KR20230086322A - Signal processing apparatus for weather radar with noise filtering function using environmental information - Google Patents

Abstract

The present invention relates to a signal processing apparatus for a weather radar that provides a noise filtering function using environmental information, and more particularly, post-processing obtained by post-processing an IF signal provided from a weather radar based on environmental characteristics in which the weather radar is located. An image similar to the image can be generated without post-processing. Supports a filtering function for the noise of the IF signal based on learning by learning environmental information on the cause of noise around the weather radar together with the noise filtering result for the IF signal. It relates to a signal processing device for a weather radar that provides a noise filtering function using environmental information of The present invention is a process of removing noise caused by environmental characteristics from an IF signal through optimal filtering conditions without post-processing a weather image by reflecting the environmental characteristics of a region where a weather radar is deployed by a user, and through the corresponding post-processing It supports the acquisition of a meteorological image that is as similar as possible to the generated image, thereby increasing the user's convenience in operating the weather radar, and has the effect of supporting the acquisition of a highly reliable meteorological image with non-weather elements removed through noise removal. there is.

Description

Signal processing apparatus for weather radar with noise filtering function using environmental information}

The present invention relates to a signal processing apparatus for a weather radar that provides a noise filtering function using environmental information, and more particularly, post-processing obtained by post-processing an IF signal provided from a weather radar based on environmental characteristics in which the weather radar is located. An image similar to the image can be generated without post-processing. Supports a filtering function for the noise of the IF signal based on learning by learning environmental information on the cause of noise around the weather radar together with the noise filtering result for the IF signal. It relates to a signal processing device for a weather radar that provides a noise filtering function using environmental information of

Meteorological radar transmits radio waves (electromagnetic waves or pulses or radar beams) and is reflected by rain or snow particles (meteoroids), which is a target, and measures the average reflection intensity of the target in the sampling volume and the received power and reception time of the returned signal. A device for measuring position.

The weather radar can observe the three-dimensional (volume) state of the atmosphere in the form of polar coordination by observing the antenna by changing the azimuth angle of 0 to 360 degrees and the altitude angle of 0 to 90 degrees.

In the process of receiving radio waves after transmitting radio waves, the radio waves are reflected on the target, radio waves reflected on the target, not meteorological factors such as snow, rain, or fog, are received or airplanes transmitting radio waves near the weather radar External radio waves transmitted by other radar equipment are received by the weather radar, and as a result, in addition to normal observation echoes generated based on radio waves reflected on targets related to weather such as precipitation, abnormal echoes due to radio wave interference are generated. It is included as noise in the IF signal provided by the weather radar.

Accordingly, it is required to filter noise included in the IF signal in a signal processing apparatus that is connected to a weather radar and generates an observation image by processing an IF signal provided by the weather radar.

However, the existing signal processing device is optimized only for the environmental characteristics set by the manufacturer of the signal processing device, and when operating a weather radar in a specific area having environmental characteristics different from the corresponding environmental characteristics, noise generated only in the environment corresponding to the specific area and it is difficult for the user to modify the filtering function of the signal processing device because it is difficult for the user to understand the noise filtering conditions of the non-manufacturer signal processing device, so that the image generated based on the IF signal provided by the weather radar For this, the user has to manually remove the noise through post-processing, which is inconvenient.

Therefore, there is a demand for a method for optimizing the filtering function of a signal processing device that performs signal processing on an IF signal of a weather radar in consideration of environmental characteristics in which the weather radar is located.

Korean Patent Registration No. 10-0922130

In the present invention, while constantly changing conditions for filtering noise included in an IF signal transmitted by a weather radar, an image generated through filtering is an image obtained by post-processing the IF signal in consideration of characteristics of an environment in which the weather radar is deployed. Noise according to the characteristics of the environment in which the weather radar is deployed based on the learned optimal filtering conditions without the need for a separate post-processing process for the IF signal. Its purpose is to support the generation of an image from which the .

A signal processing apparatus for a weather radar that provides a noise filtering function using environmental information according to an embodiment of the present invention receives an IF signal generated based on an RF signal received from the weather radar, and the IF signal is generated according to a preset filtering condition. A digital image converter for generating an I signal and a Q signal based on the filtering signal generated by filtering the signal; an analog image converter for generating an analog IF signal based on the I and Q signals; A signal processing unit that calculates a correlation coefficient based on the Q signal, calculates a weather moment based on the correlation coefficient, and calculates a calculated value for each of one or more preset weather variables based on the weather moment; A quality control unit that receives and provides environmental information of a region to be processed, and the signal processing unit after receiving the calculated value for each weather variable for the post-processed image generated by post-processing the weather image according to the calculated value for each weather variable according to user input A similarity score corresponding to the filtering condition is calculated by mutual comparison with calculated values for each weather variable received from the environment information received through the quality control unit among a plurality of learning models respectively corresponding to a plurality of different environment types set in advance. Original data generated based on the IF signal corresponding to the filtering condition, correction data generated based on the analog IF signal corresponding to the filtering condition, filtering condition, and the environment A noise learning unit that changes the filtering condition of the digital image conversion unit to the predictive filtering condition after calculating the predictive filtering condition for increasing the similarity score through the learning model while learning the information together with the similarity score corresponding to the filtering condition. can include

As an example related to the present invention, the environment information includes noise environment information and weather information received according to a user input including at least one of topographical information of a region where the weather radar is located and flight information of an airplane flying in the corresponding region. It may be characterized in that it includes at least one of weather information received from the server.

As an example related to the present invention, the digital image conversion unit includes a first filtering unit generating a filtering signal by filtering the IF signal according to the filtering condition, a first amplifier unit amplifying the filtered signal, and the first filtering unit. 1 Analog-to-digital conversion unit that converts the filtering signal amplified by the amplification unit into a digital signal to generate weather data, and a DDC (Digital Down Converter) unit that converts the weather data into I signals and Q signals and provides them to the noise learning unit can include

As an example related to the present invention, the analog image conversion unit includes a DUC (Digital Up Converter) unit that converts the I and Q signals into IF digital signals, and a digital-to-analog conversion unit that converts the IF digital signal into an analog IF signal. A second filtering unit for filtering the analog IF signal and a second amplifier for amplifying the analog IF signal filtered through the second filtering unit and providing the amplified signal to the noise learning unit.

As an example related to the present invention, the signal processing unit includes a correlation coefficient calculation unit for calculating correlation coefficient data including a dual polarization cross-correlation coefficient and a time-domain correlation coefficient based on the I signal and the Q signal, and the correlation coefficient After calculating a threshold variable based on the correlation coefficient data calculated through the calculator, a wake-up moment calculator for calculating a wake-up moment based on the threshold variable, and calculating a calculated value for each weather variable based on the wake-up moment, A weather variable calculation unit provided to the noise learning unit may be included.

As an example related to the present invention, the meteorological variable may include at least one of reflectivity, radial velocity, spectral width, cross-correlation coefficient, and differential phase difference.

As an example related to the present invention, when the predictive filtering condition is set, the digital image conversion unit filters the IF signal received from the weather radar according to the predictive filtering condition to determine the predictive filtering condition based on a filtering signal generated. The analog image conversion unit generates an analog IF signal corresponding to the prediction filtering condition based on the I signal and the Q signal corresponding to the prediction filtering condition, and the signal processing unit generates a corresponding I signal and a Q signal. , calculates a calculated value for each weather variable based on the I signal and Q signal corresponding to the prediction filtering condition, the quality management unit generates and provides environmental information corresponding to the prediction filtering condition, and the noise learning unit generates and provides environmental information corresponding to the prediction filtering condition After receiving the calculated value for each weather variable according to the post-processed image generated by post-processing the weather image according to the calculated value for each weather variable corresponding to the predictive filtering condition according to the user input, the predictive filtering received from the signal processing unit A similarity score is calculated by comparing the calculated value for each weather variable corresponding to the condition, and the prediction filtering condition is applied to the learning model corresponding to the environment type according to the environment information corresponding to the prediction filtering condition received through the quality control unit. Original data generated based on the corresponding IF signal, correction data generated based on the analog IF signal corresponding to the predictive filtering condition, environment information corresponding to the predictive filtering condition, and the predictive filtering condition While learning with similarity scores corresponding to conditions, another predictive filtering condition for increasing the similarity score is calculated through the learning model, and then the existing filtering condition set in the digital image conversion unit is changed as the other predictive filtering condition. can be done with

As an example related to the present invention, the noise learning unit determines that learning of a plurality of learning models is completed when the similarity scores for all IF signals received during a preset time are calculated to be equal to or greater than a preset reference value, and When learning of a plurality of learning models is completed, original data is generated by digitally converting a specific IF signal received after completion of learning of the plurality of learning models, and environmental information corresponding to the specific IF signal is received from the quality control unit. Then, the original data and the environment information corresponding to the specific IF signal are applied as inputs to a specific learning model corresponding to an environment type according to the environment information corresponding to the specific IF signal, and the original data is applied through the specific learning model. After calculating the optimal filtering condition corresponding to the data and environment information, it may be characterized in that it is set in the digital image conversion unit.

As an example related to the present invention, the signal processing apparatus generates and provides a weather image based on a calculated value for each weather variable received from the signal processing unit from the noise learning unit, and generates and provides a weather image according to a user input for the weather image. It may further include a post-processing data collection unit that generates the post-processed image through post-processing and then generates a calculated value for each weather variable according to the post-processed image and provides the calculated value to the noise learning unit.

As an example related to the present invention, the noise learning unit may digitally convert the IF signal to generate the original data, and digitally convert the analog IF signal to generate the correction data.

In the present invention, a weather image according to a calculated value for each weather variable obtained based on a signal obtained by filtering an IF signal received from a weather radar according to a filtering condition and a post-processed image obtained through a post-processing process by a user of the weather image Calculate the similarity of , and learn the correlation between the filtered signal obtained from the IF signal and the IF signal according to the similarity, the filtering condition, the environmental information of the area where the weather radar is located, and the filtering condition, and the filtered signal by learning the learning model. A process of post-processing the weather image by reflecting the environmental characteristics of the area where the weather radar is deployed by the user through the learning model to calculate the optimal filtering conditions in which the obtained weather image is as similar as possible to the post-processed image. It is a process of removing noise caused by environmental characteristics from the IF signal through optimal filtering conditions without any filtering, and it supports to obtain a meteorological image that is as similar as possible to the image generated through the post-processing, thereby increasing the convenience of users operating the weather radar. In addition, there is an effect of supporting to obtain a reliable meteorological image from which non-meteorological elements are removed through noise removal.

