CN105118511A - Thunder identification method - Google Patents
Thunder identification method Download PDFInfo
- Publication number
- CN105118511A CN105118511A CN201510462768.5A CN201510462768A CN105118511A CN 105118511 A CN105118511 A CN 105118511A CN 201510462768 A CN201510462768 A CN 201510462768A CN 105118511 A CN105118511 A CN 105118511A
- Authority
- CN
- China
- Prior art keywords
- thunder
- power
- sound signal
- signal
- recognition methods
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention relates to a thunder identification method, including the steps of: performing framing on an originally input audio signal; converting each frame of audio signal to obtain an audio amplitude spectrum, calculating power of a lower frequency band, and adopting a double-threshold endpoint detection method to identify an audio effective segment; calculating power of an audio signal of the effective segment, and judging a relation between a proportion of the power of the lower frequency band a predetermined threshold value, if the proportion is larger than the predetermined threshold value, judging the audio signal to be a thunder signal, and otherwise judging the audio signal to be a noise signal. The thunder identification method provided by the invention can not only overcome interference of environmental noise, but can also be suitable for thunder signals of different forms, and satisfies requirements of a thunder three-dimensional positioning system in accuracy and real-time performance.
Description
Technical field
The present invention relates to a kind of thunder recognition methods.
Background technology
Thunder and lightning is one of occurring in nature major disaster affecting mankind's activity, not only can cause casualties, and causes immeasurable economic loss also can to the industries such as China's Aero-Space, electronics industry, petrochemical complex, traffic, forest.In recent years, because of the Frequent Accidents that thunder and lightning causes, personages of various circles of society are made more and more to pay attention to Real-Time Monitoring to thunder and lightning and protection.
The basis of lightning protection mitigation to the Real-Time Monitoring of thunder and lightning, domestic in last century late nineteen eighties to begin one's study lighting location technology, and in succession establish lightning location system in more than 30 provinces from last century so far the nineties, now realize national network.The lightning electromagnetic signal that the comprehensive multiple acquisition station of lightning location system obtains, after the unified process of central station, extrapolate time, position, amplitude of lightning current etc. that thunderbolt occurs in real time, positioning precision reaches 500m.The monitoring range at lightening detection station is 30 ~ 300km, namely in the short range of 30km, there will be detection blind area.Some lightning protection key area current such as transformer station, oil depot, militant post etc. need to carry out Real-Time Monitoring to the thunder and lightning in short range, and lightning location system is difficult to meet these active demands.
The voice signal that thunder three-dimensional localization is produced when being and being struck by lightning by microphone array detection or electromagnetic signal, and estimate the distance of thunder sound source and microphone array according to the mistiming of the two arriving signal microphone array, then estimate the elevation angle and the deflection of thunder sound source according to the mistiming that thunder signal arrives each microphone.The radius of investigation of thunder 3 D positioning system is 15km, and detection accuracy reaches 20 ~ 100m, can be used as supplementing of existing lightning location system, meets the requirement to emphasis lightning protection region lightening activity Real-Time Monitoring.When detecting thunder signal, needing to carry out real-time audio signal monitoring to paid close attention to region, and therefrom filtering out the valid data for thunder three-dimensional localization, namely needing to carry out thunder identification to primary monitoring data.
Inventionbroadly, thunder identification can be included into the category that text-independent " speaker " confirms, only need judge whether voice signal is thunder.In identifying, according to the feature of voice signal in template base, choosing the characteristic parameter with better distinction, then the parameter value of sound to be identified and template sound is carried out the matching analysis, identifying whether the sound into specifying " speaker " according to matching result.Thunder signal can change various along with the form of thunder and lightning, power and far and near difference, utilizes the features such as traditional waveform, frequency cannot meet the identification of strong randomness thunder; And as the speech recognition algorithms such as BP neural network, gauss hybrid models need through great amount of samples training, algorithm complex is high, and recognition time is longer, is not suitable for the thunder identification in thunder three-dimensional localization.So in order to adapt to small scale, more accurate thunder three-dimensional localization, needing to work out and a kind ofly quick, practical carry out thunder according to thunder signal characteristic and know method for distinguishing.
Summary of the invention
The object of this invention is to provide a kind of thunder recognition methods, the method can not only overcome the interference of neighbourhood noise, and can also be applicable to multi-form thunder signal, and in accuracy and real-time, meet the requirement of thunder 3 D positioning system.
To achieve these goals, the technical solution used in the present invention is as follows: a kind of thunder recognition methods, is characterized in that, comprise the following steps:
Framing is carried out to the sound signal of original input;
Every frame sound signal is converted respectively and obtains audio amplitude spectrum, calculate the audio power spectrum of low-frequency range and the audio power proportion of described low-frequency range;
Judge the relation between aforementioned proportion and predetermined threshold, if this ratio is greater than predetermined threshold, is then judged as thunder signal, otherwise is judged as noise signal.
According to above scheme, before the described sound signal to original input carries out framing step, comprise the steps: to carry out filtering to the sound signal of original input, and make normalized.
According to above scheme, described framing is carried out to sound signal after, also comprise the steps: to carry out windowing process to every frame sound signal.
According to above scheme, the described sound signal to original input carries out filtering, and concrete steps comprise: carry out low-pass filtering to the sound signal of original input, adopts the Hz noise in trap filters signals transmission simultaneously.
According to above scheme, described every frame sound signal conversion respectively obtains audio amplitude spectrum, and the method for employing is discrete Fourier transformation.
According to above scheme, before the audio power proportion step of described calculating low-frequency range, effective section identification is carried out to described sound signal, if there is effective section, then enters next step, if there is not effective section, be then judged as noise signal.
According to above scheme, described effective section of identification, comprises the steps: the power calculating audio power spectrum low-frequency range particularly, and adopts double threshold end-point detection method to it.
According to above scheme, described double threshold end-point detection method, comprises the steps: particularly to adopt adaptive noise Power estimation method to audio power spectrum, estimates Background Noise Power spectrum, determine the value of upper and lower thresholding accordingly again.
According to above scheme, described in estimate Background Noise Power spectrum, comprise the steps: particularly
First order recursive formulae discovery is adopted to go out the smooth power spectrum of described sound signal;
Obtain the power reference value in described smooth power spectrum;
Calculate the ratio of described smooth power spectrum and described power reference value;
The probability of the effective section existence of sound signal is calculated according to above-mentioned radiometer;
The smoothing factor of ground unrest is gone out according to above-mentioned probability calculation;
Background Noise Power spectrum is upgraded according to above-mentioned smoothing factor.
According to above scheme, the power reference value value in described smooth power spectrum is the power minimum of ART network in described smooth power spectrum.
Thunder recognition methods of the present invention, first sub-frame processing is carried out to sound signal, then the power of this sound signal low-frequency range is calculated, carry out effective section identification, extract this characteristic parameter of low-frequency range audio power proportion, threshold value identification is carried out to this characteristic parameter, by judging that this ratio reaches a certain scope and realizes the screening that thunder identification completes thunder data, compared to existing recognition method, not only go for the identification of high randomness thunder, and recognition time is short, precision is high, the interference of neighbourhood noise can be overcome, and multi-form thunder signal can also be applicable to, and in accuracy and real-time, meet the requirement of thunder 3 D positioning system.
Accompanying drawing explanation
Fig. 1 is schematic flow sheet of the present invention;
Fig. 2 is double threshold end-point detection method principle schematic in the present invention;
Fig. 3 is the calculation process schematic diagram of Background Noise Power spectrum in the present invention;
Fig. 4 is the one section of original sound signal audio frequency schematic diagram chosen in the present invention;
Fig. 5 is the audio power spectrum of voice signal in Fig. 4 and the Background Noise Power spectrum schematic diagram of estimation;
Fig. 6 is the schematic flow sheet of double threshold end-point detection method in the present invention;
Fig. 7 is thunder and noise frequency section power proportion probability density curve schematic diagram in the present invention.
Embodiment
Below in conjunction with accompanying drawing and embodiment, technical scheme of the present invention is described.
As shown in Figure 1, be the schematic flow sheet of a kind of thunder recognition methods of the present invention, framing is carried out to the sound signal of original input;
Every frame sound signal is converted respectively and obtains audio amplitude spectrum, calculate the audio power spectrum of low-frequency range and the audio power proportion of described low-frequency range;
Judge the relation between aforementioned proportion and predetermined threshold, if this ratio is greater than predetermined threshold, is then judged as thunder signal, otherwise is judged as noise signal.
Thunder recognition methods of the present invention, first sub-frame processing is carried out to sound signal, then the power of this sound signal low-frequency range is calculated, carry out effective section identification, extract this characteristic parameter of low-frequency range audio power proportion, threshold value identification is carried out to this characteristic parameter, by judging that this ratio reaches a certain scope and realizes the screening that thunder identification completes thunder data, compared to existing recognition method, not only go for the identification of high randomness thunder, and recognition time is short, precision is high, the interference of neighbourhood noise can be overcome, and multi-form thunder signal can also be applicable to, and in accuracy and real-time, meet the requirement of thunder 3 D positioning system.
By studying and testing discovery, thunder signal energy mainly concentrates on low-frequency range, utilizes this feature just can distinguish thunder signal and other voice signal rapidly.In the present invention, adopt voice signal low-frequency range power proportion as thunder recognition feature parameter, judge whether voice signal comprises thunder.
When carrying out initial stage pre-service to original audio signal, specifically comprising and filtering is carried out to the sound signal of original input, and making normalized, then suitable frame length and the framing of frame in-migration are chosen to sound signal.In order to eliminate the interference of high-frequency signal, taking low-pass filtering, adopting the interference of power frequency in 50Hz trapper elimination signals transmission simultaneously.After sound signal framing, in order to reduce the ground square signal discontinuity problem of frame starting and ending, need to carry out windowing process.
When processing every frame sound signal, specifically comprise and every frame sound signal is carried out discrete Fourier transformation respectively, to obtain the audio amplitude spectrum of this sound signal, then the audio power spectrum of its low-frequency range is calculated according to audio amplitude spectrum, obtain the power of low-frequency range, calculate low-frequency range power proportion.Due in the sound signal that receives, may not thunder be comprised, therefore need to carry out effective section identification to described sound signal, if there is effective section, then enter next step, if there is not effective section, then be judged as noise signal, can recognition time be saved like this, improve recognition efficiency.Concrete, the method that effective section identification adopts is double threshold end-point detection method.
Double threshold end-point detection method easily by the interference of ground unrest, in order to improve effective section recognition efficiency, adopts adaptive noise spectrum estimation method, estimates Background Noise Power spectrum, determines the value of upper and lower thresholding accordingly again.
In described adaptive noise spectrum estimation, comprise the steps: particularly to adopt first order recursive formulae discovery to go out the smooth power spectrum of described sound signal; Obtain the power reference value in described smooth power spectrum; Calculate the ratio of described smooth power spectrum and described power reference value; The probability of the effective section existence of sound signal is calculated according to above-mentioned radiometer; The smoothing factor of ground unrest is gone out according to above-mentioned probability calculation; Upgrade Background Noise Power spectrum according to above-mentioned smoothing factor, preferably, in described smooth power spectrum, power reference value value is the power minimum of ART network in described smooth power spectrum.
Below in conjunction with concrete identifying, the present invention is described in detail.
In thunder signal recognition method, sound-detection, signals collecting and data processing three steps need be completed, relate to several hardware tools.Wherein, sound-detection adopts high sensitivity, omnibearing Electret Condencer Microphone, and its response frequency is 15Hz-20kHz, and susceptibility is-60dB, is responsible for the voice signal in Real-Time Monitoring surrounding environment; Signals collecting adopts high-performance NI data collecting card, and sample frequency is 50kHz, is responsible for the simulating signal of Real-Time Monitoring to change into discrete digital signal; Data processing adopts the computing machine of configuration Matlab software, and the data be responsible for gathering process, and complete the identification of thunder signal.By signal transmssion line, the ambient sound simulating signal of microphone real-time detection is passed to NI data collecting card, forming frequency is 50kHz, time span is the digital signal sequences of 15s, then carries out thunder identification by Matlab software to it.
After receiving original sound signal, sound can be changed into frequency 50kHz by NI data collecting card, time span is the digital signal sequences of 15s, in order to reduce undesired signal, first need to adopt Butterworth filter to carry out low-pass filtering, and utilize 50Hz trapper to remove Hz noise, normalized, then chooses suitable frame length, the capable framing of frame shift-in, is then processed every frame sound signal by Matlab software.By carrying out discrete Fourier transformation to every frame sound signal, obtain band noise amplitude spectrum Y (λ, k) frequently, wherein λ is the frame number after sound signal framing, and k is the frequency range sequence number after Fourier transform, lower same.Adopt adaptive noise spectrum estimation method, estimate Background Noise Power spectrum N (λ, k); Calculate the power of band noise frequency and ground unrest low-frequency range, if Fig. 2 is in conjunction with the feature of thunder sign mutation, provide suitable upper and lower limit threshold value amp1, amp2, and adopt double threshold end-point detection method to carry out effective section identification to band noise frequency low-frequency range power; If input audio signal does not exist effective section, be then judged as noise signal; Otherwise, continue next step.Statistics thunder signal and noise signal low-frequency range power proportion, a given suitable threshold value significantly can distinguish thunder and other voice signal; Calculate the effective section audio signal low-frequency range power proportion detected, if this ratio is greater than given threshold value, be then judged as thunder signal, otherwise be noise signal.
In order to reduce sub-frame processing to the successional impact of sound signal, suitable frame need be chosen during framing and move, and adopt hamming window to carry out windowing to every frame signal, make power be relatively concentrated in main lobe, can comparatively close to real frequency spectrum.
In order to improve the end points recognition efficiency of double-threshold comparison method, the present invention gives chapter and verse and is with the low-frequency range power of noise frequency and ground unrest to determine upper and lower limit threshold value.Wherein, the adaptive noise estimation algorithm that the calculating that Background Noise Power is composed provides with reference to Sundarrajan, calculation flow chart as shown in Figure 3.In Matlab software, by carrying out discrete Fourier transformation to the every frame sound signal after framing, obtain band noise amplitude spectrum Y (λ frequently, k), then adopt first order recursive formulae discovery band noise smooth power spectrum P (λ, k) frequently, computing formula is: P
y(λ, k)=η P
y(λ-1, k)+(1-η) | Y (λ, k) |
2; Follow the trail of band noise smooth power spectral power minimum P frequently
min(λ, k); Calculate the ratio of the power minimum of band noise frequency smooth power spectrum and its tracking, computing formula is: S
r(λ, k)=P (λ, k)/P
min(λ, k); According to S
r(λ, k) calculates the band noise probability S that frequently, effective section exists
pp(λ, k); By S
pp(λ, k) substitutes into formula α
s(λ, k)=α
d+ (1-α
d) S
ppin (λ, k), calculate the smoothing factor of ground unrest, then according to this coefficient update Background Noise Power spectrum N (λ, k), computing formula is: N (λ, k)=α
s(λ, k) N (λ-1, k)+[1-α
s(λ, k)] × | Y (λ, k) |
2.In above computation process, α
dit is a constant.When carrying out above-mentioned tracking band noise frequency smooth power spectral power minimum and upgrading Background Noise Power spectrum, in order to complete the estimation to voice signal Background Noise Power spectrum more exactly, following 2 improvement are made.
1. P
minchoosing of (λ, k) initial value: in speech recognition process, it is generally acknowledged that the head of voice signal is quiet section, thus simply using the power of the 1st frame signal as P
min(λ, k) initial value.When to the simulating signal high speed acquisition of Real-Time Monitoring and to be divided into time span be the digital signal sequences of 15s time, effective section may appear at the head of voice signal, above-mentioned process will cause failing to judge, therefore chooses the minimum frame power of the former frame of voice signal in the present invention as P
min(λ, k) initial value.
2. Background Noise Power spectrum N (λ is increased, k) renewal judges: when upgrading Background Noise Power spectrum according to the formula in Sundarrajan, there will be Background Noise Power to be greater than same frame band and to make an uproar the situation of audio power, obviously this type of renewal does not meet actual conditions.Therefore the renewal increasing Background Noise Power spectrum judges: to make an uproar audio power if the Background Noise Power after upgrading is less than same frame band, then perform renewal rewards theory, otherwise by this frame band noise power as background noise power frequently.
After making above improvement, choose one section of voice signal and test, if Fig. 4 is original voice signal, the part that in figure, amplitude is outstanding is thunder signal, i.e. the effective section of voice signal.As the Background Noise Power spectrum schematic diagram that Fig. 5 is power spectrum and the estimation thereof frequently of band noise.
The first order recursive formula adopted when upgrading Background Noise Power spectrum can follow the tracks of the signal of slowly change preferably, therefore quiet section of voice signal, the Background Noise Power of estimation can reflect the change of actual noise power in voice signal more accurately; In the effective section of voice signal, the Background Noise Power of estimation also can make suitable adjustment along with the change of voice signal amplitude, but for having the thunder signal of Characteristics of Mutation, then can present higher signal to noise ratio (S/N ratio), as shown in Figures 4 and 5 in the thunderbolt generation moment.
Existing thunder result of study and a large amount of thunder Monitoring Data show, thunder energy mainly concentrates on low-frequency range, peak power is positioned at about 200Hz, and when adding up the ratio shared by thunder and noise signal low-frequency range power, low-frequency range value should comprise 200Hz.For containing in the effective section identifying of thunder voice signal, low-frequency range power according to band noise frequency and ground unrest determines suitable upper and lower limit threshold value, and to band noise low-frequency range power employing frequently double threshold end-point detection method, not only there is effective section of higher discrimination, and have thunder and well distinguish effect.
Utilize Matlab program to realize double threshold end-point detection method, its process flow diagram marks the possible state of sound by status as shown in Figure 6, and its value 0,1,2 represents mute state, state undetermined, effective status respectively; The length of effective section and quiet section is recorded respectively by count, silence, minsilence is the shortest quiet segment length, minlen is the shortest effective segment length, namely represent as silence=minsilence and enter mute state, meet count >=minlen and then can be judged as one section of useful signal.
The characteristic parameter that the present invention proposes using low-frequency range power proportion as thunder carries out thunder identification, and its threshold value obtains according to a large amount of calculating and statistics, as shown in Figure 7.As seen from the figure, this parameter has for thunder and noise and distinguishes effect preferably.In actual application, should according to the difference of emphasis, choose that suitable threshold value makes False Rate, misdetection rate controls in rational scope.
Thunder recognition methods proposed by the invention adopts Matlab Programming with Pascal Language to realize, shown by a large amount of Historical Monitoring data test, thunder discrimination >=90% of this thunder recognition methods, misdetection rate, False Rate≤10%, every segment data recognition time≤0.5s, accuracy and real-time meet the requirement of thunder 3 D positioning system.In test process, each section for the voice data frequency of testing be 50kHz, time span is 15s, the thunder hop count of reference is distinguished by artificial audiovisual and obtains.Following table 1 is the result that the voice data of section Real-Time Monitoring sometime carries out thunder identification test:
According to upper table, thunder recognition methods of the present invention, has discrimination high, and fail to judge low with False Rate, recognition time is short, and recognition efficiency is high, is applicable to thunder 3 D positioning system to the high standard requirement identifying accuracy and real-time.
