy power prediction method and system based on multiple combination strategies integrated learning | |
| CN114493050A (en) | Multi-dimensional fusion new energy power parallel prediction method and device | |
| CN112884236B (en) | Short-term load prediction method and system based on VDM decomposition and LSTM improvement | |
| CN112215404A (en) | Prediction method and device based on phase space reconstruction and ensemble empirical mode decomposition | |
| CN117293809A (en) | Multi-time space scale new energy generation power prediction method based on large model | |
| CN117076738A (en) | Medium-long term weather prediction method, system and medium based on variable grid model | |
| CN115327662A (en) | Statistical learning-based regional ionized layer TEC prediction method | |
| CN110852415A (en) | Vegetation index prediction method, system and equipment based on neural network algorithm | |
| CN118035777A (en) | Air pollutant concentration prediction method and device based on space-time diagram data | |
| CN118013290B (en) | Ionosphere TEC forecasting method, ionosphere TEC forecasting system, computer equipment and ionosphere TEC forecasting medium | |
| CN116933083A (en) | Ionosphere total electron content prediction method, ionosphere total electron content prediction system, electronic equipment and medium | |
| CN116108743A (en) | Training method and device for ionosphere electronic forecasting model and computer equipment | |
| CN115860165A (en) | Neural network basin rainfall runoff forecasting method and system considering initial loss | |
| CN112446550B (en) | Short-term building load probability density prediction method | |
| CN113723712A (en) | Wind power prediction method, system, device and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |