CN115144312A - Indoor air fine particle measuring method and system based on Internet of things - Google Patents
Indoor air fine particle measuring method and system based on Internet of things Download PDFInfo
- Publication number
- CN115144312A CN115144312A CN202210763054.8A CN202210763054A CN115144312A CN 115144312 A CN115144312 A CN 115144312A CN 202210763054 A CN202210763054 A CN 202210763054A CN 115144312 A CN115144312 A CN 115144312A
- Authority
- CN
- China
- Prior art keywords
- matrix
- vector
- distance
- topological
- feature
- 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.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume, or surface-area of porous materials
- G01N15/06—Investigating concentration of particle suspensions
-
- 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
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
Abstract
The application relates to the field of indoor air fine particle detection, and particularly discloses an indoor air fine particle measuring method and system based on the Internet of things. In this way, the position of the undeployed sensor can be accurately predicted to form an accurate fine-grained indoor air mass distribution.
Description
Technical Field
The invention relates to the field of indoor air fine particle detection, and more particularly relates to an indoor air fine particle measurement method and system based on the Internet of things.
Background
With the increasing severity of air pollution, people pay more and more attention to air quality. Air pollution has become the largest environmental pollution, and while people pay attention to outdoor air quality, indoor air quality is easily ignored, and people in cities spend 70% of the time indoors, so that the influence of indoor air pollution on body health is more obvious.
In indoor air pollution, PM2.5 is seriously harmful, mainly refers to particles with equivalent diameter less than 2.5 μm, which can be suspended in the air for a long time, carry toxic substances, can reach the lung and enter alveoli, and carry harmful substances even can enter the body to circulate. With the aggravation of outdoor pollution and the limitation of outdoor sport places, people are more inclined to indoor sport fitness. People need to consume a large amount of oxygen when moving, and the number of breathings is more, therefore indoor environmental health is crucial to indoor body-building motion, and especially PM2.5 is more serious to the harm of people, mainly can cause respiratory system's disease, but indoor air quality worsens, only relies on vision and smell to be difficult to discover, therefore, real-time monitoring gymnasium air quality is reluctant. However, indoor air quality measurement is very challenging with respect to outdoor air quality measurement.
Therefore, in order to mine the correlation between different places from historical data to deduce the value of the position of the undeployed sensor based on the measured value so as to form a fine-grained indoor air quality distribution, an indoor air fine particle measurement method based on the internet of things is desired.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides an indoor air fine particle measurement method and system based on the Internet of things, wherein topological features between a plurality of sensors and undeployed sensors and high-dimensional correlation feature distribution among measurement values of the sensors are mined through a convolutional neural network model, and high-dimensional correlation extraction is carried out on distance features among the sensors through an encoder model, so that a correction Gaussian distribution graph can be further constructed to carry out Gaussian reformation on the measurement value features on the global level, and correction Gaussian distribution reflecting global Gaussian drift of the measurement value features is obtained to improve the accuracy of decoding regression. In this way, the position of the undeployed sensor can be accurately predicted to form an accurate fine-grained indoor air mass distribution.
According to one aspect of the application, an indoor air fine particle measurement method based on the Internet of things is provided, and comprises the following steps: constructing a topological matrix between a plurality of sensors and a position X where no sensor is deployed, wherein the value of each position on the diagonal position of the topological matrix is the distance between each sensor and the position X, and the value of each position on the off-diagonal position of the topological matrix is the distance between two corresponding sensors; passing the topological matrix through a first convolutional neural network to obtain a topological feature matrix; acquiring measurement values of the plurality of sensors; arranging the measurement values of the plurality of sensors into a measurement value vector, and multiplying the measurement value vector by the transpose of the measurement value vector to obtain a measurement value incidence matrix; passing the measured value correlation matrix through a second convolutional neural network to obtain a correlation characteristic matrix; fusing the topological characteristic matrix and the associated characteristic matrix to obtain a topological associated characteristic matrix; arranging the distances between the sensors and the position X into distance vectors, and then obtaining distance characteristic vectors through a sequence encoder comprising a one-dimensional convolutional layer and a full-connection layer; taking the distance characteristic vector as a query vector to be subjected to matrix multiplication with the topology association characteristic matrix to obtain a distance topology characteristic vector; constructing a modified Gaussian distribution map of the distance topological feature vector, wherein each position of the modified Gaussian distribution map is a Gaussian distribution, a mean vector of the modified Gaussian distribution map is the distance topological feature vector, and a variance vector of the modified Gaussian distribution map is generated based on the mean vector and an initial variance matrix constructed based on the mean vector; performing Gaussian discretization on the Gaussian distribution of each position of the correction Gaussian distribution diagram to convert the Gaussian distribution of each position of the correction Gaussian distribution diagram into a vector, and performing two-dimensional splicing on the vectors corresponding to each position of the correction Gaussian distribution diagram to obtain a Gaussian correction matrix; performing matrix multiplication on the distance topological characteristic vector and the Gaussian correction matrix to obtain a regression characteristic vector; and performing decoding regression on the regression feature vector by using a decoder to obtain a decoded value, wherein the decoded value is a measurement predicted value of the position X of the undeployed sensor.
