[go: up one dir, main page]

CN113657490B - Door and window silence detection method based on artificial intelligence - Google Patents

Door and window silence detection method based on artificial intelligence Download PDF

Info

Publication number
CN113657490B
CN113657490B CN202110939614.6A CN202110939614A CN113657490B CN 113657490 B CN113657490 B CN 113657490B CN 202110939614 A CN202110939614 A CN 202110939614A CN 113657490 B CN113657490 B CN 113657490B
Authority
CN
China
Prior art keywords
sound insulation
indexes
influence
index
performance
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110939614.6A
Other languages
Chinese (zh)
Other versions
CN113657490A (en
Inventor
张林志
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shuyang Yuanmei Decoration Material Co ltd
Original Assignee
Shuyang Yuanmei Decoration Material Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shuyang Yuanmei Decoration Material Co ltd filed Critical Shuyang Yuanmei Decoration Material Co ltd
Priority to CN202110939614.6A priority Critical patent/CN113657490B/en
Publication of CN113657490A publication Critical patent/CN113657490A/en
Application granted granted Critical
Publication of CN113657490B publication Critical patent/CN113657490B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/10—Noise analysis or noise optimisation
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Geometry (AREA)
  • Computer Hardware Design (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Software Systems (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Biophysics (AREA)
  • Architecture (AREA)
  • Civil Engineering (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Structural Engineering (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a door and window silence detection method based on artificial intelligence. The method comprises the steps of firstly obtaining influence indexes of a sound insulation material to be detected, wherein the influence indexes comprise filling rate, density and aperture ratio. Matching a batch performance index network according to the category of the sound insulation material to be detected; and inputting the influence indexes into a batch performance index network to obtain the predicted sound insulation performance indexes. And acquiring the predicted sound insulation performance index exceeding a preset threshold value, adjusting each influence index, and obtaining the adjustment trend of each influence index according to each adjusted influence index and the corresponding sound insulation performance index. The invention utilizes network training to process the sound insulation materials with similar performance in batch, thereby achieving the purpose of eliminating the inaccuracy of sampling detection.

Description

Door and window silence detection method based on artificial intelligence
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a door and window silence detection method based on artificial intelligence.
Background
With the development of economy, the requirements of people on the quality of life are higher and higher. In daily life, in addition to sound insulation of walls, sound insulation of doors and windows is also an important factor in trying to have a quiet sleeping environment or an environment where people do not disturb. In the medical field, when a medical institution performs a hearing test on a patient or a physical examiner, the hearing test is performed in a specific quiet environment, such as a soundproof room, so as to reduce the interference of noise in the surrounding environment to the test.
The conventional common door and window mute detection method is mainly based on sampling detection, wherein the sampling detection represents a sampling result and cannot represent a batch result, so that the result of the sampling detection has inaccuracy. If the accuracy of the silence detection result is to be improved, a plurality of samples need to be detected, but the detection speed is seriously affected due to too many samples, and moreover, the problems of over-fitting and under-fitting may occur, and the detection accuracy is reduced.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide a door and window silence detection method based on artificial intelligence, and the adopted technical scheme is as follows:
one embodiment of the invention provides a door and window silence detection method based on artificial intelligence, which comprises the following steps:
obtaining influence indexes of a sound insulation material to be detected, wherein the influence indexes comprise filling rate, density and aperture ratio;
matching a batch performance index network according to the category of the sound insulation material to be detected; inputting the influence indexes into the batch performance index network to obtain predicted sound insulation performance indexes;
acquiring predicted sound insulation performance indexes exceeding a preset threshold value, adjusting each influence index, and obtaining the adjustment trend of each influence index according to each adjusted influence index and the corresponding predicted sound insulation performance index;
the method for matching the batch performance index network according to the category of the sound insulation material to be detected comprises the following steps:
obtaining the influence indexes of multiple sound insulation materials, and obtaining multiple first sound insulation performance indexes according to the influence indexes;
obtaining the change interval of each influence index of the sound insulation material; obtaining a first affinity according to the similarity of the density change intervals of any two sound insulation materials; obtaining a second affinity according to the similarity of the aperture ratio change intervals of the two sound insulation materials; equally dividing intersection intervals of all the influence indexes to obtain a plurality of corresponding first sound insulation performance indexes, forming a high-dimensional vector by the plurality of first sound insulation performance indexes corresponding to each sound insulation material, and calculating cosine similarity of the high-dimensional vectors corresponding to any two sound insulation materials to serve as third affinity; obtaining an overall affinity between the two sound insulating materials according to the first affinity, the second affinity, and the third affinity;
and clustering the multiple sound insulation materials according to the total affinity to obtain multiple sound insulation material sets, and performing network training on each sound insulation material set to obtain a batch performance index network of each sound insulation material.
Preferably, the obtaining of the plurality of first sound insulation performance indexes according to the influence index includes:
establishing a filling rate equation according to a plurality of groups of different filling rates and corresponding actual sound insulation performance indexes in the sound insulation material to be detected, and obtaining filling rate performance through a least square method;
constructing a density equation according to a plurality of groups of different densities and corresponding actual sound insulation performance indexes in the sound insulation material to be detected, and obtaining density performance through a least square method;
constructing an opening rate equation according to a plurality of groups of different opening rates and corresponding actual sound insulation performance indexes in the sound insulation material to be detected, and obtaining the opening rate performance through a least square method;
constructing an initial sound insulation performance model according to the filling rate performance, the density performance and the opening rate performance;
obtaining a plurality of initial sound insulation performance indexes according to the influence indexes through the initial sound insulation performance model;
carrying out network training on a plurality of groups of influence indexes of each sound insulation material and initial sound insulation performance indexes corresponding to the influence indexes to obtain a first performance index network; each sound insulation material corresponds to a first performance index network;
inputting the filling rate performance, the density performance and the opening rate and outputting a first sound insulation performance index through the first performance index network.
Preferably, the intersection interval of the influence indexes includes:
the intersection of the density change intervals of any two sound-insulating materials, the intersection of the opening ratio change intervals and the filling ratio change interval of any one sound-insulating material.
Preferably, the equally dividing the intersection region of each influence index into a plurality of corresponding first sound insulation performance indexes includes:
carrying out mean value segmentation on the intersection intervals of the influence indexes to obtain a plurality of groups of segmentation influence indexes;
and inputting the plurality of groups of the segmentation influence indexes into the corresponding first performance index network to output a first sound insulation performance index.
Preferably, said obtaining the total affinity between the two sound-insulating materials according to the first affinity, the second affinity, and the third affinity includes:
and averaging the first affinity and the second affinity, and multiplying the average by the third affinity to obtain the total affinity.
Preferably, the clustering the plurality of sound insulation materials according to the total affinity comprises:
and clustering the plurality of sound insulation materials by taking the total affinity as a distance function of clustering, wherein the circumferential radius of the clustering is a preset radius.
Preferably, the preset threshold includes:
the preset threshold is an actual sound insulation performance index obtained through actual measurement.
Preferably, the obtaining of the predicted sound insulation performance index exceeding the preset threshold, adjusting each of the influence indexes, and obtaining an adjustment trend of each of the influence indexes according to each of the adjusted influence indexes and the corresponding predicted sound insulation performance index include:
when the predicted sound insulation performance index exceeds the preset threshold value, adjusting the density according to the density change interval, and obtaining a predicted sound insulation performance index corresponding to the density change interval; acquiring a plurality of groups of adjusted densities and corresponding predicted sound insulation performance indexes thereof, and performing straight line fitting on the plurality of groups of adjusted densities and corresponding predicted sound insulation performance indexes thereof; and taking the straight line obtained by fitting as the regulation trend to regulate the density.
Preferably, the obtaining of the predicted sound insulation performance index exceeding the preset threshold, adjusting each of the influence indexes, and obtaining an adjustment trend of each of the influence indexes according to each of the adjusted influence indexes and the corresponding predicted sound insulation performance index, further includes:
when the predicted sound insulation performance index exceeds the preset threshold value, adjusting the opening rate according to the opening rate change interval, and obtaining a predicted sound insulation performance index corresponding to the opening rate change interval; acquiring multiple groups of adjusted opening rates and corresponding predicted sound insulation performance indexes thereof, and performing straight line fitting on the multiple groups of adjusted opening rates and corresponding predicted sound insulation performance indexes; and taking the straight line obtained by fitting as an adjusting trend to adjust the aperture ratio.
The embodiment of the invention at least has the following beneficial effects:
according to the embodiment of the invention, an artificial intelligence technology is utilized, the initial sound insulation performance index is obtained according to the three influence indexes of the filling rate, the density and the opening rate of the sound insulation material, the filling rate, the density, the opening rate and the initial sound insulation performance index are subjected to network training to obtain the first performance index network, and the accuracy of the detection result, namely the sound insulation performance index is improved. The total affinity among different influence indexes of different sound insulation materials is calculated, different sound insulation materials are clustered according to the total affinity to obtain sound insulation material sets, network training is carried out on each group of sound insulation material sets to obtain a batch performance index network, and the networking training is carried out on each group of clustered sound insulation material sets, so that the generalization capability of a detection result, namely a sound insulation performance index, is improved. And obtaining a predicted sound insulation performance index exceeding the actual sound insulation performance index, adjusting each influence index, and fitting each adjusted influence index and the corresponding sound insulation performance index to obtain the adjustment trend of each influence index. The predicted sound insulation performance indexes obtained by the batch performance index network avoid the problems of over-fitting and under-fitting, improve the accuracy of a silence detection result, and can adjust the influence indexes according to the adjustment trend to obtain better influence indexes.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of a method for detecting door/window silence based on artificial intelligence according to an embodiment of the present invention;
fig. 2 is a scene diagram of a door and window silence detection method based on artificial intelligence according to an embodiment of the present invention;
FIG. 3 is a schematic illustration of a density and sound insulation performance index scatter-point fit according to an embodiment of the present invention;
fig. 4 is a graphical illustration of a scatter fit of the aperture ratio and sound insulation performance indicators in accordance with an embodiment of the present invention.
Detailed Description
To further illustrate the technical means and effects of the present invention adopted to achieve the predetermined purpose, the following detailed description will be given to the specific implementation, structure, features and effects of the artificial intelligence based door/window mute detection method according to the present invention with reference to the accompanying drawings and the preferred embodiments. In the following description, different "one embodiment" or "another embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The embodiment of the invention provides a specific implementation method of a door and window silence detection method based on artificial intelligence, the method is suitable for a door and window silence detection scene, please refer to fig. 2, the detection is carried out in a laboratory and is composed of two adjacent reverberation rooms, wherein one reverberation room is a sound source room, the other reverberation room is a receiving room, a test piece is placed between the two reverberation rooms, the test piece is an outer frame for placing a filling material, and a sound insulation material to be detected is placed inside the test piece. The detection equipment comprises a sound source system and a receiving system: the sound source system is formed by sequentially connecting a white noise or pink noise generator, an 1/3 octave filter, a power amplifier and a loudspeaker, wherein the white noise or pink noise generator is used for simulating and outputting noise, and the noise is sequentially filtered and amplified by the filter and the power amplifier and then is output and processed by the loudspeaker to form simulated noise; the receiving system is formed by sequentially connecting a microphone, an amplifier, an 1/3 octave analyzer and a recording instrument, the recording instrument receives and records output noise, the output noise is the output noise after analog noise received by the microphone is amplified and analyzed by the amplifier and the 1/3 octave analyzer in sequence and is recorded and output by the recording instrument, and the recording instrument is a meter head or a recorder. In the embodiment of the invention, the concrete discussion is that the sound insulation material filled in the door and window, such as rigid foam plastic, sound insulation felt, fiberboard and other materials, and the discussion is not about the material of the door and window outer frame, such as plastic steel, bridge-cut aluminum and the like, because the outer frame is a rigid body, the efficiency of transmitting noise is higher. In order to solve the under-fitting problem of door and window silence detection by sampling and the over-fitting problem caused by detection of each sample, the embodiment of the invention utilizes network training to process sound insulation materials with similar performance in batches, thereby achieving the purpose of eliminating the inaccuracy of sampling detection.
The following describes a specific scheme of the door and window silence detection method based on artificial intelligence in detail with reference to the accompanying drawings.
Referring to fig. 1, a flowchart illustrating steps of a method for detecting silence in a door or window based on artificial intelligence according to an embodiment of the present invention is shown, where the method includes the following steps:
and S100, acquiring influence indexes of the sound insulation material to be detected, wherein the influence indexes comprise filling rate, density and aperture ratio.
Obtaining a soundproofing material M to be tested1Fill fraction, density and open cell fraction.
First, a reference test window W is determined0The perimeter, the cross section area and the thickness of the outer frame profile are multiplied by the thickness and the cross section area to obtain a reference volume V of the outer frame0Reference detection window W0Namely, the test piece in the laboratory of fig. 2, and the sound insulating material to be tested is the reference test window W0An inner filled sound insulating material. It should be noted that, in the following description,the volume of the outer frame is calculated because the sound insulating material is filled or placed inside the outer frame, and the calculated volume of the outer frame is the volume of the space which needs to be filled for calculating the sound insulating material.
Sound insulation material M based on to be detected1The test was conducted under different filling ratios F, densities D, and opening ratios C.
For the sound insulating material M to be detected1In the test of (2), it is assumed that there are n different filling ratios { F }1,…Fi…FnIn which the filling rate FiFor the sound-insulating material M to be tested1At different filling volumes ViUnderfilling reference volume V0Ratio of outer frame (c):
Figure BDA0003214292240000051
in the present embodiment, an AccuPyc series true density instrument was used to obtain the porosity and density of the soundproofing material to be tested.
Step S200, matching a batch performance index network according to the category of the sound insulation material to be detected; and inputting the influence indexes into a batch performance index network to obtain the predicted sound insulation performance indexes.
In order to obtain the influence of the parameter changes of the three influencing indexes of the filling rate F, the density D and the opening rate C on the sound insulation performance index, a benchmark test is first performed. The purpose is to obtain the sound-insulating performance index of each sound-insulating material under a certain size, and further to normalize the sound-insulating performance index of each sound-insulating material under a unit size under the condition that the size is known. It should be noted that the size is three influencing indexes of filling rate, density and aperture ratio.
The sound insulation performance index here is defined as: on the premise of forming a laboratory by a sound source room and a receiving room, the sound insulation performance index is determined by providing a noise source in the sound source room and comparing the sound pressure ratio in two scenes with a filled sound insulation material and a non-filled sound insulation material
Figure BDA0003214292240000052
Wherein p is0Is a reference test window W0Sound pressure without filling sound insulating material inside, p1Is a reference test window W0Sound pressure when filled with sound insulating material. Wherein, the larger the soundproofing performance index P is, the better the soundproofing performance of the soundproofing material to be detected is.
The reference filling rate F is determined by the implementer0Reference density D0Reference opening ratio C0And obtaining a reference sound insulation performance index P according to the three reference influence indexes0The standard sound insulation performance index is at a standard filling rate F0Reference density D0Reference opening ratio C0Under the conditions, the actual sound insulation performance index obtained by the laboratory.
(1) An initial sound insulation performance model was constructed.
Obtaining n different filling rates { F }1,…,FnCalculating n different actual sound insulation performance indexes corresponding to n different filling rates under the condition of not changing the density D and the opening rate C
Figure BDA0003214292240000053
Obtaining the contrast relation between the filling density and the sound insulation performance index
Figure BDA0003214292240000054
And constructing a rectangular coordinate system with the filling rate as a horizontal axis and the actual sound insulation performance index as a vertical axis, and making corresponding three points according to the comparison relation. Fitting the scattered points to obtain a first-order equation YF(F) kF + c as a fill factor equation, and determining fill factor performance by least squares
Figure BDA0003214292240000061
Wherein,
Figure BDA0003214292240000062
that is, since the change of the filling factor F is linearly changed according to the sound insulation performance index when the density D and the opening factor C are not changed, the reference filling factor F can be specified under the condition of the fixed density and the fixed opening factor0The larger theThe larger the sound insulation performance index, and the effect on performance can be scaled based on different fill rates.
The implementer selects a filling rate F, a density D and an opening rate C as a reference filling rate F based on engineering experience0Reference density D0Reference opening ratio C0. Filling rate F based on the filling rate0The change in the sound insulation performance index when the density D and the opening ratio C were changed was recorded.
Controlling the reference filling factor F0Reference opening ratio C0Selecting n different densities { D }1,…,Dn} of the soundproofing material M to be tested1Testing to obtain n different actual sound insulation performance indexes corresponding to different densities
Figure BDA0003214292240000063
Obtaining the comparison relation between the density and the sound insulation performance index
Figure BDA0003214292240000064
A rectangular coordinate system with the density as the horizontal axis and the actual sound insulation performance index as the vertical axis is constructed, and a corresponding scatter diagram is made according to the comparison relation, and the reference is made to fig. 3. Fitting the scattered points to obtain a multiple equation YD(D)＝k0+k1D+k2D2+k3D3+k4D4+k5D5+k6D6+k7D7Determining density properties by least squares as a density equation
Figure BDA0003214292240000065
In the embodiment of the present invention, the multiple equation may also be referred to as 7 th degree polynomial, which is used for fitting the nonlinear mapping relationship, the highest 7 is an empirical value, and in other embodiments, the implementer may perform the inspection and adjustment according to the specific practical situation.
Controlling the reference filling factor F0Reference density D0Selecting n different opening rates { C1,…,CnH.a sound insulating material M to be detected1Testing to obtain n different sound insulation performance indexes (P) corresponding to different aperture ratiosC1,…,PCnObtaining the contrast relation of the opening rate and the sound insulation performance index { (C)1,PC1),…,(Cn,PCn)}. A rectangular coordinate system with the opening ratio as a horizontal axis and the actual sound insulation performance index as a vertical axis is constructed, and a corresponding scatter diagram is made according to the comparison relation, which is shown in fig. 4. Fitting a multiple equation Y to the scatter pointsC(C)＝k0+k1C+k2C2+k3C3+k4C4+k5C5+k6C6+k7C7Making an aperture ratio equation, and determining the aperture ratio performance by the least square method
Figure BDA0003214292240000066
In the embodiment of the present invention, the multiple equation may also be referred to as 7 th degree polynomial, which is used for fitting the nonlinear mapping relationship, the highest 7 is an empirical value, and in other embodiments, the implementer may perform the inspection and adjustment according to the specific practical situation.
In the impact index of the soundproof material, the impact on the soundproofing performance index between the density and the open area ratio is independent of each other, and therefore the density performance YD(D) And open porosity Performance YC(C) The two dimensions which are not related to each other are density and open area ratio.
Performance according to fill ratio YF(F) Density property YD(D) And open porosity Performance YC(C) And constructing an initial sound insulation performance model, and obtaining a plurality of initial sound insulation performance indexes according to each influence index through the initial sound insulation performance model.
The initial sound insulation performance model of the material is as follows:
Figure BDA0003214292240000071
wherein, YF(F) Is the fill rate capability; y isD(D) Density property;YC(C) the porosity performance; p (F, D, C) is an initial sound insulation performance index.
Due to the fact that
Figure BDA0003214292240000072
And
Figure BDA0003214292240000073
so Y in the initial sound-insulating performance modelD(D) I.e. the density performance mapped by different densities D under the reference parameter
Figure BDA0003214292240000074
Y in the initial sound-insulating performance modelC(C) I.e. the aperture ratio performance mapped by different aperture ratios C under the reference parameter
Figure BDA0003214292240000075
For rough estimation, density performance is measured
Figure BDA0003214292240000076
And open cell content performance
Figure BDA0003214292240000077
Performing weighted sum calculation, density performance in the embodiment of the invention
Figure BDA0003214292240000078
And open cell content performance
Figure BDA0003214292240000079
Are each 0.5.
And repeatedly constructing the initial sound insulation performance model of the materials for a plurality of sound insulation materials, namely, each sound insulation material corresponds to one initial sound insulation performance model of the material, and establishing an initial sound insulation performance database of different sound insulation materials.
(2) A method for constructing a batch performance index network.
Since the initial sound insulation performance model may contain a polynomial overfitting phenomenon, the overfitting phenomenon is mainly caused by the fact that a large amount of data is generated due to the fact that the ranges of the density D and the opening ratio C of different sound insulation materials are different when the collection of indexes is affected. In order to eliminate the influence of overfitting on the precision, a deep neural network is constructed to improve the generalization capability.
Sound insulating material M to be detectediCarrying out deep neural network training, i.e. first performance indicator network NiAnd (4) training. The number of layers of the deep neural network is 4, wherein the deep neural network comprises a full connection layer, two hidden layers which are also called hidden layers and an output layer.
And (3) arranging and combining the filling rate F, the density D and the opening rate C in the performance database, and substituting the arranged and combined filling rate F, density D and opening rate C into the initial sound insulation performance model of the material to obtain a corresponding initial sound insulation performance index P (F, D, C).