Claims (10)
1. a thunder recognition methods, is characterized in that, comprises the following steps:
Framing is carried out to the sound signal of original input;
Every frame sound signal is converted respectively and obtains audio amplitude spectrum, calculate the audio power spectrum of low-frequency range and the audio power proportion of described low-frequency range;
Judge the relation between aforementioned proportion and predetermined threshold, if this ratio is greater than predetermined threshold, is then judged as thunder signal, otherwise is judged as noise signal.
2. thunder recognition methods according to claim 1, is characterized in that, before the described sound signal to original input carries out framing step, comprises the steps: to carry out filtering to the sound signal of original input, and makes normalized.
3. thunder recognition methods according to claim 1, is characterized in that, described framing is carried out to sound signal after, also comprise the steps: to carry out windowing process to every frame sound signal.
4. thunder recognition methods according to claim 2, it is characterized in that, the described sound signal to original input carries out filtering, and concrete steps comprise: carry out low-pass filtering to the sound signal of original input, adopts the Hz noise in trap filters signals transmission simultaneously.
5. thunder recognition methods according to claim 1, is characterized in that, described every frame sound signal conversion respectively obtains audio amplitude spectrum, and the method for employing is discrete Fourier transformation.
6. the thunder recognition methods according to any one of Claims 1 to 5, it is characterized in that, before the audio power proportion step of described calculating low-frequency range, effective section identification is carried out to described sound signal, if there is effective section, then enter next step, if there is not effective section, be then judged as noise signal.
7. thunder recognition methods according to claim 6, is characterized in that, carry out effective section identification to described audio power spectrum, the method for employing is double threshold end-point detection method.
8. thunder recognition methods according to claim 7, it is characterized in that, comprise the steps: to adopt adaptive noise Power estimation method to audio power spectrum, estimate Background Noise Power spectrum, determine the value of upper and lower thresholding in double threshold end-point detection method accordingly.
9. thunder recognition methods according to claim 8, is characterized in that, described in estimate Background Noise Power spectrum, comprise the steps: particularly
First order recursive formulae discovery is adopted to go out the smooth power spectrum of described sound signal;
Obtain the power reference value in described smooth power spectrum;
Calculate the ratio of described smooth power spectrum and described power reference value;
The probability of the effective section existence of sound signal is calculated according to above-mentioned radiometer;
The smoothing factor of ground unrest is gone out according to above-mentioned probability calculation;
Background Noise Power spectrum is upgraded according to above-mentioned smoothing factor.