According to another aspect of the present application, there is provided an internet of things-based indoor air fine particle measurement system, including: the topological matrix construction unit is used for constructing a topological matrix between a plurality of sensors and a position X where the sensors are not deployed, wherein the value of each position on the diagonal position of the topological matrix is the distance between each sensor and the position X, and the value of each position on the off-diagonal position of the topological matrix is the distance between two corresponding sensors; the first convolution unit is used for enabling the topology matrix obtained by the topology matrix building unit to pass through a first convolution neural network so as to obtain a topology characteristic matrix; a measurement value acquisition unit for acquiring measurement values of the plurality of sensors; a measurement value correlation matrix calculation unit configured to arrange the measurement values of the plurality of sensors obtained by the measurement value obtaining unit into measurement value vectors and multiply the measurement value vectors with the transpose of the measurement value vectors to obtain a measurement value correlation matrix; the second convolution unit is used for enabling the measured value incidence matrix obtained by the measured value incidence matrix calculation unit to pass through a second convolution neural network so as to obtain an incidence characteristic matrix; a fusion unit, configured to fuse the topological feature matrix obtained by the first convolution unit and the associated feature matrix obtained by the second convolution unit to obtain a topological associated feature matrix; the encoding unit is used for arranging the distances between the sensors and the position X into distance vectors and then obtaining distance characteristic vectors through a sequence encoder comprising a one-dimensional convolutional layer and a full-connection layer; the matrix multiplication unit is used for performing matrix multiplication on the distance characteristic vector obtained by the coding unit as a query vector and the topology association characteristic matrix obtained by the fusion unit to obtain a distance topology characteristic vector; a modified gaussian distribution map constructing unit, configured to construct a modified gaussian distribution map of the distance topological feature vector obtained by the matrix multiplication unit, where each position of the modified gaussian distribution map is a gaussian distribution, a mean vector of the modified gaussian distribution map is the distance topological feature vector, and a variance vector of the modified gaussian distribution map is generated based on the mean vector and an initial variance matrix constructed based on the mean vector; the Gaussian discretization unit is used for carrying out Gaussian discretization on the Gaussian distribution of each position of the correction Gaussian distribution map obtained by the correction Gaussian distribution map construction unit so as to convert the Gaussian distribution of each position of the correction Gaussian distribution map into a vector, and carrying out two-dimensional splicing on the vectors corresponding to each position of the correction Gaussian distribution map so as to obtain a Gaussian correction matrix; a regression feature vector generation unit, configured to perform matrix multiplication on the distance topology feature vector obtained by the matrix multiplication unit and the gaussian correction matrix obtained by the gaussian discretization unit to obtain a regression feature vector; and a decoding regression unit configured to perform decoding regression on the regression feature vector obtained by the regression feature vector generation unit using a decoder to obtain a decoded value, which is a measurement prediction value of a position X where no sensor is deployed.
According to yet another aspect of the present application, there is provided a computer readable medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to perform the internet of things based indoor air fine particle measurement method as described above.