The filling factor F, density D, open porosity C and the corresponding initial sound insulation performance indicators P (F, D, C) were used as a training data set. And (3) sending the filling rate F, the density D and the opening rate C into a full connection layer, wherein the deep neural network comprises a plurality of hidden layers, the specific number of layers is not excessive, and the number of the hidden layers is 2 in the embodiment of the invention. The output layer has only one neuron, and the first performance index network NiOutputting a first sound insulation Performance index P2. And (3) supervising a first performance index network by adopting an L1 norm, wherein the supervised label is an initial sound insulation performance index P (F, D, C) obtained by filling rate F, density D and opening rate C, and the loss function of the first performance index network is a mean square error loss function. After training is finished, the first performance index network NiIs close to the result of the initial soundproofing performance index P (F, D, C) of the material.
The batch of training data is increased by using the deep neural network, so that the model learned by the network can ignore abnormal overfitting results as much as possible.
The first performance indicator network is repeatedly constructed for a plurality of sound insulating materials, i.e., each sound insulating material corresponds to one first performance indicator network.
Determining the similarity of the influence indexes between any two sound insulation materials, testing the performance of the interactive deep neural network between the two sound insulation materials, namely the interactive first performance index network, and further determining the affinity between different materials.
In particular toSelecting the ith sound insulating material MiAnd the jth soundproof material MjObtaining a sound insulating material MiAnd a sound insulating material MjCorresponding first performance indicator network NiAnd Nj。
Separately determining sound-insulating material MiAnd sound insulating material MjThe change interval of each influence index in the performance database of the sound-insulating material.
Density variation interval of density D: di∈[Di,min,Di,max]，Dj∈[Dj,min,Dj,max]；
Open area ratio change interval of open area ratio C: ci∈[Ci,min,Ci,max],Cj∈[Cj,min,Cj,max]
Wherein D isiRepresents a sound insulating material MiA corresponding density variation interval; djRepresents a sound insulating material MjA corresponding density variation interval; ciRepresents a sound insulating material MiCorresponding aperture ratio change interval; cjRepresents the sound insulating material MjCorresponding aperture ratio change interval.
According to the sound insulating material MiAnd sound insulating material MjThe first affinity is obtained by the similarity of the density change interval. In the embodiment of the invention, the similarity obtaining method is to calculate the intersection ratio of the density change intervals of the two sound insulation materials, and the intersection ratio is a concrete implementation means of similarity or coincidence degree.
Specifically, the numerator of the first affinity calculation formula is two sound insulation materials MiAnd MjThe size of the intersection of the density intervals of (a), the denominator being two sound-insulating materials MiAnd MjThe size of the union of the density intervals.
The first affinity A1The calculation formula of (2) is as follows:
Figure BDA0003214292240000081
wherein D isi,maxIs a sound insulating material MiThe maximum density of (d); dj,maxIs a sound insulating material MjThe maximum density of (d); di,minIs a sound insulating material Mi(ii) a minimum density of; dj,minIs a sound insulating material MjThe minimum density of (c).
The first affinity A1For describing the similarity of the soundproof materials in the density region, 1 is completely similar. In the present embodiment, if A1<0, then let A1＝0。
According to the sound insulating material MiAnd sound insulating material MjThe second affinity is obtained from the similarity of the aperture ratio change section(s). In the embodiment of the present invention, the similarity is obtained by calculating the intersection ratio of the variation intervals of the opening ratio of the two sound insulating materials.
Specifically, the numerator of the second affinity calculation formula is two sound insulation materials MiAnd MjThe size of the intersection of the open pore ratio intervals of (a), the denominator being two sound insulating materials MiAnd MjThe size of the union of the opening ratio sections (c).
The second affinity A2The calculation formula of (2) is as follows:
Figure BDA0003214292240000091
wherein, Ci,maxIs a sound insulating material MiThe maximum open pore ratio of (2); cj,maxIs a sound insulating material MjThe maximum open cell content of (a); ci,minIs a sound insulating material MiMinimum open cell content of (a); cj,minIs a sound insulating material MjThe minimum open cell content of (a).
The second affinity A2For describing the similarity of the sound-proofing material in the open-cell ratio region, 1 is completely similar. In the present embodiment, if A2<0, then let A2＝0。
Calculating the third affinity A3Denotes a sound insulating material MiThe density D and the open porosity C in the sound insulating material MjCorresponding first performance indicator network NjThe applicability of the data of (1).
Obtaining a Sound insulating Material MiCorresponding filling rate F in the performance databaseiThe interval of (a): fi∈[Fi,min,Fi,max]. And separately determining the sound insulating material MiAnd sound insulating material MjAnd density intersection region of (a), and sound insulating material MiAnd sound insulating material MjAperture ratio intersection interval of (a):
sound insulating material MiAnd MjThe density intersection interval of (a) is: dij∈[max(Di,min,Dj,min),min(Di,max,Dj,max]。
Sound insulating material MiAnd MjThe intersection interval of the opening rates is: cij∈[max(Ci,min,Cj,min)，min(Ci,max,Cj,max)]。
Respectively for the filling rate interval FiDensity intersection region DijAnd the intersection region of the opening ratio CijPerforming mean value cutting, uniformly cutting out Q value-taking points, and obtaining the value by the corresponding first performance index network NiAnd NjRespectively obtaining Q first sound insulation performance indexes P2. In the embodiment of the invention, the value of the value taking point is 1000, the specific value of the value taking point can be adjusted by experience of an implementer, and is preferably selected from 500-1000, because too large value can cause too slow calculation speed, and too small value can cause reduced affinity calculation precision.
Two first performance indicator networks NiAnd NjRespectively outputting the results as two high-dimensional vectors, calculating cosine similarity of the two high-dimensional vectors, and taking the obtained cosine similarity as a third affinity A3. It should be noted that the cosine similarity is normalized to the interval [0,1 ]]The similarity value of (a).
The first affinity and the second affinity are averaged, and the average and the third affinity are multiplied to obtain the total affinity.
Overall affinity AijThe calculation formula of (2) is as follows:
Figure BDA0003214292240000092
wherein A is1Is a first affinity; a. the2Is a second affinity; a. the3Is the third affinity.
It should be noted that, in the attribute of the sound-insulating material, if the value intervals of the density D and the open porosity C of the two sound-insulating materials are similar in the implementation process, it is considered that the two sound-insulating materials may be regarded as a type of sound-insulating material in the subsequent prediction, and because the influences of the density and the open porosity of the sound-insulating material on the sound-insulating performance index are independent of each other, the two sound-insulating materials will be regarded as a type of sound-insulating material
Figure BDA0003214292240000101
The term is used as the scaling coefficient of the cosine similarity, and the term value range is [0,1 ]]。
The total affinity between all the sound-insulating materials was calculated.
According to the influence index group M corresponding to various sound insulation materials by the total affinityi:{Fi,Di,CiCarry out DBSCAN clustering. Wherein, the total affinity between the two sound insulation materials is taken as a distance function of clustering, the circumferential radius of the clustering is a preset radius, and the value interval of the preset radius is [0,1 ]]. In the embodiment of the present invention, the preset radius is 0.2, and in other embodiments, an implementer may adjust the preset radius according to an actual situation to ensure that two sound insulation materials expected in a clustering result are located in the same clustering category.
