TW201701240A - Automatic recognition monitoring system for indoor noise pollution - Google Patents

Abstract

The invention relates to the field of monitoring, more particularly, to an automatic recognition monitoring system for indoor noise pollution. Sample the noise signal through the multi-channel microphone array (FPGA), preprocess, extract feature, select feature of the noise signal through the digital signal processor (DSP), and establish feature model of the noise signal in the cloud, and recognize the noise source and the corresponding noise monitoring information through the technology matching with the feature model of the noise signal, finally display the noise recognition information on the main interface in the form of chart. The system effectively makes users aware of the danger of noise pollution, and reminds users away from the noise danger and actively taking prevention measures to improve the living environment, thus has played a positive role of promoting economic development.

Description

Indoor noise pollution automatic identification and monitoring system

The invention relates to the field of monitoring, in particular to an indoor noise pollution automatic identification and monitoring system.

Noise pollution is an environmental issue that is of great concern to the whole world. Excessive environmental noise will have a certain impact on people's physical and mental health. A noisy environment can cause serious damage to people's hearing and physical health. The general noise will have a certain impact on people's daily work and life. According to statistics, noise pollution causes different degrees of harm to more than 70% of urban residents worldwide. As a developing country, the impact of noise pollution on the people in China is more prominent. Environmental noise monitoring is an important part of human beings to improve their quality of life and strengthen environmental protection. In the bustling streets and residential areas of major cities, large-scale environmental noise displays have been erected on the streets. Therefore, the detection and control of noise on the environment plays an important role in human health and new health. It is especially important to strengthen the detection of environmental noise.

At present, domestic portable noise testers are mostly expensive imported special equipments. Except for environmental protection professional departments such as sanitation and metering, they cannot be popularized as civilian products. Most of the noise monitoring instruments use integrated statistical sound level meters with single-chip processing functions. They are portable instruments. The sensitivity of these sound level meters varies with air pressure, temperature and humidity, affecting measurement accuracy, and requires frequent calibration. It is difficult to apply in daily life scenes. . Especially in indoor environmental noise monitoring, since indoor noise is mostly low-medium noise, it is difficult for traditional sound pressure meters to accurately classify and identify noise sources, and it is more difficult to provide users with accurate monitoring information of indoor environmental noise pollution.

Therefore, there is a need for a system that can automatically identify and monitor indoor noise pollution.

In view of the above problems, the present invention provides an indoor noise pollution automatic identification and monitoring system, comprising: a noise sample extraction module for capturing noise samples and converting the noise samples into digital signals; DSP digital signal processing module And connecting to the noise sample extraction module, processing the digital signal and extracting noise characteristics; and a noise classification processing module connected to the DSP digital signal processing module to extract the noise characteristic The noise feature pattern matching is performed to further classify and identify the noise sample, and the noise recognition visualization module is connected to the noise classification processing module to display the result of the classification and recognition of the noise sample.

In a preferred embodiment, the DSP digital signal processing module includes: a pre-processing module connected to the noise sample capturing module, performing noise reduction processing on the digital signal, providing information a feature extraction module, connected to the preprocessing module, selects and transforms the digital signal after the noise reduction process; and the feature selection module is connected with the feature extraction module to extract the selected and transformed The digital signal.

In the above system, the noise classification processing module includes: a classifier connected to the feature selection module, classifying the noise sample to establish a noise model library; and a noise classification recognition module respectively The device is connected to the noise recognition visual module, and performs pattern matching on the digital signal extracted by the feature selection module and the noise model library, thereby classifying and identifying the digital signal.

In the above system, the noise reduction processing includes linear filtering, median filtering, and wavelet transform.

In the above system, the feature extraction module extracts a feature vector of the digital signal by a mapping method.

In the above system, the classifier is a DHMM hidden Markov classifier.

In the above system, the noise sample capturing module includes an FPGA multi-channel microphone array signal extractor, and the FPGA multi-channel microphone array signal extractor extracts the noise sample.

In the above system, the noise sample capturing module further includes a multi-channel analog digital converter connected to the FPGA multi-channel microphone array signal extractor, and the multi-channel analog digital converter converts the noise sample Into the digital signal.

In the above system, the noise sample capturing module further includes a USB multi-channel microphone transmitter, respectively connected to the multiple analog digital converter and the DSP digital signal processing module, and package the digital signal And forming the packet, and transmitting the digital signal encapsulated into a packet form to the DSP digital signal processing module by using a USB protocol.

In the above system, the noise recognition visualization module displays the result of the classification and recognition of the noise sample in a graphical manner.

In summary, the present invention provides an indoor noise pollution automatic identification and monitoring system, which uses a multi-channel microphone array of FPGA to sample the noise signal, and performs preprocessing, feature extraction and feature selection on the noise signal through the DSP digital signal processor. The characteristic model training of the noise signal is carried out in the cloud, and the noise source and the corresponding noise monitoring information are identified by the technique of noise signal feature pattern matching, and finally the noise identification information is displayed on the main interface in a graphical manner. The system is applied to the home robot, and the home robot provides the user with noise monitoring service in a companion manner, and is accompanied by a voice recognition interaction mode, which can broadcast the indoor noise pollution in real time, effectively making the user aware of the noise pollution hazard, and Remind users to stay away from noise hazards and actively take preventive measures to improve the living environment, and thus play a positive role in promoting economic development.

In order to make the technical solutions and advantages of the present invention easier to understand, the following detailed description will be made with reference to the accompanying drawings. It should be noted that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

As shown in FIG. 1 , an indoor noise pollution automatic identification and monitoring system designed by the present invention includes: a noise sample capturing module for capturing a noise sample and converting the noise sample into a digital signal;

The DSP digital signal processing module is connected with the noise sample capturing module to process the digital signal and extract the noise characteristics; the noise classification processing module is connected with the DSP digital signal processing module to perform noise on the extracted noise characteristics. Feature pattern matching, which further classifies and identifies noise samples; the noise recognition visualization module is connected with the noise classification processing module to display the results of the classification and recognition of the noise samples, and the display is performed by DLP internal injection through a home robot or other artificial intelligence tools. The display technology displays the noise source and noise pollution index corresponding to the indoor noise recognition result to the user by means of a graphical interface display.

The DSP digital signal processing module of the system further comprises: a preprocessing module, which is connected with the noise sample capturing module, performs noise reduction processing on the digital signal to provide a signal to noise ratio; and the feature extraction module is connected with the preprocessing module. Selecting and transforming the digital signal after the noise reduction processing; the feature selection module is connected with the feature extraction module to extract the selected and transformed digital signals.

In the present invention, the noise sample extraction module refers to extracting and analyzing the noise of the living room of the target user, and classifying the noise. Specifically, the present invention provides an FPGA multi-channel microphone array signal extractor, which realizes phase synchronization capture of each microphone in the microphone array. The picker uses a multi-channel ADC to synchronously convert the multi-channel microphone signal into a digital signal, and the pick-up device realizes the key technical indicators of the phase synchronization of the multi-channel microphone through the DMA terminal.

The noise signal pre-processing module refers to performing noise reduction processing on the digital signal to provide a signal-to-noise ratio. Common noise reduction methods include linear filtering, median filtering, wavelet transform, and so on.

The signal feature extraction module means that in order to effectively implement classification and recognition, it is necessary to select and transform the noise-reduced digital signal to obtain the feature that best reflects the essence of the classification. Feature is the basis of recognition. Feature extraction refers to the most effective feature obtained by mapping method. The dimension of feature space is changed from high dimension to low dimension. Feature extraction is the key to noise target recognition. The main purpose is to extract and distinguish. The eigenvectors of various types of targets.

The noise signal feature selection module refers to the fact that since the feature extraction depends on the physical characteristics and related knowledge of the specific problem, people often record all the features that may be observed. Many of these features may not be closely related to the problem to be solved. However, the excessive amount of features leads to problems such as large amount of calculation and poor generalization ability. At the same time, when the number of noise signal feature samples is limited, many methods cannot be calculated due to problems such as ill-conditioned matrices. In order to improve the performance of the noise classifier, the noise signal feature selection realizes classification with as few features as possible while ensuring the classification effect.

In the present invention, the noise classification processing module includes: a classifier connected to the feature selection module to classify the noise samples to establish a noise model library. The classifier refers to the classification of indoor noise samples by artificial neural network in the machine, and uses DHMM hidden Markov as the classifier of environmental noise to establish the noise model library; the noise classification and identification module is respectively used in the classifier. The noise recognition visual module is connected, and the digital signal extracted by the feature selection module is matched with the noise model library, and the digital signal is classified and recognized.

The following describes the specific embodiments:

At present, indoor low-frequency noise, which uses household appliances as the main noise source, has become a non-negligible noise source. The actual monitoring shows that the household refrigerator is 35~50 decibels, the washing machine is 50~70 decibels, the electric fan is 55~70 decibels, the vacuum cleaner is 60~80 decibels, and the home theater can reach 60~80 decibels, which obviously increases the residence. The degree of noise pollution in the room. The severity of the hazards caused by indoor noise does not cause human illness or even death like air pollution and water pollution. However, since most modern people spend indoors, indoor noise pollution can affect people's psychological conditions, leading to lesions in the auditory, nervous system and endocrine system, causing greater harm to people's daily lives.

The technology of the invention can effectively overcome the problem that the sensitivity of the traditional sound pressure meter noise monitoring instrument changes with the pressure, the temperature and the humidity, affects the measurement precision, requires frequent calibration, and is difficult to accurately classify and monitor the noise source, and realizes a problem through the FPGA. The channel microphone array samples the noise signal, performs preprocessing, feature extraction, feature selection on the noise signal through the DSP digital signal processor, and performs characteristic model training on the noise signal in the cloud, and adopts the technology of noise signal feature pattern matching. The noise source and the corresponding noise monitoring information are identified, and finally the noise identification information is graphically displayed on the main interface of the robot through the DLP internal projection module. The robot provides the user with noise monitoring service in the accompanying way, and is accompanied by the interactive mode of voice recognition. It can broadcast the indoor noise pollution in real time, effectively making the user aware of the harm of noise pollution, and reminding the user to stay away from the noise hazard and actively take preventive measures. Measures to improve the living environment have played a positive role in promoting economic development.