Compared with the prior art, the indoor air fine particle measurement method and system based on the internet of things provided by the application excavate the topological characteristics between the sensors and the undeployed sensors and the high-dimensional correlation characteristic distribution among the measured values of the sensors through the convolutional neural network model, and perform high-dimensional correlation extraction on the distance characteristics among the sensors through the encoder model, so that a correction Gaussian distribution map can be further constructed to perform Gaussian reformation on the measured value characteristics on the global level, and a correction Gaussian distribution for reflecting the global Gaussian drift of the measured value characteristics is obtained, so that the accuracy of decoding regression is improved. In this way, the position of the undeployed sensor can be accurately predicted to form an accurate fine-grained indoor air mass distribution.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally indicate like parts or steps.
Fig. 1 is an application scenario diagram of an indoor air fine particle measurement method based on the internet of things according to an embodiment of the application;
fig. 2 is a flowchart of an internet of things-based indoor air fine particle measurement method according to an embodiment of the application;
fig. 3 is a schematic system architecture diagram of an internet of things-based indoor air fine particle measurement method according to an embodiment of the application;
fig. 4 is a block diagram of an internet of things-based indoor air fine particle measurement system according to an embodiment of the application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.
Scene overview
As mentioned above, as air pollution becomes more serious, people pay more attention to air quality. Air pollution has become the largest environmental pollution, and while people pay attention to outdoor air quality, indoor air quality is easily ignored, and people in cities spend 70% of the time indoors, so that the influence of indoor air pollution on body health is more obvious.
In indoor air pollution, PM2.5 is seriously harmful, mainly refers to particles with equivalent diameter less than 2.5 μm, which can be suspended in the air for a long time, carry toxic substances, can reach the lung and enter alveoli, and carry harmful substances even can enter the body to circulate. With the increasing outdoor pollution and the limited outdoor sport area, people are more inclined to indoor sports fitness. People need to consume a large amount of oxygen when moving, and the number of breathings is more, therefore indoor environmental health is crucial to indoor body-building motion, and especially PM2.5 is more serious to the harm of people, mainly can cause respiratory system's disease, but indoor air quality worsens, only relies on vision and smell to be difficult to discover, therefore, real-time monitoring gymnasium air quality is reluctant. However, indoor air quality measurement is very challenging with respect to outdoor air quality measurement.
Therefore, in order to mine the correlation between different places from the historical data to deduce the value of the position where the sensor is not deployed based on the measured value, so as to form the indoor air quality distribution with fine granularity, an indoor air fine particle measurement method based on the internet of things is desired.
Based on this, in the technical scheme of the application, a topological matrix of a plurality of sensors and a position X where the sensor is not deployed is first constructed, a diagonal line of the matrix is a distance value between a certain sensor and the position X, and a first convolution neural network is input to obtain a topological characteristic matrix.
And acquiring the measurement values of the plurality of sensors, forming a measurement value vector, multiplying the measurement value vector by the transpose of the vector to obtain a correlation matrix, and inputting the correlation matrix into the second convolutional neural network to obtain a correlation characteristic matrix.
And multiplying the topological characteristic matrix and the incidence characteristic matrix to obtain a topological incidence matrix.
And forming distance vectors by using distances between the sensors and the position X, inputting the distance vectors into a time sequence encoder to obtain distance characteristic vectors, and multiplying the distance characteristic vectors serving as query vectors and the topology incidence matrix to obtain distance topology vectors.
Considering that the measured values of a plurality of devices approximately obey gaussian (Gauss) distribution on the overall spatial distribution relative to the measuring device, by performing gaussian reforming on the measured value characteristics on the global level, a corrected gaussian distribution reflecting the global gaussian drift of the measured value characteristics can be obtained.
Inputting the measured value vector into a time sequence encoder to obtain a measured characteristic vector, multiplying the measured characteristic vector by a topological correlation matrix to obtain a measured topological vector, taking the vector as a mean vector mu of Gaussian distribution X-mu (mu, sigma), and taking a variance matrix between every two characteristic values of the vector as an initial variance matrix of the Gaussian distribution
And based on the mean vector mu and the initial variance matrix
The product of (a) yields a variance vector σ. Then, the modified Gaussian distribution is assigned to the Gaussian distribution x of each position i ～N i (μ i ,σ i ) And performing Gaussian discretization to obtain a Gaussian correction matrix.