Finally obtaining a plurality of types of sound insulation material sets through DBSCAN clustering: { G1,…GN}。
In order to improve the generalization ability, N groups of sound-insulating material are combined { G1,…GNCarry out N deep neural networks
Figure BDA0003214292240000102
The N deep neural networks
Figure BDA0003214292240000103
That is, the batch performance index network, the number of layers of the deep neural network is 4, which comprisesA fully connected layer, two hidden layers also called hidden layers and an output layer.
Wherein the i-th group of sound insulating materials is combined with the GiFor example, in the property database, the sound insulating material set G is collectediThe influence indexes { F, D, C } corresponding to each sound-insulating material and the first sound-insulating performance indexes P corresponding to the groups of influence indexes2And combining the sound insulation materials into the same set, and using the set as a training data set of the sound insulation material set.
The influence indexes { F, D and C } corresponding to each sound insulation material are input into the full connection layer, the deep neural network comprises a plurality of hidden layers, the specific number of layers is not too large, and the number of the hidden layers is 2 in the embodiment of the invention. The output layer has only one neuron and outputs the predicted sound insulation performance index PpredMonitoring a batch performance index network by adopting an L1 norm, wherein a monitoring label is a first sound insulation performance index P obtained by filling rate F, density D and aperture ratio C2The loss function of the batch performance indicator network is a mean square error loss function. After training is completed, batch performance index network
Figure BDA0003214292240000104
Is close to the result of the first performance indicator network.
And repeatedly obtaining batch performance index networks corresponding to the multiple sound insulation material sets, namely, each sound insulation material set corresponds to one batch performance index network.
And step S300, acquiring the predicted sound insulation performance indexes exceeding the preset threshold value, adjusting each influence index, and obtaining the adjustment trend of each influence index according to each adjusted influence index and the corresponding predicted sound insulation performance index.
A known density D and open porosity C were obtained. Determining a test window W of a certain sizexThe perimeter, the cross section area and the thickness of the outer frame section are multiplied by the cross section area to obtain the volume V of the outer framexCalculating the test window W from the volume of the outer frame and the reference volumexThe corresponding filling factor F. Specifically, the calculation method of the filling ratio F and the acquisition method of the density and the open-cell ratio are given in step S100.
For ease of understanding, assume a test window WxThe sound-insulating material used was M1And a sound insulating material M1Set G of sound insulating material1Then the fill factor F, density D and open area C are input into the batch Performance indicator network at this time
Figure BDA0003214292240000111
Output predicted sound insulation performance index
Figure BDA0003214292240000112
Obtaining a predicted soundproofing performance index P exceeding a preset thresholdpredThe preset threshold value is the actual measured sound insulation material M1Corresponding actual sound insulation performance index PD1。
If the sound insulation performance index P is predictedpredExceeds the actual sound insulation performance index PD1I.e. the actual soundproofing performance index PD1Does not exceed the predicted sound insulation performance index PpredIt is indicated that the actual soundproofing performance index needs to be adjusted by changing the filling rate F, the density D and the opening rate C.
Thus, in addition to the practitioner empirically determining the variation intervals for density D and open cell content C, a batch performance indicator network is further used
Figure BDA0003214292240000113
And (6) performing prediction. Adjusting the density [ D ] in certain stepsi,min,…,Di,max]And obtaining a predicted soundproofing performance index corresponding thereto
Figure BDA0003214292240000114
Thereby obtaining a plurality of predicted sample groups
Figure BDA0003214292240000115
Constructing a rectangular coordinate system with the density as a horizontal axis and the prediction sound insulation performance index as a vertical axis, making corresponding scattered points according to a plurality of predicted sample groups, and performing linear fitting on the scattered points to obtain a straight line Plinear(D) An inclination corresponding to the straight lineRate KDA straight line P obtained by fittinglinear(D) As a trend of the adjustment, the density was adjusted. The change section of the density D is determined empirically by the practitioner, and this density change section is defined as a local linear section.
Taking the straight line obtained by fitting as the trend of density adjustment, specifically, when KD>0 indicates that the higher the density, the better the sound-insulating performance, whereas the higher the density, the worse the sound-insulating performance.
Adjusting the opening ratio [ C ] in a certain step lengthi,min,…,Ci,max]And obtaining a predicted soundproofing performance index corresponding thereto
Figure BDA0003214292240000116
Thereby obtaining a plurality of predicted sample groups
Figure BDA0003214292240000117
Making corresponding scattered points according to the plurality of predicted sample groups, and performing linear fitting on the scattered points to obtain a straight line Plinear(C) A slope K corresponding to the straight lineCA straight line P obtained by fittinglinear(C) As a tendency of the adjustment, the porosity was adjusted. The operator empirically determines the variation interval of the opening ratio C, and this opening ratio variation interval is defined as a local linear interval.
Taking the fitted straight line as the trend of adjusting the aperture ratio, specifically, when K isC>When 0, the sound-insulating property is considered to be better as the open porosity is larger, whereas when the open porosity is larger, the sound-insulating property is considered to be worse.
Since the filling factor F itself is linearly increased, the filling factor F can be directly increased to offset in order to significantly achieve the expected performance, and thus the detailed steps are not described again.
In summary, the embodiment of the present invention utilizes an artificial intelligence technique, obtains an initial sound insulation performance index according to three influence indexes, i.e., the filling rate, the density, and the aperture ratio, of the sound insulation material, performs network training on the filling rate, the density, the aperture ratio, and the initial sound insulation performance index to obtain a first performance index network, and inputs the filling rate, the density, and the aperture ratio to output the first sound insulation performance index from the first performance index network. Calculating the total affinity among different influence indexes of different sound insulation materials, clustering the different sound insulation materials according to the total affinity to obtain sound insulation material sets, carrying out network training on each group of sound insulation material sets to obtain a batch performance index network, and inputting the filling rate, the density and the aperture ratio to output the predicted sound insulation performance index through the batch performance index network. And acquiring a predicted sound insulation performance index exceeding the actual sound insulation performance index, adjusting each influence index, and fitting each adjusted influence index with the corresponding sound insulation performance index to obtain the adjustment trend of each influence index.
It should be noted that: the precedence order of the above embodiments of the present invention is only for description, and does not represent the merits of the embodiments. And specific embodiments thereof have been described above. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
All the embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from other embodiments.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (5)

