CN113126038B - High-frequency ground wave radar working frequency optimization method, system, storage medium and application - Google Patents
High-frequency ground wave radar working frequency optimization method, system, storage medium and application Download PDFInfo
- Publication number
- CN113126038B CN113126038B CN202110371066.1A CN202110371066A CN113126038B CN 113126038 B CN113126038 B CN 113126038B CN 202110371066 A CN202110371066 A CN 202110371066A CN 113126038 B CN113126038 B CN 113126038B
- Authority
- CN
- China
- Prior art keywords
- frequency
- spectrum
- window
- data
- algorithm
- 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
- 238000000034 method Methods 0.000 title claims abstract description 38
- 238000005457 optimization Methods 0.000 title claims abstract description 16
- 238000001228 spectrum Methods 0.000 claims abstract description 109
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 37
- 238000012545 processing Methods 0.000 claims abstract description 26
- 238000012549 training Methods 0.000 claims abstract description 19
- 238000011156 evaluation Methods 0.000 claims abstract description 14
- 230000008569 process Effects 0.000 claims abstract description 8
- 230000006870 function Effects 0.000 claims description 26
- 238000010586 diagram Methods 0.000 claims description 23
- 238000012544 monitoring process Methods 0.000 claims description 19
- 238000004458 analytical method Methods 0.000 claims description 17
- 230000015654 memory Effects 0.000 claims description 16
- 239000013598 vector Substances 0.000 claims description 11
- 230000007787 long-term memory Effects 0.000 claims description 9
- 230000004913 activation Effects 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 210000002569 neuron Anatomy 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 210000004027 cell Anatomy 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 239000000758 substrate Substances 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 2
- 230000006872 improvement Effects 0.000 abstract description 3
- 238000013519 translation Methods 0.000 description 7
- 230000000694 effects Effects 0.000 description 6
- 230000006403 short-term memory Effects 0.000 description 6
- 230000007547 defect Effects 0.000 description 3
- 238000003491 array Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000008034 disappearance Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000008713 feedback mechanism Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000035939 shock Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/36—Means for anti-jamming, e.g. ECCM, i.e. electronic counter-counter measures
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/88—Radar or analogous systems specially adapted for specific applications
-
- 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)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention belongs to the technical field of radars, and discloses a high-frequency ground wave radar working frequency optimization method, a system, a storage medium and application thereof, wherein the frequency spectrum data is acquired, and a two-dimensional OS algorithm and a double threshold algorithm are adopted to process the frequency spectrum data so as to obtain frequency spectrum information of anti-impact interference processing; constructing an LSTM network, and performing model training and prediction deviation evaluation; a conic trapezoidal window is selected, the window function is slid on the whole frequency spectrum to extract the average power spectrum and fluctuation information in the window, the total score of the sliding window is calculated, and then the optimal working frequency is selected according to the score. The improvement of the frequency spectrum prediction scheme in the invention enhances the instantaneity and accuracy of frequency selection; the LSTM model predicts the time sequence, trains and learns the frequency spectrum information at the current moment instead of simple data frequency shift, so that the prediction is closer to a true value, the reliability is greatly enhanced, sliding window frequency selection is carried out through a conic trapezoid window, and an interference frequency band can be effectively avoided.
Description
Technical Field
The invention belongs to the technical field of radars, and particularly relates to a method, a system, a storage medium and application for optimizing working frequency of a high-frequency ground wave radar.
Background
At present: the high-frequency ground wave radar has the advantages of over-the-horizon, all weather and the like for sea monitoring due to low cost and easy maintenance, but the working frequency band (3-30 MHz) electromagnetic environment is complex, a large amount of short wave communication service, radio station interference, impact interference and noise exist, and the working performance of the high-frequency ground wave radar is affected. Searching for "silent frequencies" suitable for its operation can effectively improve radar performance, so a method is sought that can recommend a better operating frequency for high frequency ground wave radar. Long Short-Term Memory network LSTM (Long Short-Term Memory) is a type of feedback neural network. The cyclic neural network (RNN) has preliminary memory capacity due to the introduction of a feedback mechanism, but has limited length of information which can be processed due to the need of preserving the information of all states, and can solve the problems of gradient disappearance and gradient explosion when the update weight is reversely propagated, so that the long-term memory capacity is insufficient. And a long-short-term memory network (LSTM) with input gates, output gates and forgetting gates is introduced, memory units are introduced, a path for long-time continuous flow of gradients is generated, and the weight of the self-loop is updated in each iteration. The design can avoid information loss caused by transverse depth, and also solves the problem of gradient disappearance which is easy to generate when the RNN model updates weight. The forgetting gate controls the storage of information by deleting part of history information in the memory unit; the input gate screens the output state at the current time and the last time, and adjusts the useful information entering the LSTM cell. The working mode ensures that an algorithm based on a long-term and short-term memory network model is good at processing long-sequence data, and can realize time sequence prediction.