And finally, multiplying the distance topological vector by the Gaussian correction matrix to obtain a regression characteristic vector, and obtaining an inference result through decoder regression.
Based on this, the application provides an indoor air fine particle measurement method based on thing networking, and it includes: constructing a topological matrix between a plurality of sensors and a position X where no sensor is deployed, wherein the value of each position on the diagonal position of the topological matrix is the distance between each sensor and the position X, and the value of each position on the off-diagonal position of the topological matrix is the distance between two corresponding sensors; passing the topological matrix through a first convolutional neural network to obtain a topological feature matrix; acquiring measurement values of the plurality of sensors; arranging the measured values of the plurality of sensors into a measured value vector and multiplying the measured value vector by the transpose of the measured value vector to obtain a measured value incidence matrix; passing the measured value correlation matrix through a second convolutional neural network to obtain a correlation characteristic matrix; fusing the topological characteristic matrix and the associated characteristic matrix to obtain a topological associated characteristic matrix; arranging the distances between the sensors and the position X into distance vectors, and then obtaining distance characteristic vectors through a sequence encoder comprising a one-dimensional convolutional layer and a full-connection layer; taking the distance characteristic vector as a query vector to be subjected to matrix multiplication with the topology association characteristic matrix to obtain a distance topology characteristic vector; constructing a modified Gaussian distribution map of the distance topological feature vector, wherein each position of the modified Gaussian distribution map is a Gaussian distribution, a mean vector of the modified Gaussian distribution map is the distance topological feature vector, and a variance vector of the modified Gaussian distribution map is generated based on the mean vector and an initial variance matrix constructed based on the mean vector; performing Gaussian discretization on the Gaussian distribution of each position of the correction Gaussian distribution map to convert the Gaussian distribution of each position of the correction Gaussian distribution map into a vector, and performing two-dimensional splicing on the vectors corresponding to each position of the correction Gaussian distribution map to obtain a Gaussian correction matrix; performing matrix multiplication on the distance topological characteristic vector and the Gaussian correction matrix to obtain a regression characteristic vector; and performing decoding regression on the regression feature vector by using a decoder to obtain a decoded value, wherein the decoded value is a measurement predicted value of the position X of the undeployed sensor.
Fig. 1 illustrates an application scenario of an internet of things-based indoor air fine particle measurement method according to an embodiment of the application. As shown in fig. 1, in this application scenario, first, a plurality of sensors for monitoring air quality (e.g., P1 to Pn as illustrated in fig. 1) are deployed within a specific site, for example, a gymnasium, but no sensors are deployed at any of some locations (e.g., location X as illustrated in fig. 1). Then, a topological matrix of distances between the plurality of sensors and a location where no sensors are deployed is obtained and measurements of the plurality of sensors are obtained. The obtained topological matrix and the measured values of the plurality of sensors are then input into a server (e.g., S as illustrated in fig. 1) deployed with an internet of things based indoor air fine particle measurement algorithm, wherein the server is capable of processing the topological matrix and the measured values of the plurality of sensors with the internet of things based indoor air fine particle measurement algorithm to generate measurement predicted values for a location X where no sensor is deployed.
Having described the general principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary method
Fig. 2 illustrates a flow chart of an internet of things-based indoor air fine particle measurement method. As shown in fig. 2, an indoor air fine particle measurement method based on the internet of things according to an embodiment of the present application includes: s110, constructing a topological matrix between a plurality of sensors and a position X where the sensors are not deployed, wherein the value of each position on the diagonal position of the topological matrix is the distance between each sensor and the position X, and the value of each position on the off-diagonal position of the topological matrix is the distance between two corresponding sensors; s120, passing the topological matrix through a first convolutional neural network to obtain a topological characteristic matrix; s130, obtaining the measurement values of the plurality of sensors; s140, arranging the measurement values of the plurality of sensors into measurement value vectors, and multiplying the measurement value vectors by the transpose of the measurement value vectors to obtain a measurement value incidence matrix; s150, enabling the measured value correlation matrix to pass through a second convolutional neural network to obtain a correlation characteristic matrix; s160, fusing the topological characteristic matrix and the associated characteristic matrix to obtain a topological associated characteristic matrix; s170, arranging the distances between the sensors and the position X into distance vectors, and then obtaining distance characteristic vectors through a sequence encoder comprising a one-dimensional convolutional layer and a full-connection layer; s180, taking the distance characteristic vector as a query vector to be subjected to matrix multiplication with the topology association characteristic matrix to obtain a distance topology characteristic vector; s190, constructing a modified Gaussian distribution map of the distance topological feature vector, wherein each position of the modified Gaussian distribution map is a Gaussian distribution, a mean vector of the modified Gaussian distribution map is the distance topological feature vector, and a variance vector of the modified Gaussian distribution map is generated based on the mean vector and an initial variance matrix constructed based on the mean vector; s200, carrying out Gaussian discretization on the Gaussian distribution of each position of the correction Gaussian distribution map so as to convert the Gaussian distribution of each position of the correction Gaussian distribution map into a vector, and carrying out two-dimensional splicing on the vectors corresponding to each position of the correction Gaussian distribution map so as to obtain a Gaussian correction matrix; s210, performing matrix multiplication on the distance topological characteristic vector and the Gaussian correction matrix to obtain a regression characteristic vector; and S220, performing decoding regression on the regression feature vector by using a decoder to obtain a decoded value, wherein the decoded value is a measurement predicted value of the position X of the undeployed sensor.
Fig. 3 illustrates an architecture diagram of an indoor air fine particle measurement method based on the internet of things according to an embodiment of the application. As shown in fig. 3, in the network architecture of the internet of things-based indoor air fine particle measurement method, first, the obtained topology matrix (e.g., M as illustrated in fig. 3) is passed through a first convolutional neural network (e.g., CNN1 as illustrated in fig. 3) to obtain a topology feature matrix (e.g., MF1 as illustrated in fig. 3); then, arranging the obtained measurement values (for example, P1 as illustrated in fig. 3) of the plurality of sensors into a measurement value vector (for example, V1 as illustrated in fig. 3) and multiplying the measurement value vector by the transpose of the measurement value vector to obtain a measurement value association matrix (for example, M1 as illustrated in fig. 3); then, passing the measurement correlation matrix through a second convolutional neural network (e.g., CNN2 as illustrated in fig. 3) to obtain a correlation feature matrix (e.g., MF2 as illustrated in fig. 3); then, fusing the topological feature matrix and the associated feature matrix to obtain a topological associated feature matrix (e.g., MF as illustrated in fig. 3); then, arranging distances (e.g., P2 as illustrated in fig. 3) between the plurality of sensors and the position X as a distance vector (e.g., V2 as illustrated in fig. 3) and then passing through a sequence encoder (e.g., E as illustrated in fig. 3) including one-dimensional convolutional layers and fully-connected layers to obtain a distance feature vector (e.g., VF1 as illustrated in fig. 3); then, matrix-multiplying the distance feature vector as a query vector with the topology association feature matrix to obtain a distance topology feature vector (for example, VF2 as illustrated in fig. 3); then, constructing a modified gaussian distribution map of the distance topological feature vector (e.g., GD as illustrated in fig. 3); next, performing gaussian discretization on the gaussian distribution of each position of the modified gaussian distribution map to convert the gaussian distribution of each position of the modified gaussian distribution map into one vector (for example, V as illustrated in fig. 3), and performing two-dimensional stitching on the vectors corresponding to each position of the modified gaussian distribution map to obtain a gaussian modification matrix (for example, MG as illustrated in fig. 3); then, matrix-multiplying the distance topological feature vector with the gaussian correction matrix to obtain a regression feature vector (e.g., VC as illustrated in fig. 3); and, finally, decoding the regression feature vector using a decoder (e.g., D as illustrated in fig. 3) to obtain a decoded value, which is a measured predicted value of the position X of the undeployed sensor.