1. A door and window silence detection method based on artificial intelligence is characterized by comprising the following steps:
obtaining influence indexes of a sound insulation material to be detected, wherein the influence indexes comprise filling rate, density and aperture ratio;
matching a batch performance index network according to the category of the sound insulation material to be detected; inputting the influence indexes into the batch performance index network to obtain predicted sound insulation performance indexes;
acquiring predicted sound insulation performance indexes exceeding a preset threshold value, adjusting each influence index, and obtaining the adjustment trend of each influence index according to each adjusted influence index and the corresponding predicted sound insulation performance index;
the method for matching the batch performance index network according to the category of the sound insulation material to be detected comprises the following steps:
obtaining the influence indexes of multiple sound insulation materials, and obtaining multiple first sound insulation performance indexes according to the influence indexes;
obtaining the change interval of each influence index of the sound insulation material; obtaining a first affinity according to the similarity of density change intervals of any two sound insulation materials; obtaining a second affinity according to the similarity of the aperture ratio change intervals of the two sound insulation materials; equally dividing intersection intervals of all the influence indexes to obtain a plurality of corresponding first sound insulation performance indexes, forming a high-dimensional vector by the plurality of first sound insulation performance indexes corresponding to each sound insulation material, and calculating cosine similarity of the high-dimensional vectors corresponding to any two sound insulation materials to serve as third affinity; obtaining an overall affinity between the two sound insulating materials according to the first affinity, the second affinity, and the third affinity;
clustering a plurality of sound insulation materials according to the total affinity to obtain a plurality of sound insulation material sets, and performing network training on each sound insulation material set to obtain a batch performance index network of each sound insulation material;
wherein the intersection interval is: the intersection of the density change intervals, the intersection of the opening rate change intervals and the filling rate change interval of any two sound insulation materials;
wherein the process of obtaining a plurality of first sound insulation performance indicators from the impact indicators is: establishing a filling rate equation according to a plurality of groups of different filling rates and corresponding actual sound insulation performance indexes in the sound insulation material to be detected, and obtaining filling rate performance through a least square method; constructing a density equation according to a plurality of groups of different densities and corresponding actual sound insulation performance indexes in the sound insulation material to be detected, and obtaining density performance through a least square method; constructing an opening rate equation according to a plurality of groups of different opening rates and corresponding actual sound insulation performance indexes in the sound insulation material to be detected, and obtaining the opening rate performance through a least square method; constructing an initial sound insulation performance model according to the filling rate performance, the density performance and the opening rate performance; obtaining a plurality of initial sound insulation performance indexes according to the influence indexes through the initial sound insulation performance model; carrying out network training on a plurality of groups of influence indexes of each sound insulation material and initial sound insulation performance indexes corresponding to the influence indexes to obtain a first performance index network; each sound insulation material corresponds to a first performance index network; inputting the filling rate performance, the density performance and the opening rate and outputting a first sound insulation performance index through the first performance index network;
wherein, equally dividing the intersection interval of each influence index to obtain a plurality of corresponding first sound insulation performance indexes comprises: carrying out mean value segmentation on the intersection intervals of the influence indexes to obtain a plurality of groups of segmentation influence indexes; inputting a plurality of groups of the segmentation influence indexes into the corresponding first performance index network to output a first sound insulation performance index;
the method for acquiring the total affinity comprises the following steps: and averaging the first affinity and the second affinity, and multiplying the average by the third affinity to obtain the total affinity.
2. The artificial intelligence based door and window silence detection method of claim 1, wherein the clustering the plurality of sound insulation materials according to the overall affinity comprises:
and clustering the plurality of sound insulation materials by taking the total affinity as a distance function of clustering, wherein the circumferential radius of the clustering is a preset radius.
3. The artificial intelligence based door and window silence detection method of claim 1, wherein the presetting of the threshold value comprises:
the preset threshold is an actual sound insulation performance index obtained through actual measurement.
4. The artificial intelligence based door and window silence detection method of claim 1, wherein the obtaining of the predicted sound insulation performance index exceeding a preset threshold, adjusting each of the influence indexes, and obtaining the adjustment trend of each influence index according to each adjusted influence index and the corresponding predicted sound insulation performance index comprises:
when the predicted sound insulation performance index exceeds the preset threshold value, adjusting the density according to the density change interval, and obtaining a predicted sound insulation performance index corresponding to the density change interval; acquiring a plurality of groups of adjusted densities and corresponding predicted sound insulation performance indexes thereof, and performing straight line fitting on the plurality of groups of adjusted densities and corresponding predicted sound insulation performance indexes thereof; and taking the straight line obtained by fitting as the regulation trend to regulate the density.
5. The artificial intelligence-based door and window silence detection method according to claim 1, wherein the obtaining of the predicted soundproofing performance index exceeding the preset threshold, adjusting each of the influence indexes, and obtaining the adjustment trend of each influence index according to each adjusted influence index and the corresponding predicted soundproofing performance index, further comprises:
when the predicted sound insulation performance index exceeds the preset threshold value, adjusting the opening rate according to the opening rate change interval, and obtaining a predicted sound insulation performance index corresponding to the opening rate change interval; acquiring multiple groups of adjusted opening rates and corresponding predicted sound insulation performance indexes thereof, and performing straight line fitting on the multiple groups of adjusted opening rates and corresponding predicted sound insulation performance indexes; and taking the straight line obtained by fitting as an adjusting trend to adjust the aperture ratio.
CN202110939614.6A 2021-08-16 2021-08-16 Door and window silence detection method based on artificial intelligence Active CN113657490B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110939614.6A CN113657490B (en) 2021-08-16 2021-08-16 Door and window silence detection method based on artificial intelligence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110939614.6A CN113657490B (en) 2021-08-16 2021-08-16 Door and window silence detection method based on artificial intelligence