In addition, the present invention automatically recognizes filtering conditions having criteria similar to those set by the user for a post-processed image generated through post-processing of a weather image in consideration of the environmental characteristics of the region where the weather radar is placed, based on learning. Through this, the signal processing device can automatically set the optimal filtering conditions for the IF signal of the weather radar connected to the signal processing device by automatically identifying the environmental characteristics adaptively for different weather radars deployed in various regions. It has the effect of significantly increasing the convenience and efficiency of operation.

1 is a block diagram of a signal processing apparatus for a weather radar providing a noise filtering function using environmental information according to an embodiment of the present invention.
2 is an exemplary view of noise appearing in a weather image according to an embodiment of the present invention;
3 is a detailed configuration diagram of a signal processing apparatus for a weather radar providing a noise filtering function using environmental information according to an embodiment of the present invention.
4 is a detailed configuration diagram of a noise learning unit according to an embodiment of the present invention.
5 is an exemplary view of a weather image generated by a signal processing apparatus for a weather radar providing a noise filtering function using environment information according to an embodiment of the present invention;

Hereinafter, detailed embodiments of the present invention will be described with reference to the drawings.

1 is a configuration environment diagram of a signal processing apparatus for a weather radar (hereinafter referred to as a signal processing apparatus) providing a noise filtering function using environmental information according to an embodiment of the present invention.

As shown, the signal processing apparatus 100 may receive a radio frequency (RF) signal, which is a reflected wave received after a radio wave transmitted from a weather radar 1 is reflected by a target, and the signal processing apparatus (100) may generate and provide a weather image based on an Intermediate Frequency (IF) signal obtained from the RF signal.

At this time, as shown in FIG. 2, in the process of the radio wave transmitted from the weather radar 1 being reflected by the target and received by the weather radar 1, it is not a meteorological element such as precipitation, but a terrain or feature. Various types of noise such as reflected non-weather echo, interference signals with external radio waves other than radio waves transmitted by the weather radar 1, and clutter reflected by the ground or sea surface are the IF signal included in

In addition, such noise is the environment of the region where the weather radar 1 is deployed, including the weather change pattern of the region where the weather radar 1 is deployed, the topography of the region, and radio waves transmitted from airplanes flying in the region. It is closely related to the environmental characteristics for , and accordingly, these noises also appear differently between a plurality of IF signals provided from a plurality of weather radars 1 located in different environments.

Accordingly, when a filtering condition for filtering noise from an IF signal is applied to the other weather radar 1 as it is in consideration of the environmental characteristics in which one of the plurality of weather radars 1 is located, the other weather radar 1 is positioned noise is not normally removed because it is not suitable for the environmental characteristics.

Accordingly, conventionally, a separate post-processing process of manually removing noise from a weather image generated based on an IF signal in the signal processing apparatus 100 has to be performed, which is cumbersome.

That is, when the conventional signal processing device 100 uses a filtering condition specialized for a specific environment, it is difficult to remove noise that does not appear in the specific environment in other environments with the corresponding filtering condition. A post-processing process of removing a region visually determined to be noise must be additionally performed.

Therefore, the signal processing apparatus 100 according to an embodiment of the present invention is obtained by filtering the IF signal under various filtering conditions in order to remove noise appearing in the IF signal according to the environmental characteristics of the area where the weather radar 1 is located. The IF signal filtered according to the filtering condition and the original IF signal received from the weather radar 1 and the filtering condition to learn the filtering condition that increases the similarity between the weather image and the post-processed image generated from the corresponding weather image through the post-processing process. And information on the corresponding environmental characteristics is learned in a preset learning model, and when the learning of this learning model is completed, an optimal filtering condition optimized for the environmental characteristics in which the weather radar (1) is located is set, and a separate post-processing process is required. It operates to provide a weather image from which undesirable noise is removed, which will be described with reference to the drawings below.

3 is a detailed configuration diagram of a signal processing device 100 according to an embodiment of the present invention.

As shown, the signal processing apparatus 100 according to an embodiment of the present invention includes an RF receiver 110, a digital image converter 120, an analog image converter 140, a data storage unit 150, a quality It may be configured to include a management unit 170, a post-processing data collection unit 160, and a control unit 130.

The RF receiver 110 may communicate with the weather radar 1, receive an RF signal from the weather radar 1, and generate an IF signal from the RF signal to convert the digital image to the image converter 120. can be provided to

In addition, the digital image converter 120 filters the IF signal received from the weather radar 1 through the RF receiver 110 according to a preset filtering condition, and generates an I signal and an I signal based on a filtering signal. A Q signal may be generated and the I and Q signals may be provided to the control unit 130 .

To this end, the digital image conversion unit 120 includes a first filtering unit 121, a first amplification unit 122, an analog digital converter (ADC) unit 123, and a digital down converter (DDC). converter) unit 124 may be configured.

At this time, the first filtering unit 121 may filter the IF signal according to the filtering condition set in advance to generate a filtering signal that is a filtered IF signal, by the control unit 130 described below. When a prediction filtering condition is set in the first filtering unit 121, a filtering signal may be generated by filtering the IF signal according to the prediction filtering condition.

Also, the first amplifier 122 may amplify the filtered signal and provide the amplified signal to the analog-to-digital converter 123 .

Also, the analog-to-digital conversion unit 123 may generate weather data by converting the filtering signal amplified by the first amplification unit 122 into a digital signal.

In addition, the DDC unit 124 may convert the meteorological data into I signals and Q signals and provide the converted signals to the control unit 130 .

At this time, the DDC unit 124 may provide the I signal and the Q signal to the noise learning unit 132 included in the control unit 130 .

Meanwhile, the analog image conversion unit 140 may generate an analog IF signal based on the I signal and the Q signal, and may provide the analog IF signal to the control unit 130 .

To this end, the analog image conversion unit 140 includes a digital up converter (DUC) unit 144, a digital analog converter (DAC) unit 143, a second filtering unit 142, and a second filtering unit 142. It may be configured to include 2 amplifiers 141.

At this time, the DUC unit 144 may convert the I and Q signals into IF digital signals.

In addition, the digital-to-analog converter 143 may convert the IF digital signal into an analog IF signal.

In addition, the second filtering unit 142 filters the analog IF signal and provides the filtered analog IF signal to the second amplifying unit 141, and the second amplifying unit 141 filters the second filtering unit 142. The analog IF signal filtered through may be amplified and provided to the control unit 130 .

At this time, the second filtering unit 142 may provide the amplified analog IF signal to the noise learning unit 132 included in the control unit 130 .

That is, the analog image conversion unit 140 may generate an analog IF signal corresponding to the filtering signal by inversely transforming the I signal and the Q signal.

Meanwhile, the control unit 130 performs overall control functions of the signal processing device 100, and the control unit 130 may include RAM, ROM, CPU, GPU, and a bus, and may include RAM, ROM, CPU, GPUs and the like may be connected to each other through a bus.

At this time, at least one of the plurality of components constituting the signal processing device 100 may be included in the control unit 130.

The controller 130 may receive the IF signal from the RF receiver 110, digitally convert the original IF signal received from the RF receiver 110 to generate original data, and then store the data. The original data may be stored in the original DB included in the unit 150, and the analog IF signal generated by inversely converting the I signal and the Q signal corresponding to the original IF signal from the analog image conversion unit 140 is received. After digital conversion is performed to generate correction data, the correction data may be stored in a correction DB included in the data storage unit 150 .

At this time, the data storage unit 150 may store various types of information, and the data storage unit 150 may be configured in various forms such as a hard disk drive (HDD) and a solid state drive (SSD).

Also, the quality management unit 170 may communicate with a weather server that manages weather sensors through a communication network, and may provide a user interface for receiving a user input.

Accordingly, the quality control unit 170 may receive meteorological information from the meteorological server, and according to the user input, topographical information of the region where the weather radar 1 is located and the flight of an airplane flying in the region Information (or training information), etc. may be received.

In addition, when the I and Q signals are received by the control unit 130 in association with the control unit 130, the quality control unit 170 determines when the I and Q signals are received by the control unit 130. Weather information corresponding to is provided as environment information, or environmental information including noise environment information, such as topographical information of the region where the weather radar 1 is located or flight information of airplanes flying in the region, is provided as environment information by the control unit 130. ) can be provided.

In this case, the environment information may include at least one of the weather information and noise environment information.

In addition, the quality control unit 170 interlocks with the noise learning unit 132 to be described below, and when the I and Q signals are received by the noise learning unit 132, the quality control unit 170 corresponds to the I and Q signals. Environment information may be generated and provided to the noise learning unit 132 .

In addition, the control unit 130 may include a signal processing unit 131 and a noise learning unit 132.

First, the signal processing unit 131 calculates a correlation coefficient based on the I signal and Q signal provided from the digital image conversion unit 120, and then calculates a wake-up moment based on the correlation coefficient. Based on the moment, one or more preset calculated values for each weather variable may be calculated.

To this end, the signal processor 131 may include a correlation coefficient calculator 1311, a weather moment calculator 1312, and a weather variable calculator 1313.

First, the correlation coefficient calculation unit 1311 may calculate correlation coefficient data including a dual polarized wave cross-correlation coefficient and a time-domain correlation coefficient based on the I signal and the Q signal.

In addition, the wake-up moment calculator 1312 may calculate a threshold variable based on the correlation coefficient data calculated through the correlation coefficient calculator 1311 and then calculate the wake-up moment based on the threshold variable. .

Also, the weather variable calculating unit 1313 may calculate the calculated value for each weather variable based on the weather moment, and provide the calculated value for each weather variable to the noise learning unit 132 .

In this case, the one or more meteorological variables may include reflectivity, radial velocity, spectral width, cross-correlation coefficient, differential phase difference, and the like.

In addition, the noise learning unit 132 receives a post-processed image generated by post-processing an image according to a calculated value for each weather variable according to a user input, and then receives a post-processed image for each weather variable received from the signal processing unit 131. A similarity score corresponding to the filtering condition is calculated by mutual comparison with a weather image according to the calculated value, environmental information corresponding to the filtering condition is received from the quality control unit 170, and the IF signal corresponding to the filtering condition is received. (original IF signal), the analog IF signal, the filtering condition, and the environment information are trained in a preset learning model along with a similarity score corresponding to the filtering condition, and a predictive filtering condition in which the similarity score increases through the learning model. After calculation, the filtering condition of the digital image conversion unit 120 may be changed to the prediction filtering condition.

At this time, the noise learning unit 132 may change an existing filtering condition set in the first filtering unit 121 of the digital image conversion unit 120 to the prediction filtering condition.

If this is explained in detail based on the configuration of FIG. 4, as shown, the noise learning unit 132 includes an analog signal filtering management unit 1321, an input information management unit 1322, a prediction unit 1323, and a learning model verification unit. (1324) may be configured.

First, the analog signal filtering management unit 1321 may receive the I signal and the Q signal generated through filtering of the IF signal from the digital image conversion unit 120 and transmit the received I signal and Q signal to the input information management unit 1322.

In addition, the analog signal filtering management unit 1321 may set a filtering condition for filtering the IF signal in the digital image conversion unit 120, and condition information on the filtering condition may be transferred to the input information management unit 1322. can be sent to

At this time, in the analog signal filtering management unit 1321, setting information for a filtering condition initially set in the digital image conversion unit 120 may be set in advance, and the input information management unit 1322 uses the setting information as the condition information. ) can be provided.

In addition, the analog signal filtering management unit 1321 includes condition information on the filtering condition set in the digital image conversion unit 120 and an I signal provided from the digital image conversion unit 120 in which the filtering condition according to the condition information is set. And matching information obtained by mutually matching the Q signal may be provided to the input information management unit 1322 .

In this case, the matching information may include an I signal, a Q signal, and the condition information.

In addition, the input information management unit 1322 may provide the I signal and the Q signal according to the matching information received from the analog signal filtering management unit 1321 to the signal processing unit 131, and the signal processing unit 131 Calculated values for each of one or more weather variables corresponding to the I signal and the Q signal may be received.

In addition, the input information management unit 1322 provides I and Q signals according to the matching information to the analog image conversion unit 140, based on the I and Q signals from the analog image conversion unit 140. The analog IF signal generated by the analog image conversion unit 140 can be received.

That is, the analog image converter 140 restores the filtered signal, which is an IF signal filtered on the basis of the I and Q signals generated by filtering the IF signal in the digital image converter 120 according to filtering conditions. Thus, an analog IF signal corresponding to the filtering signal may be provided to the input information management unit 1322.

In addition, when receiving the matching information, the input information management unit may transmit request information for requesting environment information corresponding to the matching information to the quality management unit 170 .

Accordingly, the quality control unit 170 communicates with the weather server to receive weather information, or provides a user interface to receive noise environment information such as terrain information and flight training information according to user input through the user interface. can do.

In addition, the quality management unit 170 may provide environment information including at least one of the weather information and noise environment information to the input information management unit of the noise learning unit 132 .

Also, the input information management unit 1322 may provide the calculated value for each weather variable corresponding to the matching information received from the analog signal filtering management unit 1321 to the learning model verification unit.

In addition, the learning model verification unit may provide calculated values for each weather variable corresponding to the matching information to the post-processing data collection unit 160 configured in the signal processing device 100 .

Meanwhile, the post-processing data collection unit 160 receives calculated values for each weather variable corresponding to the matching information from the learning model verification unit of the noise learning unit 132, and calculates the calculated value for each weather variable corresponding to the matching information. After generating a post-processed image by performing post-processing on the weather image corresponding to the matching information according to a user input, the calculated value for each weather variable according to the post-processed image is converted into the learning model. can be provided to the verifier.

That is, the post-processing data collection unit 160 may remove noise from the weather image by the user, generate a calculated value for each weather variable according to the post-processed image, and provide the calculated value to the learning model verification unit. .

In addition, the learning model verification unit mutually compares the calculated value for each weather variable of the weather image corresponding to the matching information with the calculated value for each weather variable of the post-processed image corresponding to the matching information, A similarity score for similarity between the image and the post-processed image may be calculated to correspond to the matching information and provided to the prediction unit 1323 .

Meanwhile, the input information management unit 1322 of the noise learning unit 132 receives the original IF signal from the RF receiver 110, digitally converts the original IF signal to generate original data, and then stores the data. The original data may be stored in the original DB included in the unit 150, and the analog IF signal generated by inversely converting the I signal and the Q signal corresponding to the original IF signal from the analog image conversion unit 140 is received. After digital conversion is performed to generate correction data, the correction data may be stored in a correction DB included in the data storage unit 150 .

In addition, when receiving the matching information, the input information management unit 1322 extracts original data corresponding to the matching information from the original DB, extracts correction data corresponding to the matching information from the correction DB, and then extracts the matching information. Original data and correction data corresponding to the information, condition information included in the matching information, and environment information corresponding to the I signal and the Q signal according to the matching information may be provided to the prediction unit 1323 .

In addition, the prediction unit 1323 may have a learning model set in advance, and original data and correction data provided from the input information management unit 1322 in conjunction with the input information management unit 1322 and the learning model verification unit. Learning data may be generated by mutually matching condition information and environment information with a similarity score provided from the learning model verification unit, and the learning data may be trained in a preset learning model.

In this case, the learning model may be composed of a deep learning algorithm, and the deep learning algorithm may be composed of one or more neural network models.

In addition, the neural network model (or neural network) described in the present invention may be composed of an input layer, one or more hidden layers, and an output layer, and the neural network model includes a deep neural network (DNN) , various types of neural networks such as a recurrent neural network (RNN), a convolutional neural network (CNN), and a support vector machine (SVM) may be applied.

In the above-described configuration, the input information management unit may include weather conditions (heavy rain, typhoon, winter, fog, etc.) or external environmental conditions (geographical information, islands or mountains located in the monitoring area, flight information of airplanes, etc.) according to environmental information. wave height, etc.) may be divided into a plurality of different environment types set in advance, and any one of a plurality of preset environment types may be set in the environment information received from the quality control unit 170 .

In addition, the prediction unit 1323 may create learning models for each of the plurality of different environment types in conjunction with the input information management unit, and accordingly, the prediction unit 1323 corresponds to a plurality of different environment types, respectively. A plurality of learning models to be can be set in advance.

Accordingly, the prediction unit 1323 performs learning data based on environment information included in learning data generated by matching original data, correction data, condition information, and environment information received from the input information management unit with the similarity score. An environment type corresponding to is identified, and the learning data is allowed to be learned in a learning model corresponding to the identified environment type among the plurality of learning models.

In addition, in the configuration described above, the quality control unit 170 may set a filtering learning type for learning a subject requiring filtering to the learning model in the environment information according to a user input.

As this type of filtering learning, various filtering purposes such as clutter filtering, radio noise (or interference) filtering, terrain clutter filtering, echo filtering, and the like can be included.

Accordingly, the prediction unit 1323 may generate a plurality of learning models respectively corresponding to a plurality of learning target types including at least one of a plurality of environment types and a plurality of filtering learning types, and the learning data The learning data may be trained in a learning model corresponding to at least one of an environment type according to included environment information and a filtering learning type.

In addition, the learning model learns the learning data and at the same time, based on the learning data received as an input, a weather image generated to correspond to a filtering condition for converting the original IF signal into the analog IF signal and a weather radar by a user. The degree of similarity between the post-processed images generated by post-processing the weather image according to the environmental characteristics of the region where (1) is located is determined through the similarity score, and the similarity score is calculated based on the algorithm generated through learning. A prediction filtering condition for increasing can be calculated as an output.

In this case, the prediction filtering condition may be a filtering condition considering environmental information corresponding to the learning model and a filtering condition for which a filtering setting value for filtering a subject requiring filtering in the corresponding environment information is set.

Also, the prediction unit 1323 may transmit the prediction filtering condition calculated through the learning model to the analog signal filtering management unit 1321 .

In addition, the analog signal filtering management unit 1321 may replace the existing filtering condition set in the digital image conversion unit 120 with the prediction filtering condition, and through this, the existing filtering condition set in the digital image conversion unit 120 A condition may be changed to the prediction filtering condition.

At this time, the analog signal filtering management unit 1321 may change the existing filtering condition set in the setting information to the prediction filtering condition.

In addition, when the analog signal filtering management unit 1321 receives the I signal and Q signal generated by filtering the IF signal with the predictive filtering condition from the digital image converter 120 for which the predictive filtering condition is set, the predictive filtering condition Matching information including I signals and Q signals corresponding to conditions and condition information on prediction filtering conditions according to the setting information may be generated and provided to the input information management unit 1322 .

According to the configuration described above, the digital image converter 120 filters the original IF signal, which is an IF signal received through the RF receiver 110 after the predictive filtering condition is set based on the predictive filtering condition. An I signal and a Q signal corresponding to the prediction filtering condition may be generated and then provided to the control unit 130 .

In addition, the noise learning unit 132 of the control unit 130 includes the digital image conversion unit 120, the quality management unit 170, the analog image conversion unit 140, and the post-processing data collection unit 160. It is possible to generate learning data corresponding to the prediction filtering condition in the same manner as the above-described process of generating learning data according to filtering conditions in conjunction with, and environment according to environment information included in the learning data among the plurality of learning models. By applying the learning data as an input to a specific learning model corresponding to a type, another predictive filtering condition corresponding to the generated learning data corresponding to the predictive filtering condition may be calculated through the specific learning model.

At this time, the post-processing data collection unit 160 receives the calculated value for each weather variable calculated to correspond to the prediction filtering condition from the control unit 130, and calculates the calculated value for each weather variable corresponding to the prediction filtering condition. The calculated value for each weather variable according to the post-processed image generated by post-processing the weather image according to the user input may be provided to the control unit 130 .

In addition, the noise learning unit 132 of the control unit 130 calculates a calculated value for each weather variable according to a weather image corresponding to the predictive filtering condition and a calculated value for each weather variable according to a post-processed image corresponding to the predictive filtering condition. are compared to each other to calculate a similarity score between the weather image corresponding to the predictive filtering condition and the post-processed image, and the calculated similarity score, original data and correction data corresponding to the predictive filtering condition, and the quality management unit 170 Learning data corresponding to the predictive filtering condition including environment information corresponding to the predictive filtering condition received from the controller may be generated and trained in the specific learning model.

Accordingly, the noise learning unit 132 learns the learning data according to the predictive filtering conditions and similarity scores between a weather image generated according to the predictive filtering condition and a post-processed image obtained by post-processing the weather image by the user. Based on , if the similarity score of the predictive filtering condition increases compared to the existing filtering condition, other prediction filtering conditions in which the parameters of some attributes that can increase the similarity score are changed while maintaining some of the parameters for each attribute included in the predictive filtering condition. , and when the similarity score of the predictive filtering condition decreases compared to the existing filtering condition, another predictive filtering condition can be created while setting parameters for each property included in the predictive filtering condition as exclusion targets.

That is, the noise learner 132 may proceed with learning in the direction of generating prediction filtering conditions that increase the similarity score through learning of the learning model.

In addition, the noise learning unit 132 replaces the existing prediction filtering condition set in the digital image conversion unit 120 with the other prediction filtering condition when calculating the other prediction filtering condition, so that the digital image conversion unit 120 The existing predictive filtering condition set in may be changed to the other predictive filtering condition.

In addition, the noise learning unit 132 repeatedly calculates a different predictive filtering condition through the above-described operation whenever a similarity score that is less than a preset reference value is calculated, and converts the calculated predictive filtering condition to the digital filtering condition. It is set in the image conversion unit 120 to perform learning on the learning model through the above-described process.

On the other hand, when the noise learning unit 132 or the prediction unit 1323 included in the noise learning unit 132 calculates that the similarity scores for all IF signals received during a preset time are equal to or greater than a preset reference value. It may be determined that learning of a plurality of learning models is completed.

In addition, the input information management unit 1322 of the noise learning unit 132 interlocks with the prediction unit 1323 to complete the learning of the plurality of learning models, and after completion of learning of the plurality of learning models, the RF receiver The specific IF signal, which is the original IF signal received through 110, is digitally converted to generate original data, and after receiving environmental information corresponding to the specific IF signal from the quality control unit 170, the specific IF signal is corresponded to. The original data and the environment information are transmitted to the prediction unit 1323, and the prediction unit 1323 transmits the original data and environment information corresponding to the specific IF signal to a specific environment type corresponding to the environment information. After applying the learning model as an input, an optimal filtering condition corresponding to the specific IF signal is calculated through the specific learning model, and then transmitted to the analog signal filtering management unit 1321.

In addition, the analog signal filtering management unit 1321 sets the optimal filtering condition in the digital image conversion unit 120, and the digital image conversion unit 120 filters the IF signal according to the optimal filtering condition. can make this happen.

5 shows a weather image before filtering the IF signal according to the optimal filtering condition calculated by the signal processing apparatus 100 according to an embodiment of the present invention and a weather image after filtering the IF signal according to the optimal filtering condition. is the drawing shown.

As shown, it can be seen that noise appearing in the weather image before filtering is largely removed from the weather image generated based on the filtered IF signal after filtering the IF signal according to the optimal filtering condition.

As described above, the present invention provides a weather image according to a calculated value for each weather variable obtained based on a signal obtained by filtering an IF signal received from a weather radar according to a filtering condition and a post-processing process by a user for the weather image. The degree of similarity between the obtained post-processed images is calculated, and the correlation between the degree of similarity and the filtered signal obtained from the IF signal according to the filtering condition and the environmental information of the region where the weather radar is located and the filtering condition and the IF signal is converted into a learning model. A learning model can be trained to calculate an optimal filtering condition in which a weather image obtained from the filtered signal is as similar as possible to a post-processed image, and through this, a weather image by reflecting the environmental characteristics of a region where a weather radar is deployed by a user It is a process of removing noise caused by environmental characteristics from the IF signal through optimal filtering conditions without post-processing. In addition to enhancing user convenience, it is possible to support acquisition of a highly reliable weather image from which non-weather elements are removed through noise removal.

In addition, the present invention automatically recognizes filtering conditions having criteria similar to those set by the user for a post-processed image generated through post-processing of a weather image in consideration of the environmental characteristics of the region where the weather radar is placed, based on learning. Through this, the signal processing device can automatically set the optimal filtering conditions for the IF signal of the weather radar connected to the signal processing device by automatically identifying the environmental characteristics adaptively for different weather radars deployed in various regions. The convenience and efficiency of operation can be greatly improved.

Components described in the embodiments of the present invention, for example, a storage unit such as a memory, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable One or more general-purpose computers or hardware, such as gate arrays, programmable logic units (PLUs), microprocessors, software containing instruction sets, or combinations thereof, or any other device capable of executing and responding to instructions. It can be implemented using a special purpose computer.

The foregoing may be modified and modified by those skilled in the art without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention, but to explain, and the scope of the technical idea of the present invention is not limited by these embodiments. The protection scope of the present invention should be construed according to the claims below, and all technical ideas within the equivalent range should be construed as being included in the scope of the present invention.

1: weather radar 100: signal processing device
110: RF receiver 120: digital image converter
121: first filtering unit 122: first amplifying unit
123: analog-to-digital conversion unit
124: DDC unit 130: control unit
131: signal processing unit 1311: correlation coefficient calculation unit
1312: weather moment calculation unit 1313: weather variable calculation unit
132: noise learning unit 1321: analog signal filtering management unit
1322: input information management unit 1323: prediction unit
1324: learning model verification unit 140: analog image conversion unit
141: second amplifier 142: second filtering unit
143: digital-to-analog converter
144: DUC unit 150: data storage unit
160: post-processing data collection unit 170: quality control unit

Claims (10)

a digital image conversion unit for receiving an IF signal generated based on an RF signal received from a weather radar, filtering the IF signal according to a preset filtering condition, and generating an I signal and a Q signal based on the generated filtering signal;
an analog image conversion unit generating an analog IF signal based on the I and Q signals;
a signal processing unit which calculates a correlation coefficient based on the I signal and Q signal, calculates a wake-up moment based on the correlation coefficient, and calculates a calculated value for each of one or more preset weather variables based on the wake-up moment;
a quality management unit receiving and providing environmental information of a region where the weather radar is located; and
After receiving the calculated value for each weather variable for the post-processed image generated by post-processing the weather image according to the calculated value for each weather variable according to the user input, the calculated value for each weather variable received from the signal processing unit is compared with the calculated value for each weather variable. A similarity score corresponding to a filtering condition is calculated, and a learning model corresponding to an environment type according to environment information received through the quality control unit among a plurality of learning models respectively corresponding to a plurality of different environment types set in advance is selected according to the filtering condition. Learning original data generated based on the IF signal corresponding to , correction data generated based on the analog IF signal corresponding to the filtering condition, filtering condition, and the environment information together with a similarity score corresponding to the filtering condition. noise learning unit for changing the filtering condition of the digital image conversion unit to the predictive filtering condition after calculating a predictive filtering condition for increasing the similarity score through the learning model while
A signal processing device for a weather radar that provides a noise filtering function using environmental information comprising a.
The method of claim 1,
The environmental information may include at least one of noise environment information received according to a user input including at least one of topographical information of an area where the weather radar is located and flight information of an airplane flying in the corresponding area and weather information received from a weather server. A signal processing device for a weather radar that provides a noise filtering function using environmental information, comprising one.
The method of claim 1,
The digital image conversion unit,
a first filtering unit generating a filtered signal by filtering the IF signal according to the filtering condition;
a first amplifier to amplify the filtered signal;
an analog-to-digital conversion unit converting the filtering signal amplified by the first amplification unit into a digital signal to generate meteorological data; and
a DDC (Digital Down Converter) unit converting the meteorological data into an I signal and a Q signal and providing the converted signals to the noise learning unit;
A signal processing device for a weather radar providing a noise filtering function using environmental information, comprising:
The method of claim 1,
The analog image conversion unit,
a Digital Up Converter (DUC) unit that converts the I and Q signals into IF digital signals;
a digital-to-analog converter converting the IF digital signal into an analog IF signal;
a second filtering unit filtering the analog IF signal; and
A second amplifier for amplifying the analog IF signal filtered through the second filtering unit and providing it to the noise learning unit
A signal processing device for a weather radar providing a noise filtering function using environmental information, comprising:
The method of claim 1,
The signal processing unit,
a correlation coefficient calculator calculating correlation coefficient data including a dual polarization cross-correlation coefficient and a time-domain correlation coefficient based on the I and Q signals;
a wake-up moment calculator calculating a critical variable based on the correlation coefficient data calculated through the correlation coefficient calculator and then calculating a wake-up moment based on the critical variable; and
A weather variable calculation unit for calculating a calculated value for each weather variable based on the weather moment and providing the calculated value to the noise learning unit
A signal processing device for a weather radar providing a noise filtering function using environmental information, comprising:
The method of claim 1,
The weather variable is a signal processing device for a weather radar that provides a noise filtering function using environmental information, characterized in that it includes at least one of reflectivity, radial velocity, spectral width, cross-correlation coefficient, and differential phase difference.
The method of claim 1,
When the predictive filtering condition is set, the digital image conversion unit filters the IF signal received from the weather radar according to the predictive filtering condition, and generates an I signal and a Q signal corresponding to the predictive filtering condition based on a filtering signal generated. create,
The analog image conversion unit generates an analog IF signal corresponding to the prediction filtering condition based on the I signal and the Q signal corresponding to the prediction filtering condition;
The signal processing unit calculates a calculated value for each weather variable based on the I signal and the Q signal corresponding to the prediction filtering condition;
The quality control unit generates and provides environmental information corresponding to the prediction filtering condition,
The noise learning unit receives a calculated value for each weather variable according to a post-processed image generated by post-processing a weather image according to a calculated value for each weather variable corresponding to the prediction filtering condition according to a user input, and then receives the result from the signal processing unit. A similarity score is calculated by comparing the calculated value for each weather variable corresponding to the predicted filtering condition, and the learning model corresponding to the environment type according to the environmental information corresponding to the predicted filtering condition received through the quality control unit Original data generated based on the IF signal corresponding to the predictive filtering condition, correction data generated based on the analog IF signal corresponding to the predictive filtering condition, environment information corresponding to the predictive filtering condition, and the predictive filtering condition while learning with a similarity score corresponding to the predictive filtering condition, another predictive filtering condition in which the similarity score increases is calculated through the learning model, and then the existing filtering condition set in the digital image conversion unit is used as the other predictive filtering condition. A signal processing device for a weather radar that provides a noise filtering function using environmental information, characterized in that for changing.
The method of claim 1,
The noise learning unit determines that the learning of a plurality of learning models is completed when the similarity scores for all IF signals received during a predetermined time are calculated to be equal to or greater than a predetermined reference value, and learning of the plurality of learning models is completed. When completed, the specific IF signal received after completion of learning for the plurality of learning models is digitally converted to generate original data, and after receiving environmental information corresponding to the specific IF signal from the quality control unit, the response to the specific IF signal is performed. The original data and the environment information corresponding to the specific IF signal are applied as inputs to a specific learning model corresponding to the type of environment according to the environment information corresponding to the specific IF signal, and the optimum corresponding to the original data and environment information is applied through the specific learning model. A signal processing device for a weather radar that provides a noise filtering function using environmental information, characterized in that filtering conditions are calculated and set in the digital image conversion unit.
The method of claim 1,
The noise learning unit generates and provides a weather image based on the calculated value for each weather variable received from the signal processor, generates the post-processed image through post-processing according to a user input for the weather image, and then generates the post-processed image. A signal processing device for a weather radar that provides a noise filtering function using environmental information, further comprising a post-processing data collection unit generating a calculated value for each weather variable according to the processed image and providing the calculated value to the noise learning unit.
The method of claim 1,
The noise learning unit digitally converts the IF signal to generate the original data, and digitally converts the analog IF signal to generate the correction data. signal processing device.

Images