Constructing a topological matrix between a plurality of sensors and a position X where no sensor is deployed, wherein the value of each position on the diagonal position of the topological matrix is the distance between each sensor and the position X, and the value of each position on the off-diagonal position of the topological matrix is the distance between two corresponding sensors, and the topological matrix is passed through a first convolutional neural network to obtain a topological characteristic matrix. As previously mentioned, it should be understood that in the solution of the present application, in order to make an accurate prediction of the undeployed sensor position, it is desirable to mine the correlation between different sites from the historical data to infer the value of the undeployed sensor position based on the measured value, thereby forming a fine-grained indoor air quality distribution.
That is, specifically, in the technical solution of the present application, first, a topological matrix between the plurality of sensors and a position X where no sensor is deployed is obtained according to a distance between the respective sensors. Here, the value of each position on the diagonal position of the topology matrix is the distance between each sensor and the position X, and the value of each position on the off-diagonal position of the topology matrix is the distance between the corresponding two sensors. And then, processing the topological matrix in a first convolution network to extract topological characteristic information among the sensors, so as to obtain the topological characteristic matrix. It should be understood that the high-dimensional features extracted by the convolutional neural network and embodying the correlation information between the input data are used for calculating instead of the original data, so that the influence of the error of the original data on the data dimension can be eliminated.
Specifically, in the embodiment of the present application, the process of passing the topology matrix through a first convolutional neural network to obtain a topology feature matrix includes: firstly, each layer except the last layer of the first convolutional neural network performs convolution processing, pooling processing and activation processing on input data to output a topological feature map by the last second layer of the first convolutional neural network, wherein the input of the first layer of the first convolutional neural network is the topological matrix. Then, the last layer of the first convolutional neural network performs convolution processing, global mean pooling along channel dimensions, and activation processing on the topological feature map to generate the topological feature matrix. It should be appreciated that pooling can be used for feature dimension reduction, mitigation of the risk of overfitting, and reduction of the hypersensitivity of the convolutional layer to the detection information, while mean pooling is a means preserving computation in a pooling kernel for highlighting important parts of the feature map. That is, by performing global mean pooling on the topological feature map, the number of parameters can be reduced to increase the training speed, and further, the whole network structure is normalized to prevent overfitting.
In steps S130 and S140, the measurement values of the plurality of sensors are acquired, and the measurement values of the plurality of sensors are arranged into a measurement value vector and then multiplied by the transpose of the measurement value vector to obtain a measurement value correlation matrix. It should be appreciated that in order to mine the correlation between different sites from historical data to infer the value of the undeployed sensor's location based on the measurements, a fine-grained indoor air quality distribution is formed. Therefore, in the technical solution of the present application, it is further required to obtain the measurement values of the plurality of sensors, and multiply the measurement values of the plurality of sensors after being arranged as measurement value vectors by the transpose of the measurement value vectors, so as to obtain a measurement value correlation matrix with global correlation characteristic information, so as to perform high-dimensional correlation characteristic extraction on the measurement value correlation matrix in the following.
In steps S150 and S160, the measured value correlation matrix is passed through a second convolutional neural network to obtain a correlation feature matrix, and the topological feature matrix and the correlation feature matrix are fused to obtain a topological correlation feature matrix. That is, in the technical solution of the present application, after the measured value correlation matrix is obtained, similarly, the measured value correlation matrix is further processed through a second convolutional neural network to extract high-dimensional correlation features between the measured values of the plurality of sensors, it should be understood that the high-dimensional features extracted by the convolutional neural network and representing the associated information between the input data are used for computing instead of the original data, so that the influence of errors of the original data on the data dimension can be eliminated. Accordingly, in one specific example, each layer of the second convolutional neural network except the last layer performs convolution processing, pooling processing and activation processing on input data to output a correlation feature map by the last second layer of the second convolutional neural network, and an input of the first layer of the second convolutional neural network is the measurement value correlation matrix; and the last layer of the second convolutional neural network performs convolution processing, global mean pooling along channel dimensions, and activation processing on the associated feature map to generate the associated feature matrix. Then, the topological feature matrix and the correlation feature matrix may be matrix-multiplied to fuse the topological feature information and the measured value correlation feature information to obtain a topological correlation feature matrix.