10. thunder recognition methods according to claim 9, it is characterized in that, the determination of power reference value in described smooth power spectrum, specifically comprises following steps: if the power of present frame is less than the power reference value of former frame, then using the power reference value of the power of present frame as present frame; Otherwise, the power reference value basis of former frame adopts single order Tuning function to carry out ART network to the power reference value of present frame.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510462768.5A CN105118511A (en) | 2015-07-31 | 2015-07-31 | Thunder identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510462768.5A CN105118511A (en) | 2015-07-31 | 2015-07-31 | Thunder identification method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105118511A true CN105118511A (en) | 2015-12-02 |
Family
ID=54666474
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510462768.5A Pending CN105118511A (en) | 2015-07-31 | 2015-07-31 | Thunder identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105118511A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107192892A (en) * | 2017-06-16 | 2017-09-22 | 国网电力科学研究院武汉南瑞有限责任公司 | The automatic triggering method of thunder alignment system based on lightning electromagnetic signal identification technology |
| CN109829407A (en) * | 2019-01-23 | 2019-05-31 | 中国科学技术大学 | Intelligent lightening recognition method based on convolutional neural networks |
| CN110033757A (en) * | 2019-04-04 | 2019-07-19 | 行知技术有限公司 | A kind of voice recognizer |
| CN110782687A (en) * | 2019-08-30 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Data processing method and device and electronic equipment |
| CN112004133A (en) * | 2020-09-04 | 2020-11-27 | 成都极米科技股份有限公司 | Sound and picture synchronization method and device, projection equipment and readable storage medium |
| CN112017674A (en) * | 2020-08-04 | 2020-12-01 | 杭州联汇科技股份有限公司 | Method for detecting noise in broadcast audio signal based on audio features |
| CN112309419A (en) * | 2020-10-30 | 2021-02-02 | 浙江蓝鸽科技有限公司 | Noise reduction and output method and system for multi-channel audio |
| CN112687280A (en) * | 2020-12-25 | 2021-04-20 | 浙江弄潮儿智慧科技有限公司 | Biodiversity monitoring system with frequency spectrum-time space interface |
| CN112702551A (en) * | 2020-12-21 | 2021-04-23 | 上海眼控科技股份有限公司 | Lightning field video recording method, device, equipment and medium |
| CN113960375A (en) * | 2021-10-18 | 2022-01-21 | 特斯联科技集团有限公司 | Artificial intelligent lightning detection system and method for early warning of forest and grassland fire |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102509546A (en) * | 2011-11-11 | 2012-06-20 | 北京声迅电子股份有限公司 | Noise reduction and abnormal sound detection method applied to rail transit |
| CN102522082A (en) * | 2011-12-27 | 2012-06-27 | 重庆大学 | Recognizing and locating method for abnormal sound in public places |
| US20130246062A1 (en) * | 2012-03-19 | 2013-09-19 | Vocalzoom Systems Ltd. | System and Method for Robust Estimation and Tracking the Fundamental Frequency of Pseudo Periodic Signals in the Presence of Noise |
| CN104575513A (en) * | 2013-10-24 | 2015-04-29 | 展讯通信（上海）有限公司 | Burst noise processing system and burst noise detection and suppression method and device |
-
2015
- 2015-07-31 CN CN201510462768.5A patent/CN105118511A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102509546A (en) * | 2011-11-11 | 2012-06-20 | 北京声迅电子股份有限公司 | Noise reduction and abnormal sound detection method applied to rail transit |
| CN102522082A (en) * | 2011-12-27 | 2012-06-27 | 重庆大学 | Recognizing and locating method for abnormal sound in public places |
| US20130246062A1 (en) * | 2012-03-19 | 2013-09-19 | Vocalzoom Systems Ltd. | System and Method for Robust Estimation and Tracking the Fundamental Frequency of Pseudo Periodic Signals in the Presence of Noise |
| CN104575513A (en) * | 2013-10-24 | 2015-04-29 | 展讯通信（上海）有限公司 | Burst noise processing system and burst noise detection and suppression method and device |
Non-Patent Citations (1)
| Title |
|---|
| 卫克晶等: "一种雷声实时识别方法", 《实验室研究与探索》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107192892A (en) * | 2017-06-16 | 2017-09-22 | 国网电力科学研究院武汉南瑞有限责任公司 | The automatic triggering method of thunder alignment system based on lightning electromagnetic signal identification technology |
