CN-122001503-A - Short-wave communication highest available frequency prediction method and system based on long-short-term prediction data fusion processing

CN122001503ACN 122001503 ACN122001503 ACN 122001503ACN-122001503-A

Abstract

The application discloses a short-wave communication highest available frequency prediction method based on long-term and short-term prediction data fusion processing, which comprises the steps of collecting F2 layer critical frequency in vertical detection data as long-term prediction data, identifying and marking abnormal values in original data, replacing the abnormal values by long-term prediction data at the same time of the month, taking data in the vicinity of the replaced abnormal values as short-term prediction data, carrying out data fusion on the long-term prediction data and the short-term prediction data to obtain continuous time sequence data, and inputting the continuous time sequence data into a trained LSTM neural network model to obtain short-wave highest available frequency in future time. The application uses Kalman filtering recursion to estimate real-time fusion observation, dynamically suppresses wild values, maintains the optimal steady state of the linear system, predicts the highest available frequency by adopting a data fusion technology, combines the characteristics of short-term prediction and long-term prediction, and can more accurately predict the highest available frequency of short-wave communication in a longer time.

Inventors

XIONG TAO
XU XIAOTAO
LU XUN
CUI MINGYU
ZHANG JIYUAN
Liang Xuenan

Assignees

中国人民解放军信息支援部队工程大学

Dates

Publication Date: 20260508
Application Date: 20251231

Claims (10)

1. A short-wave communication highest available frequency prediction method based on long-short-term prediction data fusion processing is characterized by comprising the following steps: s101, collecting critical frequency of an F2 layer in vertical detection data as long-term prediction data; s201, identifying and marking abnormal values in original data, replacing the abnormal values by long-term prediction data at the same time of the month, and taking data of a near day after replacing the abnormal values as short-term prediction data; S301, carrying out data fusion on the long-term prediction data and the short-term prediction data to obtain continuous time sequence data; s401, inputting the continuous time series data into a trained LSTM neural network model to obtain the shortwave highest available frequency in future time.
2. The method for predicting the highest available frequency of short-wave communication as set forth in claim 1, wherein the step S201 specifically includes: Identifying and marking abnormal values in the original data, using N values which are consecutive before the abnormal values as inputs of Kalman filtering prediction, and replacing the abnormal values by output values of the Kalman filtering prediction to ensure the continuity and integrity of time series data, wherein N is a positive integer.
3. The method for predicting the highest available frequency for short-wave communication as recited in claim 2, further comprising: And (3) performing ionosphere parameter inversion on the critical frequency of the F2 layer of the vertical detection on the near day to obtain the highest available frequency on the near day.
4. The method for predicting the highest available frequency of short-wave communication as set forth in claim 1, wherein step S301 specifically includes: and splicing and fusing the long-term prediction data and the highest available frequency obtained by the vertical detection data to obtain continuous time sequence data.
5. The method for predicting the highest available frequency of short-wave communication as claimed in claim 1, wherein the step S401 further comprises: Creating an LSTM neural network model, and configuring the LSTM layer number, the neuron number and the activation function; The continuous time series of highest available frequencies of the fusion data is taken as input, and the highest available frequencies of the prediction days are taken as output.
6. The method for predicting the highest available frequency of short-wave communication according to claim 5, wherein the training of the LSTM neural network model specifically comprises: dividing data into a training set, a verification set and a test set; Training the LSTM neural network model by using training set data, and adjusting network weights by a back propagation algorithm to minimize a prediction error; and respectively verifying and testing the trained LSTM neural network model by using a verification set and a test set.
7. The method for predicting the highest available frequency of short-wave communication according to claim 6, wherein the workflow of the LSTM neural network model specifically comprises: S4011, receiving input data in each time step, and hiding state and memory unit state in last time step by using LSTM neural network; s4012, determining information added to the memory unit and information deleted from the memory unit by an input gate and a forget gate of the LSTM neural network; S4013, updating the state of the memory unit, and updating and integrating the information in the memory unit; S4014, the output gate determines the information read from the memory unit and takes it as the output information of the current time step.
8. A short-wave communication highest available frequency prediction system based on long-short-term prediction data fusion processing, comprising: A long-term prediction data acquisition module configured to collect an F2 layer critical frequency in the vertical detection data as long-term prediction data; A short-term predicted data acquisition module configured to identify and flag an outlier in the original data, replace the outlier with long-term predicted data at the same time of the month, and use data of a near day after replacement of the outlier as short-term predicted data; The data fusion module is configured to perform data fusion on the long-term prediction data and the short-term prediction data to obtain continuous time sequence data; a data prediction module configured to input the continuous time series data into a trained LSTM neural network model resulting in a shortwave highest available frequency for future time.
9. A short-wave communication highest available frequency prediction device based on a long-short term prediction data fusion process, characterized by comprising at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program which, when executed by the processing unit, causes the processing unit to perform the steps of the method according to any of claims 1-7.
10. A computer program product comprising computer programs/instructions which, when executed by a processor, implement the steps of the method of any of claims 1 to 7.

Description

Short-wave communication highest available frequency prediction method and system based on long-short-term prediction data fusion processing Technical Field The application relates to the technical field of communication, in particular to a short-wave communication highest available frequency prediction method and system based on long-short-period prediction data fusion processing. Background The highest available frequency (Maximum Usable Frequency, MUF) in short-wave communications refers to the highest frequency of radio waves that can be reflected by the ionosphere back to the ground under certain ionosphere conditions and communication distances. ‌ if the frequency of the wave exceeds the MUF, the wave penetrates the ionosphere into space, so that sky wave propagation cannot be realized, while the wave with the frequency lower than the MUF can be reflected, but multipath effect and signal attenuation are considered. The value of ‌ MUF is affected by a number of factors including ionospheric electron density distribution, solar activity period, season, diurnal variation, and communication distance. ‌ A In the prior art, the average absolute error of the short-term prediction MUF is large by adopting a long-term prediction wild value mode. Because long-term predictive models employ, for example, month median, critical frequencies of ionosphere layers, etc. While ionospheric models built by integrating these data can represent the average ionospheric state relatively reasonably, such models do not have high prediction accuracy when predicting ionospheric changes at specific times. This is because the ionosphere undergoes significant changes in a short period of time, which changes are not only reflected in its spatial distribution of longitude and latitude, but also by factors such as the daytime, season and solar cycle. These factors act together to cause the state of the ionosphere to change drastically in a short time, making accurate predictions difficult by simple models. The average error of short-term prediction MUF by long-term prediction horizon processing is large. Along with the deep ionosphere research and the continuous improvement of the requirements of system application on the quality of short-wave communication, the improvement of the prediction precision of the highest available frequency (MUF) of the short-wave communication becomes a problem to be solved in the field of short-wave communication. Disclosure of Invention Aiming at least one defect or improvement demand of the prior art, the invention provides a method, a system, equipment and a storage medium for predicting the highest available frequency of short-wave communication based on long-term and short-term prediction data fusion processing, which are characterized in that abnormal values are replaced by Kalman filtering to ensure continuity and accuracy of time series data in short-term prediction, the advantages of long-term prediction and short-term prediction are fully utilized by adopting a data fusion method, the long-term series data is predicted by using an LSTM model, and the above various means are organically combined, so that the prediction precision of the highest available frequency (MUF) of short-wave communication is greatly improved. In order to achieve the above object, according to a first aspect of the present invention, there is provided a method for predicting a highest available frequency of short-wave communication based on a long-short-term prediction data fusion process, comprising the steps of: s101, collecting critical frequency of an F2 layer in vertical detection data as long-term prediction data; s201, identifying and marking abnormal values in original data, replacing the abnormal values by long-term prediction data at the same time of the month, and taking data of a near day after replacing the abnormal values as short-term prediction data; S301, carrying out data fusion on the long-term prediction data and the short-term prediction data to obtain continuous time sequence data; s401, inputting the continuous time series data into a trained LSTM neural network model to obtain the shortwave highest available frequency in future time. Further, in the above method for predicting the highest available frequency of short-wave communication, step S201 specifically includes: Identifying and marking abnormal values in the original data, using N values which are consecutive before the abnormal values as inputs of Kalman filtering prediction, and replacing the abnormal values by output values of the Kalman filtering prediction to ensure the continuity and integrity of time series data, wherein N is a positive integer. Further, the method for predicting the highest available frequency of short-wave communication further comprises the following steps: And (3) performing ionosphere parameter inversion on the critical frequency of the F2 layer of the vertical detection on the near day to obtain the highest avai