In practical engineering application, most of the adopted frequency spectrum prediction schemes are current translation prediction models. And this model translates the current time data as the next time spectral data. The working environment of the high-frequency ground wave radar is complex, and although the natural electric interference is eliminated through an anti-impact interference algorithm, a large amount of noise and interference still exist, and the adoption of the scheme directly shifts the frequency spectrum to ignore the complex and changeable electromagnetic environment, so that the effect of real-time monitoring cannot be achieved, and the accuracy of frequency selection is affected.
The 'multiple threshold-average power scheme' actually adopted by the sliding window frequency selection scheme is to perform rough measurement frequency selection by selecting three thresholds which are gradually reduced, and the finally selected frequency band is positioned in a range with relatively low interference. The accuracy is improved to a certain extent by adopting three thresholds, but the proper threshold parameters can be obtained only by multiple experimental tests and accumulation, and the method lacks certain flexibility in complex and changeable electromagnetic environments.
Through the above analysis, the problems and defects existing in the prior art are as follows:
(1) The existing spectrum prediction scheme ignores complex and changeable electromagnetic environments, cannot achieve the effect of real-time monitoring, and influences the accuracy of frequency selection.
(2) The existing sliding window frequency selection scheme needs multiple experimental tests to accumulate to acquire proper threshold parameters, and has low flexibility.
The difficulty of solving the problems and the defects is as follows: because the electromagnetic environment is complex and changeable, if the currently monitored frequency spectrum is used for selecting the 'silence frequency' to be applied to the radar, the 'silence frequency' cannot achieve a better effect when the frequency spectrum is changed greatly. A scheme capable of efficiently predicting the frequency spectrum at the next time based on the existing frequency information is necessary. The sliding window frequency selection is scored according to a certain criterion, and the selected criterion has reference significance as a frequency selection element and is little affected by small interference. In summary, a scheme is needed to find a frequency spectrum prediction method that can effectively predict the frequency spectrum at the next time according to the current frequency spectrum information and can screen out the optimal frequency according to a certain frequency selection principle.
The meaning of solving the problems and the defects is as follows: the frequency spectrum is effectively predicted in a complex electromagnetic environment, more accurate data information is provided for searching the optimal working frequency later, and the effect of real-time frequency selection can be approximately achieved. The proper sliding window frequency selection scheme can efficiently utilize the characteristic information of the frequency spectrum to search for the preferred frequency, and the radar performance working on the preferred frequency is improved.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention provides a method, a system, a storage medium and application for optimizing the working frequency of a high-frequency ground wave radar.
The invention is realized in such a way that a high-frequency ground wave radar working frequency optimization method, a system, a storage medium and an application thereof comprise:
step one, acquiring spectrum data: the frequency monitoring subsystem processes the received signals in parallel by utilizing a plurality of receiving channels, respectively carries out Fourier analysis on each channel, then combines the analysis results of the channels to obtain complete spectrum data, and uses the complete spectrum data as spectrum information of shock resistance processing to provide raw data for subsequent processing;
step two, anti-impact interference treatment: the two-dimensional OS algorithm and the double threshold algorithm are adopted to process the frequency spectrum data, so that the frequency spectrum information of the impact interference processing is obtained, the influence of the impact interference on raising the full-frequency band frequency spectrum is eliminated, and the relatively real frequency spectrum information is provided for the frequency spectrum prediction;
step three, predicting a frequency spectrum by a long-term and short-term memory model: and constructing an LSTM network, and performing model training and prediction deviation evaluation. The LSTM gating structure can balance short-term and long-term dependence on a sequence time dimension on a time sequence, has higher prediction precision than a current prediction model, and provides spectrum information with high prediction precision for a sliding window frequency selection module;
step four, sliding window frequency selection of the average power spectrum-variance frequency selection scheme: a conic trapezoidal window is selected, the window function is slid on the whole frequency spectrum to extract the average power spectrum and fluctuation information in the window, the total score of the sliding window is calculated, and then the optimal working frequency is selected according to the score. Considering the frequency band and electromagnetic information on two sides of the frequency band, the optimal frequency can be provided for radar selection by avoiding the interference frequency band under the condition of variable electromagnetic environment.
Further, in the first step, each receiving channel has a different carrier frequency, so as to cover the entire frequency band.
Further, in the second step, the anti-impact interference treatment specifically includes:
(1) The method comprises the steps of adopting a two-dimensional OS algorithm, firstly sorting each point according to amplitude of each time batch data in a frequency dimension to find a noise substrate, and then adopting the OS algorithm to find the noise substrate of a time-frequency diagram in the time dimension;
(2) And (3) adopting a double-threshold algorithm, setting an amplitude threshold and a pulse width threshold to judge interference pulses, eliminating frequency monitoring data subjected to impact interference or replacing the frequency monitoring data with adjacent non-interference batch data, and then splicing and integrating the rest effective spectrum data to obtain the spectrum information of impact interference processing.
Further, step three, the long-term memory model prediction spectrum includes:
(1) Data preparation, namely reading spectrum data according to batches, dividing the spectrum data into a training set and a testing set, and normalizing the data;
(2) Constructing an LSTM network, setting the node number of each layer, and defaulting an activation function to a sigmoid function and a tanh function;
(3) Initializing network parameters, and setting forgetting rate, learning rate, neuron number and maximum iteration times;
(4) Model training and evaluation indexes, carrying out model training by adopting an Adam algorithm, and measuring the prediction deviation of the model by using root mean square error as the evaluation index. Root mean square error calculation formula:
wherein the method comprises the steps of
Representing the predicted value of the model, x i Representing the true value of the sequence and n representing the number of predicted values.
Further, the calculation formula of the long-term and short-term memory model is as follows:
i t ＝sigmoid(W i [x t ,h t-1 ]+b i )；
f t ＝sigmoid(W f [x t ,h t-1 ]+b f )；
o t ＝sigmoid(W o [x t ,h t-1 ]+b o )；
h t ＝o t ·tanh(c t )；
in each time step t, x t For inputting vectors c t Is a cell state vector, h t Is according to c t An output hidden state vector, where W * Representing a weight matrix, b * Is the bias vector, e { i, f, o, c }; the sigmoid function is used as input gate i t Forgetting door f t And an output gate o t Is in the range of [0,1 ]]The ratio of the output information of each gate is controlled;
and h t The tanh function is used as the activation function.
Further, the average power spectrum-variance frequency selection scheme sliding window frequency selection specifically comprises:
firstly, determining the width of a conic trapezoidal window according to the bandwidth of a radar receiving channel and the bandwidth of a demodulation signal, sliding the designed sliding window on the whole frequency spectrum data, scoring the sliding window by a scoring formula, and selecting an optimal frequency point in a working frequency band.
Further, the width and the shape of the sliding window are selected to be matched with the bandwidth of a radar receiving channel and the bandwidth of a demodulation signal, and the average power and the variance are used as frequency selection elements to represent the average amplitude and the deviation degree of information in a certain accumulated time window.
Further, the scoring formula is:
x 1 ＝n F1 x F1 +n F2 x F2 ；
m is in F1 ，m F2 Taking m as a proportionality coefficient F1 ＝20，m F2 ＝5；C MIN ，V MIN Represents a least mean square value; faverage is at x F1 The average value of the sliding window power is expressed in the formula and is shown as x F2 The variance value expressed as a sliding window in the formula; x is x 1 Is the total score of the sliding window;x F1 mean power score for sliding window; x is x F2 A sliding window variance score; n is n F1 Taking 0.7 as a proportionality coefficient of the average power score of the sliding window; n is n F2 Taking 0.3 as the proportionality coefficient of the sliding window variance value.
Another object of the present invention is to provide a computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
acquiring spectrum data: the frequency monitoring subsystem processes the received signals in parallel by utilizing a plurality of receiving channels, respectively carries out Fourier analysis on each channel, and combines analysis results of the channels to obtain complete frequency spectrum data;
impact interference resistance treatment: processing the frequency spectrum data by adopting a two-dimensional OS algorithm and a double threshold algorithm to obtain frequency spectrum information of anti-impact interference processing;
long-term memory model prediction spectrum: constructing an LSTM network, and carrying out model training and prediction deviation evaluation;
average power spectrum-variance frequency selection scheme sliding window frequency selection: a conic trapezoidal window is selected, the window function is slid on the whole frequency spectrum to extract the average power spectrum and fluctuation information in the window, the total score of the sliding window is calculated, and then the optimal working frequency is selected according to the score.
Another object of the present invention is to provide a high-frequency ground wave radar operation frequency optimization system for the high-frequency ground wave radar operation frequency optimization method, the high-frequency ground wave radar operation frequency optimization system comprising:
the spectrum data acquisition module is used for processing the received signals in parallel by utilizing multiple paths of receiving channels, performing Fourier analysis on each channel respectively, and then combining analysis results of the channels to obtain complete spectrum data;
the anti-impact interference processing module is used for processing the frequency spectrum data to obtain frequency spectrum information of anti-impact interference processing;
the LSTM network model prediction spectrum module is used for constructing an LSTM network, and performing model training and prediction deviation evaluation;
and the sliding window frequency selection module is used for selecting a proper window function, sliding the window function on the whole frequency spectrum to extract average power spectrum and fluctuation information in the window, calculating the total score of the sliding window, and further selecting the optimal working frequency according to the score.
Another object of the present invention is to provide a high-frequency ground wave radar equipped with the high-frequency ground wave radar operating frequency optimization system.
By combining all the technical schemes, the invention has the advantages and positive effects that: the improvement of the frequency spectrum prediction scheme in the invention enhances the instantaneity and accuracy of frequency selection. The time sequence is predicted through the LSTM model, and the frequency spectrum information at the current moment is trained and learned instead of simple data frequency shift. Comparing the LSTM prediction on the full frequency band with the RMSE of the current translation prediction model, the method finds that compared with the traditional prediction method, the prediction is closer to a true value by performing spectrum prediction based on the LSTM, and the reliability is greatly enhanced.
According to the quadratic curve trapezoidal window disclosed by the invention, sliding window frequency selection is carried out according to an average power spectrum-variance frequency selection scheme, average power and variance are selected as index evaluation, spectrum information is comprehensively reflected, and an interference frequency band can be effectively avoided by the quadratic curve trapezoidal window.
Comparing the frequency selection simulation results of the rectangular window, the trapezoidal window and the conic trapezoidal window, finding that when the rectangular window is scored, frequency peaks exist at the boundaries of two sides, the electromagnetic environment is changeable, and the frequency peaks possibly influence the detection power of the radar. The trapezoidal window and the conic trapezoidal window can not only reflect data information in a frequency band, but also give consideration to electromagnetic environment information on two sides. Compared with a trapezoidal window, the quadratic curve trapezoidal window is adopted for frequency selection, so that the small interference judgment and avoidance effects are more excellent.
The invention provides a root mean square error comparison graph of the LSTM and the current translation prediction result, which verifies the superiority of the LSTM model prediction scheme; in addition, the result of the sliding window frequency selection of the rectangular window and the conic trapezoid window is compared, and the conic trapezoid window is found to consider information on two sides of a frequency band, so that the adaptability is higher in a complex electromagnetic environment. Errors in the schematic diagram of the long-short-term memory model are corrected.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following description will briefly explain the drawings needed in the embodiments of the present application, and it is obvious that the drawings described below are only some embodiments of the present application, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a preferred method for operating a high-frequency ground wave radar according to an embodiment of the present invention.
FIG. 2 is a schematic diagram of a long-short term memory model provided by an embodiment of the present invention.
Fig. 3 is a schematic diagram of sliding window frequency selection according to an embodiment of the present invention.
Fig. 4 is a diagram of acquired spectrum data containing impulse interference according to an embodiment of the present invention.
Fig. 5 is a spectrum chart of the embodiment of the present invention after impact data is removed by using an impact-interference algorithm.
Fig. 6 is a diagram of a "time-frequency-power spectrum" of obtaining measured data in practical application according to an embodiment of the present invention.
FIG. 7 is a root mean square error plot of the prediction results of the long-short term memory model provided by the embodiment of the invention.
Fig. 8 is a root mean square error diagram of a current translation model prediction result according to an embodiment of the present invention.
FIG. 9 is a root mean square error comparison chart of prediction results of a long-short-term memory model and a current translation model according to an embodiment of the present invention.
Fig. 10 is a functional structure diagram of a conic trapezoid window according to an embodiment of the present invention.
Fig. 11 is a functional structure diagram of a rectangular window according to an embodiment of the present invention.
Fig. 12 is a schematic diagram of rectangular window frequency selection according to an embodiment of the present invention.
Fig. 13 is a schematic diagram of a conic trapezoid window according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the following examples in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
In view of the problems existing in the prior art, the present invention provides a method, a system, a storage medium and an application for optimizing the working frequency of a high-frequency ground wave radar, and the present invention is described in detail below with reference to the accompanying drawings.
As shown in fig. 1, the method for optimizing the working frequency of the high-frequency ground wave radar according to the embodiment of the invention includes:
s101, acquiring spectrum data: the frequency monitoring subsystem processes the received signals in parallel by utilizing a plurality of receiving channels, respectively carries out Fourier analysis on each channel, and then combines the analysis results of the channels to obtain complete frequency spectrum data;
s102, impact interference resistance processing: processing the frequency spectrum data by adopting a two-dimensional OS algorithm and a double threshold algorithm to obtain frequency spectrum information of anti-impact interference processing;
s103, predicting a frequency spectrum by a long-term and short-term memory model: constructing an LSTM network, and carrying out model training and prediction deviation evaluation;
s104, sliding window frequency selection of an average power spectrum-variance frequency selection scheme: sliding a conic trapezoid window on the whole frequency spectrum, extracting average power spectrum and fluctuation information in the window, and selecting the optimal working frequency according to a scoring scheme.
Those skilled in the art may implement other steps in the method for optimizing the operating frequency of the high-frequency ground wave radar according to the present invention, and the method for optimizing the operating frequency of the high-frequency ground wave radar according to the present invention shown in fig. 1 is merely a specific embodiment.
The technical scheme of the invention is further described below with reference to the accompanying drawings.
1. As shown in fig. 2, a schematic frame diagram of a long-short-term memory model (LSTM) is introduced into an input gate, an output gate and a forget gate to control the memory and forget of time sequence information. In each time step t, x t For inputting vectors c t Is directed to the cell stateQuantity, h t Is according to c t An output hidden state vector, where W * Representing a weight matrix, b * Is the bias vector, e { i, f, o, c }. The sigmoid function is used as input gate i t Forgetting door f t And an output gate o t Is in the range of [0,1 ]]The ratio of the output information of each gate is controlled;
and h t As the activation function, a hyperbolic tangent tanh function is generally used. The following formula is the calculation formula of LSTM:
(1)i t ＝sigmoid(W i [x t ,h t-1 ]+b i )；
(2)f t ＝sigmoid(W f [x t ,h t-1 ]+b f )；
(3)o t ＝sigmoid(W o [x t ,h t-1 ]+b o )；
(4)
(5)
(6)h t ＝o t ·tanh(c t )；
the invention sets the node number, forgetting rate, learning rate, neuron number and maximum iteration times of each layer by establishing the LSTM network. And (3) inputting one part of the spectrum information subjected to the impact interference resistance treatment into the LSTM network as a training set for learning, and predicting the spectrum by using the other part as a prediction set.
2. Average power spectrum-variance frequency selection scheme frequency selection
Fig. 3 is a schematic diagram of sliding window frequency selection, wherein the principle of sliding window frequency selection is to select a proper window function, score according to a certain criterion, and select the optimal working frequency by the score. Selecting the width and shape of the sliding window to match the bandwidth of the radar receiving channel and demodulatingThe bandwidth of the signal uses the average power and variance as frequency-selective elements to represent the average amplitude and deviation of the information in a certain cumulative time window. M is in F1 ，m F2 For the proportionality coefficient, take m F1 ＝20，m F2 ＝5；C MIN ，V MIN Represents a least mean square value; faverage represents the average value of the sliding window power in equation (7) and the variance value of the sliding window in equation (8); x is x 1 Is the total score of the sliding window; x is x F1 Mean power score for sliding window; x is x F2 A sliding window variance score; n is n F1 Taking 0.7 as a proportionality coefficient of the average power score of the sliding window; n is n F2 Taking 0.3 as the proportionality coefficient of the sliding window variance value. The formula is as follows:
(7)
(8)
(9)x 1 ＝n F1 x F1 +n F2 x F2 ；
firstly, determining the width of a conic trapezoidal window according to the bandwidth of a radar receiving channel and the bandwidth of a demodulation signal, sliding the designed sliding window on the whole frequency spectrum data, scoring the sliding window by the scoring formula, and selecting the optimal frequency point in the working frequency band. When the rectangular window is scored, frequency peaks exist at the boundaries of two sides, the electromagnetic environment is changeable, and the frequency peaks possibly influence the detection power of the radar, so that the quadratic curve trapezoidal window is adopted for frequency selection. The method can not only embody the data information in the frequency band, but also give consideration to the electromagnetic environment information on both sides.
The technical effects of the present invention will be described in detail with reference to simulation experiments.
1. Acquiring spectrum data: the frequency monitoring subsystem processes the received signals in parallel by utilizing multiple paths of receiving channels, and each path of receiving channel has different carrier frequencies so as to cover the whole frequency band. And respectively carrying out Fourier analysis on each channel, and then merging analysis results of all the channels to obtain complete spectrum data.
2. Impact interference handling
I. And (3) sorting the data of each time batch according to the amplitude in the frequency dimension by adopting a two-dimensional OS (order statistics) algorithm, namely a two-dimensional ordered statistic algorithm, taking 1/4 of the sorted samples as an estimation threshold of a noise base, and using an OS algorithm base for the obtained base sequence to obtain the noise base of the time-frequency diagram.
II, adopting a double threshold algorithm, setting an amplitude threshold and a pulse width threshold to judge interference pulses, eliminating frequency monitoring data subjected to impact interference or replacing the frequency monitoring data with adjacent non-interference batch data, and then splicing and integrating the rest effective spectrum data to obtain the spectrum information of impact interference processing.
Fig. 4 and 5 are spectra before and after the impact interference.
3. Prediction spectrum of long-short-term memory model
I. Data preparation. And (3) reading the spectrum data according to batches, dividing the spectrum data into a training set and a testing set, and normalizing the data.
And II, constructing an LSTM network. Setting the node number of each layer, and defaulting the activation function to a sigmoid function and a tanh function.
Initializing network parameters. Setting forgetting rate, learning rate, neuron number and maximum iteration number.
IV, model training and evaluating indexes. Model training is performed by adopting an Adam algorithm, and the prediction deviation of the model is measured by using root mean square error (Root Mean Square Error, RMSE) as an evaluation index in the experiment.
The current recording frequency range is 4.3MHz to 4.9MHz, the frequency interval is 0.5kHz, and the total frequency ranges are 1200. The time average interval was set to 5 minutes, and a total of 700 time batch points were set for each frequency point, with 90% of each 700 time batch being the training set and 10% being the test set, and the results of the 70 points were predicted. The forgetting rate is set to 0.1, the learning rate is set to 0.005, the number of neurons is set to 200, and the maximum number of iterations is 250. Fig. 6 is a diagram of the read spectrum information, fig. 7 is a diagram of the rms error value of the LSTM prediction model, fig. 8 is a diagram of the rms error value of the current translation prediction model, and fig. 9 is a diagram of the rms error comparison of the LSTM and the current translation prediction result.
4. Sliding window frequency selection for average power spectrum-variance frequency selection scheme
A conic trapezoid window (figure 10) is selected, the window function is slid on the whole frequency spectrum to extract average power spectrum and fluctuation information in the window, the total score of the sliding window is calculated, and then the optimal working frequency is selected according to the score. Fig. 11 is a rectangular window structure diagram, and fig. 12 and 13 are schematic diagrams of frequency selection of rectangular windows and conic trapezoidal windows respectively.
It should be noted that the embodiments of the present invention can be realized in hardware, software, or a combination of software and hardware. The hardware portion may be implemented using dedicated logic; the software portions may be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or special purpose design hardware. Those of ordinary skill in the art will appreciate that the apparatus and methods described above may be implemented using computer executable instructions and/or embodied in processor control code, such as provided on a carrier medium such as a magnetic disk, CD or DVD-ROM, a programmable memory such as read only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The device of the present invention and its modules may be implemented by hardware circuitry, such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., as well as software executed by various types of processors, or by a combination of the above hardware circuitry and software, such as firmware.
The foregoing is merely illustrative of specific embodiments of the present invention, and the scope of the invention is not limited thereto, but any modifications, equivalents, improvements and alternatives falling within the spirit and principles of the present invention will be apparent to those skilled in the art within the scope of the present invention.
Claims (9)
1. A high-frequency ground wave radar operation frequency optimization method, characterized in that the high-frequency ground wave radar operation frequency optimization method comprises:
acquiring spectrum data: the frequency monitoring subsystem processes the received signals in parallel by utilizing a plurality of receiving channels, respectively carries out Fourier analysis on each channel, and combines analysis results of the channels to obtain complete frequency spectrum data;
impact interference resistance treatment: processing the frequency spectrum data by adopting a two-dimensional OS algorithm and a double threshold algorithm to obtain frequency spectrum information of anti-impact interference processing;
long-term memory model prediction spectrum: constructing an LSTM network, and carrying out model training and prediction deviation evaluation;
average power spectrum-variance frequency selection scheme sliding window frequency selection: sliding a conic trapezoid window on the whole frequency spectrum, extracting average power spectrum and fluctuation information in the window, and selecting the optimal working frequency according to a scoring scheme;
the anti-impact interference treatment specifically comprises the following steps:
(1) A two-dimensional OS algorithm, namely a two-dimensional ordered statistic algorithm, is adopted, data of each time batch are ordered according to amplitude in a frequency dimension, 1/4 of the ordered samples are taken as an estimation threshold of a noise base, and the obtained base sequence is subjected to the OS algorithm base, so that the noise base of a time-frequency diagram is obtained;
(2) And (3) adopting a double-threshold algorithm, setting an amplitude threshold and a pulse width threshold to judge interference pulses, eliminating frequency monitoring data subjected to impact interference or replacing the frequency monitoring data with adjacent non-interference batch data, and splicing and integrating the rest effective spectrum data to obtain the spectrum information of impact interference processing.
2. The method of optimizing the operating frequency of a high frequency ground wave radar as recited in claim 1, wherein each receiving channel has a different carrier frequency covering the entire frequency band.
3. The high frequency ground wave radar operating frequency optimization method of claim 1, wherein the long-short term memory model prediction spectrum comprises:
(1) Data preparation, namely reading spectrum data according to batches, dividing the spectrum data into a training set and a testing set, and normalizing the data;
(2) Constructing an LSTM network, setting the node number of each layer, and defaulting an activation function to a sigmoid function and a tanh function;
(3) Initializing network parameters, and setting forgetting rate, learning rate, neuron number and maximum iteration times;
(4) Model training and evaluation indexes, carrying out model training by adopting an Adam algorithm, using root mean square error as an evaluation index, measuring the prediction deviation of the model, and calculating a formula of the root mean square error:
wherein the method comprises the steps of
Representing the predicted value of the model, x i Representing the true value of the sequence and n representing the number of predicted values.
4. The method for optimizing the operating frequency of a high-frequency ground wave radar according to claim 1, wherein the calculation formula of the long-short-term memory model is as follows:
i t ＝sigmoid(W i [x t ,h t-1 ]+b i )；
f t ＝sigmoid(W f [x t ,h t-1 ]+b f )；
o t ＝sigmoid(W o [x t ,h t-1 ]+b o )；
h t ＝o t ·tanh(c t )；
in each time step t, x t For inputting vectors c t Is a cell state vector, h t Is according to c t An output hidden state vector, where W * Representing a weight matrix, b * Is the bias vector, e { i, f, o, c }; the sigmoid function is used as input gate i t Forgetting door f t And an output gate o t Is in the range of [0,1 ]]The ratio of the output information of each gate is controlled;
and h t The tanh function is used as the activation function.
5. The method for optimizing the operating frequency of the high-frequency ground wave radar according to claim 1, wherein the average power spectrum-variance frequency selection scheme sliding window frequency selection specifically comprises: firstly, determining the width of a conic trapezoidal window according to the bandwidth of a radar receiving channel and the bandwidth of a demodulation signal, sliding the designed sliding window on the whole frequency spectrum data, scoring the sliding window by a scoring formula, and selecting an optimal frequency point in a working frequency band;
the width and the shape of the sliding window are selected to be matched with the bandwidth of a radar receiving channel and the bandwidth of a demodulation signal, and the average power and the variance are used as frequency selection elements to represent the average amplitude and the deviation degree of information in a certain accumulated time window.
6. The high frequency ground wave radar operating frequency optimization method of claim 5, wherein the scoring formula is:
x 1 ＝n F1 x F1 +n F2 x F2 ；
m is in F1 ，m F2 Taking m as a proportionality coefficient F1 ＝20，m F2 ＝5；C MIN ，V MIN Represents a least mean square value; faverage is at x F1 The average value of the sliding window power is expressed in the formula and is shown as x F2 The variance value expressed as a sliding window in the formula; x is x 1 Is the total score of the sliding window; x is x F1 Mean power score for sliding window; x is x F2 A sliding window variance score; n is n F1 Taking 0.7 as a proportionality coefficient of the average power score of the sliding window; n is n F2 Taking 0.3 as the proportionality coefficient of the sliding window variance value.
7. A computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
acquiring spectrum data: the frequency monitoring subsystem processes the received signals in parallel by utilizing a plurality of receiving channels, respectively carries out Fourier analysis on each channel, and combines analysis results of the channels to obtain complete frequency spectrum data;
impact interference resistance treatment: calculating a noise substrate by adopting a two-dimensional OS algorithm, judging and removing impact pulses by adopting a double threshold algorithm, and obtaining the frequency spectrum information of impact interference processing;
long-term memory model prediction spectrum: constructing an LSTM network, and carrying out model training and prediction deviation evaluation;
average power spectrum-variance frequency selection scheme sliding window frequency selection: sliding a conic trapezoid window on the whole frequency spectrum to extract average power spectrum and fluctuation information in the window, calculating total score of the sliding window, and selecting the optimal working frequency according to the score;
the anti-impact interference treatment specifically comprises the following steps:
(1) A two-dimensional OS algorithm, namely a two-dimensional ordered statistic algorithm, is adopted, data of each time batch are ordered according to amplitude in a frequency dimension, 1/4 of the ordered samples are taken as an estimation threshold of a noise base, and the obtained base sequence is subjected to the OS algorithm base, so that the noise base of a time-frequency diagram is obtained;
(2) And (3) adopting a double-threshold algorithm, setting an amplitude threshold and a pulse width threshold to judge interference pulses, eliminating frequency monitoring data subjected to impact interference or replacing the frequency monitoring data with adjacent non-interference batch data, and splicing and integrating the rest effective spectrum data to obtain the spectrum information of impact interference processing.
8. A high-frequency ground wave radar operation frequency optimization system for use in the high-frequency ground wave radar operation frequency optimization method of any one of claims 1 to 6, characterized in that the high-frequency ground wave radar operation frequency optimization system comprises:
the spectrum data acquisition module is used for processing the received signals in parallel by utilizing multiple paths of receiving channels, performing Fourier analysis on each channel respectively, and then combining analysis results of the channels to obtain complete spectrum data;
the anti-impact interference processing module is used for processing the frequency spectrum data to obtain frequency spectrum information of anti-impact interference processing;
the LSTM network model prediction spectrum module is used for constructing an LSTM network, and performing model training and prediction deviation evaluation;
the sliding window frequency selection module is used for selecting a proper window function, sliding the window function on the whole frequency spectrum to extract average power spectrum and fluctuation information in the window, calculating total score of the sliding window, and further selecting the optimal working frequency according to the score;
the anti-impact interference treatment specifically comprises the following steps:
(1) A two-dimensional OS algorithm, namely a two-dimensional ordered statistic algorithm, is adopted, data of each time batch are ordered according to amplitude in a frequency dimension, 1/4 of the ordered samples are taken as an estimation threshold of a noise base, and the obtained base sequence is subjected to the OS algorithm base, so that the noise base of a time-frequency diagram is obtained;
(2) And (3) adopting a double-threshold algorithm, setting an amplitude threshold and a pulse width threshold to judge interference pulses, eliminating frequency monitoring data subjected to impact interference or replacing the frequency monitoring data with adjacent non-interference batch data, and splicing and integrating the rest effective spectrum data to obtain the spectrum information of impact interference processing.
9. A high-frequency ground wave radar, wherein the high-frequency ground wave radar is equipped with the high-frequency ground wave radar operation frequency optimization system according to claim 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110371066.1A CN113126038B (en) | 2021-04-07 | 2021-04-07 | High-frequency ground wave radar working frequency optimization method, system, storage medium and application |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110371066.1A CN113126038B (en) | 2021-04-07 | 2021-04-07 | High-frequency ground wave radar working frequency optimization method, system, storage medium and application |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113126038A CN113126038A (en) | 2021-07-16 |
| CN113126038B true CN113126038B (en) | 2023-05-30 |
Family
ID=76775082
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110371066.1A Active CN113126038B (en) | 2021-04-07 | 2021-04-07 | High-frequency ground wave radar working frequency optimization method, system, storage medium and application |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113126038B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114325625B (en) * | 2021-11-23 | 2024-03-12 | 哈尔滨工业大学（威海） | Ocean-ionosphere chaotic dynamics characteristic extraction method, device, medium and application |
| CN114740432B (en) * | 2022-04-22 | 2023-04-14 | 哈尔滨工业大学 | Composite multi-factor high-frequency ground wave radar comprehensive frequency selection method |
| CN115356694B (en) * | 2022-08-26 | 2023-08-22 | 哈尔滨工业大学（威海） | High-frequency ground wave radar anti-impact interference method, system, computer equipment and application |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1819496A (en) * | 2006-03-09 | 2006-08-16 | 武汉大学 | Method for realizing frequency monitoring by high-frequency radar receiver |
| CN102879766A (en) * | 2011-07-11 | 2013-01-16 | 哈尔滨工业大学 | Method and apparatus for detecting and tracking faint target of high frequency ground wave radar |
| CN102879774A (en) * | 2011-07-11 | 2013-01-16 | 哈尔滨工业大学 | Method and apparatus for synthesizing short flight paths |
| CN110045337A (en) * | 2019-05-10 | 2019-07-23 | 武汉大学 | High-frequency ground wave radar radio frequency interference suppressing method based on tensor subspace projection |
| CN112198484A (en) * | 2020-09-24 | 2021-01-08 | 清华大学 | Anti-interference-oriented radar frequency agility method |
-
2021
- 2021-04-07 CN CN202110371066.1A patent/CN113126038B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1819496A (en) * | 2006-03-09 | 2006-08-16 | 武汉大学 | Method for realizing frequency monitoring by high-frequency radar receiver |
| CN102879766A (en) * | 2011-07-11 | 2013-01-16 | 哈尔滨工业大学 | Method and apparatus for detecting and tracking faint target of high frequency ground wave radar |
| CN102879774A (en) * | 2011-07-11 | 2013-01-16 | 哈尔滨工业大学 | Method and apparatus for synthesizing short flight paths |
| CN110045337A (en) * | 2019-05-10 | 2019-07-23 | 武汉大学 | High-frequency ground wave radar radio frequency interference suppressing method based on tensor subspace projection |
| CN112198484A (en) * | 2020-09-24 | 2021-01-08 | 清华大学 | Anti-interference-oriented radar frequency agility method |
Non-Patent Citations (1)
| Title |
|---|
| Research on the integrated working mode based on positive and negative frequency modulation in radar communication integration;Quanrui Z;《nternational Workshop on Advances in Communications and Computing for Internet-of- Things.2019》;408-414 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113126038A (en) | 2021-07-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113126038B (en) | High-frequency ground wave radar working frequency optimization method, system, storage medium and application | |
| CN110133599B (en) | Intelligent radar radiation source signal classification method based on long-time and short-time memory model | |
| CN110222748B (en) | OFDM radar signal identification method based on 1D-CNN multi-domain feature fusion | |
| CN113687433B (en) | Bi-LSTM-based magnetotelluric signal denoising method and system | |
| CN109065176B (en) | Blood glucose prediction method, device, terminal and storage medium | |
| CN115758212A (en) | Mechanical equipment fault diagnosis method based on parallel network and transfer learning | |
| CN112580539A (en) | Long-term drift suppression method for electronic nose signals based on PSVM-LSTM | |
| CN114417942B (en) | Clutter recognition method, system, device and medium | |
| CN113791389B (en) | Method and system for identifying unknown radar radiation source individuals | |
| Zheng et al. | Reliability analysis of multi-stage degradation with stage-varying noises based on the nonlinear Wiener process | |
| CN117610428A (en) | New energy automobile lithium battery working condition parameter prediction method, device and medium based on fuzzy time sequence | |
| CN117034782A (en) | Receiver fault diagnosis method based on Weibull distribution and long-short-term memory network | |
| CN117991178A (en) | High-precision array direction finding method and system based on data driving | |
| CN117574835A (en) | Method, equipment and storage medium for simulating Nand false LLR distribution | |
| CN115062526A (en) | Deep learning-based three-dimensional ionosphere electron concentration distribution model training method | |
| CN116090352A (en) | Full waveform inversion method based on gate cycle unit and attention mechanism | |
| CN113687143B (en) | Fitting method of electromagnetic wave signal amplitude attenuation and propagation distance relation curve | |
| CN113095215B (en) | Solar radio filtering method and system based on improved LSTM network | |
| CN115629413A (en) | Physical and data driving-based receiving function inversion method | |
| CN111431645B (en) | Spectrum sensing method based on small sample training neural network | |
| CN111160419B (en) | Deep learning-based electronic transformer data classification prediction method and device | |
| Dong et al. | Multi-Classification of Ultra-Wideband Non-Line-of-Sight Ranging Signals using Deep Learning | |
| CN117637063B (en) | Water quality measurement method, device, equipment and storage medium | |
| CN115659162B (en) | Method, system and equipment for extracting intra-pulse characteristics of radar radiation source signals | |
| CN115903023A (en) | Hot dry rock microseism event detection method based on residual error network |
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 |