Publications (2)

Publication Number Publication Date
CN113657490A CN113657490A (en) 2021-11-16
CN113657490B true CN113657490B (en) 2022-05-31

Family

ID=78480418

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110939614.6A Active CN113657490B (en) 2021-08-16 2021-08-16 Door and window silence detection method based on artificial intelligence

Country Status (1)

Country Link
CN (1) CN113657490B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115331414A (en) * 2022-10-13 2022-11-11 江苏濠玥电子科技有限公司 Temperature early warning method for zirconium oxide ceramic nozzle workpiece

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8015454B1 (en) * 2008-06-02 2011-09-06 Quest Software, Inc. Computer systems and methods for predictive performance management of data transactions
CN106487571A (en) * 2015-09-02 2017-03-08 中国移动通信集团公司 A kind of method and device of assessment network performance index variation tendency
CN106802977A (en) * 2016-12-14 2017-06-06 同济大学 One kind is used for sintering performance index prediction and Quality evaluation method
CN108256689A (en) * 2018-02-06 2018-07-06 华中科技大学 A kind of neural network prediction method of non-crystaline amorphous metal thermoplastic forming performance
CN110674996A (en) * 2019-09-27 2020-01-10 河南大学 An urban traffic noise prediction method
CN111198808A (en) * 2019-12-25 2020-05-26 东软集团股份有限公司 Method, device, storage medium and electronic equipment for predicting performance index
CN112033789A (en) * 2020-09-02 2020-12-04 张海军 Method and system for detecting strength of building material
CN112036581A (en) * 2019-05-15 2020-12-04 上海杰之能软件科技有限公司 Performance detection method and device of vehicle air conditioning system, storage medium and terminal
WO2021004198A1 (en) * 2019-07-10 2021-01-14 江苏金恒信息科技股份有限公司 Plate performance prediction method and apparatus
CN112508275A (en) * 2020-12-07 2021-03-16 国网湖南省电力有限公司 Power distribution network line load prediction method and equipment based on clustering and trend indexes

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI272334B (en) * 2004-07-16 2007-02-01 Takayuki Ueda Adjustable special concrete floor plate with sound insulation properties and device and application method thereof
CN111507228B (en) * 2020-04-10 2022-11-18 中国人民解放军陆军装甲兵学院 Alloy steel mechanical property combined nondestructive testing method based on deep learning

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8015454B1 (en) * 2008-06-02 2011-09-06 Quest Software, Inc. Computer systems and methods for predictive performance management of data transactions
CN106487571A (en) * 2015-09-02 2017-03-08 中国移动通信集团公司 A kind of method and device of assessment network performance index variation tendency
CN106802977A (en) * 2016-12-14 2017-06-06 同济大学 One kind is used for sintering performance index prediction and Quality evaluation method
CN108256689A (en) * 2018-02-06 2018-07-06 华中科技大学 A kind of neural network prediction method of non-crystaline amorphous metal thermoplastic forming performance
CN112036581A (en) * 2019-05-15 2020-12-04 上海杰之能软件科技有限公司 Performance detection method and device of vehicle air conditioning system, storage medium and terminal
WO2021004198A1 (en) * 2019-07-10 2021-01-14 江苏金恒信息科技股份有限公司 Plate performance prediction method and apparatus
CN110674996A (en) * 2019-09-27 2020-01-10 河南大学 An urban traffic noise prediction method
CN111198808A (en) * 2019-12-25 2020-05-26 东软集团股份有限公司 Method, device, storage medium and electronic equipment for predicting performance index
CN112033789A (en) * 2020-09-02 2020-12-04 张海军 Method and system for detecting strength of building material
CN112508275A (en) * 2020-12-07 2021-03-16 国网湖南省电力有限公司 Power distribution network line load prediction method and equipment based on clustering and trend indexes

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Stock market prediction by using artificial neural network;Yunus Yetis等;《WAC》;20141027;第132-136页 *
基于集成径向基网络的制造系统性能预测方法;朱海平等;《计算机集成制造系统》;20090215(第02期);第150-155页 *

Also Published As

Publication number Publication date
CN113657490A (en) 2021-11-16

Similar Documents

Publication Publication Date Title
KR100905586B1 (en) Performance Evaluation System and Method of Microphone for Remote Speech Recognition in Robots
CN111968677B (en) Voice quality self-assessment method for fitting-free hearing aids
CN113657490B (en) Door and window silence detection method based on artificial intelligence
CN116347318A (en) Intelligent production test method and system for sound equipment
Götz et al. Neural network for multi-exponential sound energy decay analysis
CN114387987B (en) Method, device, terminal and storage medium for measuring ecological noise source
Foy et al. Mean absorption estimation from room impulse responses using virtually supervised learning
CN108920854A (en) It is a kind of based on wireless interconnected and noise inline diagnosis harmony method for evaluating quality and system of athe portable client
CN112200238A (en) Hard rock tension-shear fracture identification method and device based on sound characteristics
Xie et al. Research and development of sound quality in portable testing and evaluation system based on self-adaptive neural network
CN115683318A (en) Method for evaluating sound quality and determining limit value of driving motor system
CN109933933B (en) Noise treatment method and equipment
Loh et al. Toward child-appropriate acoustic measurement methods in primary schools and daycare centers
CN109671430A (en) Voice processing method and device
CN118645101B (en) Intelligent sound box control system
CN111554325A (en) Voice recognition method and system
CN115512718A (en) Voice quality evaluation method, device and system for stock voice file
CN115346560A (en) Sound level weighting method for subjective annoyance degree comparison of train station hall
CN117280415A (en) Apparatus and method for pure dialog loudness estimation based on deep neural network
CN114220455B (en) Vehicle door closing sound quality evaluation method and system
CN117278895B (en) Microphone signal optimization method and system based on data enhancement
CN118609602B (en) A method and system for judging the degree of environmental reverberation based on speech signals
CN117612566B (en) Audio quality assessment method and related product
Siripool et al. Blind Estimation of Room Volume from Reverberant Speech Based on the Modulation Transfer Function
CN113691924A (en) Quantitative evaluation method for active noise reduction effect of TWS (time and frequency) headset ANC (acoustic control and noise cancellation)

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant