A device for airport noise identification and measurement based on convolutional neural network
By using an airport noise recognition device based on a convolutional neural network, combined with a Raspberry Pi device, real-time and accurate identification and monitoring of airport noise has been achieved, solving the problems of mobility and misidentification of traditional devices and improving recognition efficiency.
Patent Information
- Application Number
- CN202210397679.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-04-15
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2042-04-15
AI Technical Summary
Traditional noise identification equipment is power-consuming, bulky, difficult to move, expensive, and has a high false identification rate, making it unable to accurately identify airport noise sources.
An airport noise identification and measurement device based on convolutional neural networks is adopted. It is connected to a Raspberry Pi through a microphone, decibel meter and display screen. The convolutional neural network is used to extract and classify audio data features, and a 1/3 octave band filtering algorithm is combined to extract and identify noise features.
It enables real-time identification and accurate monitoring of airport noise, reduces the false identification rate, improves identification efficiency, and reduces human resource consumption.
Smart Images

Figure CN114758675B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of speech information processing, and in particular relates to an airport noise recognition and measurement device based on a convolutional neural network. Background Art
[0002] In recent years, with the rapid development of the civil aviation industry, the airport throughput capacity and land area have been continuously expanded. At the same time, the urbanization process in the areas surrounding the airport has accelerated. As the airport land is getting closer and closer to the urban land, the conflicts and disputes caused by the noise pollution of the airport are increasing. With the improvement of people's quality of life, people's awareness of legal rights protection is getting stronger. Some airports have seen incidents where nearby residents broke into the airport to prevent aircraft from taking off and landing because they could not bear the airport noise, which seriously affected the flight safety of the aircraft. Therefore, under the premise of vigorously advocating the sustainable development of airports, the research on aircraft noise identification has become an important issue facing civil aviation administrative departments at all levels, airport authorities, and airlines.
[0003] Traditional noise identification equipment is based on PCs, which consume high power, are bulky, difficult to move, prone to crashes, and expensive. At the same time, it mainly determines whether the noise is generated by an aircraft by identifying the intensity and duration of audio data, without considering the actual sound source, so there is a problem of misidentification. Summary of the invention
[0004] In view of the problems existing in the prior art, the present invention proposes an airport noise identification and measurement device based on a convolutional neural network. The convolutional neural network method extracts complex feature information in the audio, reducing the probability of misidentification. At the same time, the feature extraction and classification process of different noise types is realized through historical audio data training. The network model can extract features that are not easy to notice, and has strong practicality and versatility. At the same time, the noise intensity of the aircraft is tested by a decibel meter and output on the display screen, which can better monitor the airport noise.
[0005] The purpose of the present invention is to provide an airport noise identification and measurement device based on a convolutional neural network, comprising:
[0006] Model construction module, builds a noise classification model based on convolutional neural network; the specific construction process is:
[0007] Collect historical audio data of airport aircraft noise and non-aircraft noise, and obtain feature vectors after preprocessing operations; input the feature vectors into the convolutional neural network, and output the probability that the noise data belongs to aircraft noise and non-aircraft noise, that is, the category matrix; calculate the loss based on the output and annotation information, and back-propagate to update the model parameters, and finally obtain a noise classification model based on the convolutional neural network;
[0008] Device building module, build a noise recognition device; the specific construction process is:
[0009] Connect the microphone, decibel meter and display screen to the Raspberry Pi via USB to form a noise recognition device.
[0010] A transplantation module is used to compress the noise classification model and then transplant it into the Raspberry Pi to obtain an aircraft noise recognition device;
[0011] The audio acquisition module uses a microphone to collect airport noise data and obtain the audio to be tested;
[0012] Noise data preprocessing module; specifically:
[0013] The collected noise signal is segmented, normalized, interference noise is removed, useless background noise is removed, the signal-to-noise ratio is enhanced, and characteristic information in the noise is highlighted;
[0014] Feature extraction module; specifically:
[0015] With the help of 1 / 3 octave filtering algorithm, the sound pressure level spectrum analysis method is used to extract the characteristics of the recorded noise from the perspective of the main frequency distribution of the noise, and the characteristic vector of the audio data of the noise to be tested is obtained;
[0016] Classification and recognition module; specifically:
[0017] The noise classification model is used to classify the feature vector, thereby determining whether the noise is generated by an aircraft.
[0018] Preferably, it also includes: a result output module, if the noise is generated by an aircraft, using a decibel meter to test the noise intensity, and finally outputting the noise intensity on a display screen, otherwise, no processing is performed.
[0019] In summary, the advantages and positive effects of the present invention are:
[0020] The present invention identifies the noise type through a convolutional neural network, and further tests the sound intensity of the noise generated by the aircraft. A convolutional neural network, i.e., CNN, is used to extract features from the original audio data to obtain a CNN noise recognition model, and the model is implanted into a Raspberry Pi. At the same time, a Raspberry Pi mini microphone and a decibel meter are connected to the Raspberry Pi through a USB direct plug-in method to form a noise recognition device. The noise recognition device collects noise audio data in real time, processes the collected noise signal, and sends the pre-processed feature vector into a convolutional neural network for classification to obtain the probability that the noise data belongs to different category attributes. The pre-processing operation segments and normalizes the collected noise signal, removes interference noise, removes useless background noise, enhances the signal-to-noise ratio, highlights the feature information in the noise, and optimizes the recognition effect. If it is found that the noise is generated by an aircraft, the decibel meter is used to monitor the noise sound intensity, and finally outputs it on a display screen. The present invention realizes real-time detection of different airport noise categories, reduces the probability of misidentification, and improves the accuracy of aircraft noise identification.
[0021] The present invention can collect the noise in the airport in real time and simultaneously identify whether the noise is generated by an airplane, thereby greatly reducing the human resource consumption of the airport and improving the efficiency of noise identification. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 This is a flow chart of CNN aircraft noise recognition in a preferred embodiment of the present invention;
[0023] Figure 2 This is a schematic diagram of an aircraft noise identification device in a preferred embodiment of the present invention;
[0024] Figure 3 Flowchart of a preferred embodiment of the present invention. DETAILED DESCRIPTION
[0025] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0026] Noise recognition based on convolutional neural networks is essentially a classification problem. The sound source of the current audio can be detected by extracting and classifying historical audio features. For historical noise audio data, it is necessary to establish a classification model based on the features to classify the audio data. The feature vector obtained after preprocessing the historical audio data is sent to the convolutional neural network for feature extraction and training to obtain a CNN noise classifier. The CNN noise classification model obtained through pre-training can classify different noise audio data.
[0027] The aircraft noise identification method proposed here is to effectively evaluate airport noise and avoid the waste of resources caused by the storage and transmission of a large amount of non-aircraft noise data at noise monitoring points. Aircraft noise has the following characteristics: ① The volume of aircraft noise is high. The aircraft engine is the main noise source. When the aircraft takes off, it may exceed 140 decibels; ② The sound source of aircraft noise is complex. The sound source of aircraft noise comes from three sources: mechanical noise, that is, the rotation of engine parts; aerodynamic noise, that is, the airflow around the surface of the aircraft, especially when flying at high speed and low speed; aircraft system, that is, the noise generated by the cockpit and cabin pressurization and regulation system and auxiliary power unit; ③ The contribution value of each noise source of different types of aircraft is different. The main noise source of jet aircraft is jet noise, the main noise of helicopters is the propeller of the aircraft, and the main noise source of modern large passenger aircraft is the fan and the fuselage.
[0028] According to the characteristics of different noises at the airport, the present invention realizes real-time detection of different noise categories, reduces the probability of misidentification, and improves the accuracy of aircraft noise identification.
[0029] See also Figures 1 to 3 , an airport noise identification and measurement device based on convolutional neural network requires two implementation stages, namely the convolutional neural network training stage and the aircraft noise identification device detection stage. Including:
[0030] The model building module collects historical audio data of airport aircraft noise and non-aircraft noise, and obtains feature vectors after preprocessing. The feature vectors are input into the convolutional neural network (CNN), and the final output is the probability that the noise data belongs to aircraft noise and non-aircraft noise, that is, the category matrix. The loss is calculated based on the output and annotation information, and the model parameters are updated by back propagation, and finally a noise classification model based on the convolutional neural network is obtained.
[0031] Device building module, as attached Figure 2 As shown in the figure, the Raspberry Pi mini microphone, decibel meter and display screen are connected to the Raspberry Pi via USB direct plug-in to form a noise recognition device.
[0032] The transplantation module compresses the noise classification model and transplants it to the Raspberry Pi to obtain an aircraft noise recognition device. In the device, the Raspberry Pi mini microphone is used as input, the noise classification model is used to process the noise audio data, and the decibel meter is used to display the sound intensity corresponding to the aircraft noise.
[0033] The audio acquisition module uses a Raspberry Pi mini microphone to collect airport noise data to obtain the audio to be tested;
[0034] The noise data preprocessing module segments and normalizes the collected noise signals, removes interference noise, removes useless background noise, enhances the signal-to-noise ratio, and highlights the characteristic information in the noise;
[0035] The feature extraction module, in order to obtain sufficient information from the noise event, uses the 1 / 3 octave filtering algorithm and the sound pressure level spectrum analysis method to extract the features of the recorded noise from the perspective of the main frequency distribution of the noise, and obtains the feature vector of the noise audio data to be tested;
[0036] The classification and recognition module uses the trained noise classification model to classify the audio data to be detected, so as to determine whether the noise is generated by the aircraft;
[0037] The result output module, if the noise is generated by an aircraft, uses a decibel meter to test the noise intensity and finally outputs the noise intensity on the display screen.
[0038] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When the use is implemented in whole or in part in the form of a computer program product, the computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL) or wireless (e.g., infrared, wireless, microwave, etc.) mode) to another website site, computer, server or data center. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk SolidState Disk (SSD)), etc.
[0039] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. An airport noise identification and measurement device based on convolutional neural network, characterized in that: include: Model building module, building a noise classification model based on convolutional neural network; The specific construction process is: Collect historical audio data of airport aircraft noise and non-aircraft noise, and obtain feature vectors after preprocessing operations; input the feature vectors into the convolutional neural network, and output the probability that the noise data belongs to aircraft noise and non-aircraft noise, that is, the category matrix; calculate the loss based on the output and annotation information, and back-propagate to update the model parameters, and finally obtain a noise classification model based on the convolutional neural network; Device building module, build noise recognition device; The specific construction process is: Connect the microphone, decibel meter and display screen to the Raspberry Pi via USB to form a noise recognition device. A transplantation module is used to compress the noise classification model and then transplant it into the Raspberry Pi to obtain an aircraft noise recognition device; The audio acquisition module uses a microphone to collect airport noise data and obtain the audio to be tested; Noise data preprocessing module; Specifically: The collected noise signal is segmented, normalized, interference noise is removed, useless background noise is removed, the signal-to-noise ratio is enhanced, and characteristic information in the noise is highlighted; Feature extraction module; Specifically: With the help of 1 / 3 octave filtering algorithm, the sound pressure level spectrum analysis method is used to extract the characteristics of the recorded noise from the perspective of the main frequency distribution of the noise, and the characteristic vector of the audio data of the noise to be tested is obtained; Classification and recognition module; specifically: The noise classification model is used to classify the feature vector, thereby determining whether the noise is generated by an aircraft.
2. The airport noise identification and measurement device based on convolutional neural network according to claim 1, characterized in that: Also includes: The result output module uses a decibel meter to test the noise intensity if the noise is generated by an aircraft, and finally outputs the noise intensity on the display screen. Otherwise, no processing is performed.
3. The airport noise identification and measurement device based on convolutional neural network according to claim 1, characterized in that: The microphone is a Raspberry Pi mini microphone.
Citation Information
Patent Citations
Speech recognition method and device with adaptive noise reduction capability
CN113205803A
Method and system for detecting voice activity innoisy conditions
WO2020043160A1