| CN107192892B (en) * | 2017-06-16 | 2019-06-18 | 国网电力科学研究院武汉南瑞有限责任公司 | Thunder positioning system automatic trigger method based on lightning electromagnetic signal identification technology |
| CN109829407A (en) * | 2019-01-23 | 2019-05-31 | 中国科学技术大学 | Intelligent lightening recognition method based on convolutional neural networks |
| CN110033757A (en) * | 2019-04-04 | 2019-07-19 | 行知技术有限公司 | A kind of voice recognizer |
| CN110782687A (en) * | 2019-08-30 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Data processing method and device and electronic equipment |
| CN110782687B (en) * | 2019-08-30 | 2022-07-19 | 腾讯科技（深圳）有限公司 | Data processing method and device and electronic equipment |
| CN112017674A (en) * | 2020-08-04 | 2020-12-01 | 杭州联汇科技股份有限公司 | Method for detecting noise in broadcast audio signal based on audio features |
| CN112017674B (en) * | 2020-08-04 | 2024-02-02 | 杭州联汇科技股份有限公司 | Method for detecting noise in broadcast audio signal based on audio characteristics |
| CN112004133A (en) * | 2020-09-04 | 2020-11-27 | 成都极米科技股份有限公司 | Sound and picture synchronization method and device, projection equipment and readable storage medium |
| CN112309419A (en) * | 2020-10-30 | 2021-02-02 | 浙江蓝鸽科技有限公司 | Noise reduction and output method and system for multi-channel audio |
| CN112702551A (en) * | 2020-12-21 | 2021-04-23 | 上海眼控科技股份有限公司 | Lightning field video recording method, device, equipment and medium |
| CN112687280A (en) * | 2020-12-25 | 2021-04-20 | 浙江弄潮儿智慧科技有限公司 | Biodiversity monitoring system with frequency spectrum-time space interface |
| CN112687280B (en) * | 2020-12-25 | 2023-09-12 | 浙江弄潮儿智慧科技有限公司 | Biodiversity monitoring system with frequency spectrum-time space interface |
| CN113960375A (en) * | 2021-10-18 | 2022-01-21 | 特斯联科技集团有限公司 | Artificial intelligent lightning detection system and method for early warning of forest and grassland fire |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105118511A (en) | Thunder identification method | |
| CN109357749B (en) | DNN algorithm-based power equipment audio signal analysis method | |
| CN109949823B (en) | DWPT-MFCC and GMM-based in-vehicle abnormal sound identification method | |
| CN102163427B (en) | Method for detecting audio exceptional event based on environmental model | |
| CN108896878B (en) | Partial discharge detection method based on ultrasonic waves | |
| CN102606891B (en) | Water leakage detector, water leakage detecting system and water leakage detecting method | |
| CN103646649A (en) | High-efficiency voice detecting method | |
| CN101430882B (en) | Method and apparatus for restraining wind noise | |
| CN103440872B (en) | The denoising method of transient state noise | |
| CN106373589B (en) | A kind of ears mixing voice separation method based on iteration structure | |
| CN106601230B (en) | Logistics sorting place name voice recognition method and system based on continuous Gaussian mixture HMM model and logistics sorting system | |
| CN107402131B (en) | A kind of bullet train kinematic parameter recognition methods based on rail side acoustic signal time-frequency crestal line | |
| CN104916289A (en) | Quick acoustic event detection method under vehicle-driving noise environment | |
| CN103646648B (en) | A kind of noise power estimation method | |
| CN101494049A (en) | Method for extracting audio characteristic parameter of audio monitoring system | |
| CN111060965A (en) | Seismic facies picking and event detection method based on convolutional neural network | |
| CN104167207A (en) | Equipment sound identification method based on transformer substation patrol inspection robot | |
| CN103278801A (en) | Noise imaging detection device and detection calculation method for transformer substation | |
| CN102760444A (en) | Support vector machine based classification method of base-band time-domain voice-frequency signal | |
| CN109300483A (en) | A kind of intelligent audio abnormal sound detection method | |
| CN106548786A (en) | A kind of detection method and system of voice data | |
| CN106504760B (en) | Broadband ambient noise and speech Separation detection system and method | |
| CN105825857A (en) | Voiceprint-recognition-based method for assisting deaf patient in determining sound type | |
| CN104103280A (en) | Dynamic time warping algorithm based voice activity detection method and device | |
| CN105916090A (en) | Hearing aid system based on intelligent speech recognition technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20151202 |
|
| RJ01 | Rejection of invention patent application after publication |