As shown in FIG. 1 , the present invention provides a robot indoor noise automatic identification and monitoring system, which realizes the automatic indoor identification and indoor noise pollution monitoring and provides a visual display of indoor noise pollution for the user.

Indoor noise pollution has brought a lot of inconvenience to people's lives, which has greatly reduced people's quality of life. In order to effectively monitor indoor noise pollution, the robot indoor noise automatic identification and monitoring system first captures the audio signal of the microphone array through the FPGA multi-channel microphone array signal extractor. The FGPA multi-channel microphone array extractor synchronously converts the multi-channel microphone signal into a digital signal through a multi-channel ADC, and then realizes phase synchronization of the multi-channel microphone by means of DMA interruption. Subsequently, the USB multi-channel microphone transmitter encapsulates the microphone signal captured in the DMA into a packet, and transmits the data packet to the DSP digital signal processing main controller through the USB protocol. After the noise sample is taken, the DSP main controller performs preprocessing and feature extraction on the noise signal. In order to prevent the influence of excessive feature signals on the noise model training, the noise characteristics need to be selected. After the feature extraction of the noise signal is completed, the noise feature sample set is sent to the cloud through the network for feature machine model training. Automatic identification of indoor environmental noise can be performed after the completion of the training of the noise characteristic model. After the feature extraction of the noise signal is completed, the noise feature is sent to the cloud to enter the noise feature pattern matching through the network protocol, thereby identifying the corresponding noise source type and noise value. Subsequently, the cloud returns the noise recognition result to the DSP main controller through the network protocol. The DSP main controller displays the noise recognition result on the main interface of the robot through the DLP internal projection module in a graphical manner.

Exemplary embodiments of the specific structure of the specific embodiments are given by way of illustration and the accompanying drawings, and other transitions are possible in accordance with the spirit of the invention. Although the above invention proposes a prior preferred embodiment, these are not intended to be limiting.

Various changes and modifications will no doubt become apparent to those skilled in the <RTIgt; Accordingly, the appended claims are to cover all such modifications and modifications The scope and content of any and all equivalents are intended to be within the scope and spirit of the invention.

no

Embodiments of the present invention are described more fully with reference to the accompanying drawings. However, the attached drawings are for illustration and illustration only and are not intended to limit the scope of the invention. Figure 1 is a schematic view showing the structure of the system of the present invention.

Claims (10)

An indoor noise pollution automatic identification monitoring system includes: a noise sample extraction module for capturing a noise sample and converting the noise sample into a digital signal; a DSP digital signal processing module, and the noise sample The module is connected to process the digital signal and extract a noise feature; the noise classification processing module is connected to the DSP digital signal processing module, and performs noise feature pattern matching on the extracted noise feature, and then The noise sample is classified and identified; the noise recognition visualization module is connected to the noise classification processing module, and the result of the classification and recognition of the noise sample is displayed. The system of claim 1, wherein the DSP digital signal processing module comprises: a preprocessing module, connected to the noise sample capturing module, performing noise reduction processing on the digital signal, providing information a feature extraction module, connected to the preprocessing module, selects and transforms the digital signal after the noise reduction process; the feature selection module is connected with the feature extraction module, and selects the selected and transformed The digital signal. The system of claim 2, wherein the noise classification processing module comprises: a classifier, connected to the feature selection module, and classifying the noise sample to establish a noise model library; noise classification and recognition The module is connected to the noise recognition visual module, and the digital signal extracted by the feature selection module is matched with the noise model library to further classify and identify the digital signal. The system of claim 2, wherein the noise reduction process comprises linear filtering, median filtering, and wavelet transform. The system of claim 2, wherein the feature extraction module extracts a feature vector of the digital signal by a mapping method. The system of claim 3, wherein the classifier is a DHMM hidden Markov classifier. The system of claim 1, wherein the noise sample capturing module comprises an FPGA multi-channel microphone array signal extractor, and the FPGA multi-channel microphone array signal extractor extracts the noise sample. The system of claim 7, wherein the noise sample capturing module further comprises a multi-channel analog digital converter connected to the FPGA multi-channel microphone array signal extractor, the multi-channel analog digital bit A converter converts the noise sample into the digital signal. The system of claim 8, wherein the noise sample capturing module further comprises a USB multi-channel microphone transmitter, respectively connected to the multi-channel analog digital converter and the DSP digital signal processing module. And encapsulating the digital signal into a packet, and transmitting the digital signal encapsulated into a packet form to the DSP digital signal processing module by using a USB protocol. In the system of claim 1, the noise recognition visualization module displays the result of the classification and identification of the noise sample in a graphical manner.