In steps S170 and S180, after the distances between the plurality of sensors and the position X are arranged as a distance vector, a distance feature vector is obtained by a sequence encoder including a one-dimensional convolutional layer and a full-link layer, and the distance feature vector is used as a query vector to be matrix-multiplied by the topology correlation feature matrix to obtain a distance topology feature vector. That is, in the technical solution of the present application, the obtained distances between the plurality of sensors and the position X are further arranged as a distance vector, and the distance vector is encoded by a sequence encoder including a one-dimensional convolutional layer and a full link layer to extract high-dimensional implicit features of feature values of each position in the distance feature vector, thereby obtaining the distance feature vector. In this way, the distance feature vector can be used as a query vector to perform matrix multiplication with the topology associated feature matrix so as to map the distance feature vector into the feature space of the topology associated feature matrix, thereby obtaining the distance topology feature vector.
Specifically, in this embodiment, the process of arranging the distances between the sensors and the position X as a distance vector and then obtaining a distance feature vector by a sequence encoder including a one-dimensional convolutional layer and a fully-connected layer includes: first, the distance vector is full-concatenation-encoded using a full-concatenation of the sequence encoder to extract high-dimensional implicit features of feature values of respective positions in the distance vector in the following formula
Where X is the distance vector, Y is the output vector, W is the weight matrix, B is the offset vector,
representing a matrix multiplication. Then, using a one-dimensional convolution layer of the sequence encoder, performing one-dimensional convolution encoding on the distance vector by using the following formula to extract high-dimensional implicit correlation features of the correlation between feature values of each position in the distance vector, wherein the formula is as follows:
wherein, a is the width of the convolution kernel in the x direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, and w is the size of the convolution kernel.
In steps S190 and S200, a modified gaussian distribution map of the distance topological feature vector is constructed, wherein each position of the modified gaussian distribution map is a gaussian distribution, a mean vector of the modified gaussian distribution map is the distance topological feature vector, a variance vector of the modified gaussian distribution map is generated based on the mean vector and an initial variance matrix constructed based on the mean vector, the gaussian distribution of each position of the modified gaussian distribution map is subjected to gaussian discretization to convert the gaussian distribution of each position of the modified gaussian distribution map into a vector, and vectors corresponding to each position of the modified gaussian distribution map are subjected to two-dimensional stitching to obtain a gaussian modified matrix. It will be appreciated that, taking into account the measurement errors of the measurements of the individual sensors, which are determined by the characteristics of the sensors themselves, the measurements of the plurality of devices, over a long period of statistical analysis, approximately follow a gaussian (Gauss) distribution over the overall spatial distribution with respect to the measuring device. Therefore, in the technical solution of the present application, a modified gaussian distribution for reflecting a global gaussian drift of the measured value feature can be obtained by further performing gaussian reforming on the measured value feature on a global level.
That is, specifically, in the technical solution of the present application, the distance topological feature vector is taken as a mean vector μ of gaussian distributions X to N (μ, σ), and a variance matrix between every two eigenvalues of the vector is taken as an initial variance matrix μ of the gaussian distributions
And based on the mean vector mu and the initial variance matrix
The product of which yields the variance vector sigma. Then, the modified Gaussian distribution is assigned to the Gaussian distribution x of each position i ～N i (μ i ,σ i ) And carrying out Gaussian discretization, and carrying out two-dimensional splicing on vectors corresponding to all the positions of the obtained correction Gaussian distribution map so as to obtain a Gaussian correction matrix.
Specifically, in the embodiment of the present application, the process of constructing the modified gaussian distribution map of the distance topological feature vector includes: constructing a modified Gaussian distribution map of the distance topological feature vector X to N (mu, sigma) by the following formula, wherein the mean vector mu is the distance topological feature vector, and the variance vector sigma is an initial variance matrix of Gaussian distribution by first taking a variance matrix between every two eigenvalues of the mean vector as the variance matrix
Then, the mean vector mu and the initial variance matrix are calculated
The product of (c) is generated.
In steps S210 and S220, the distance topology feature vector is matrix-multiplied by the gaussian correction matrix to obtain a regression feature vector, and a decoder is used to perform decoding regression on the regression feature vector to obtain a decoded value, where the decoded value is a measured predicted value of the position X where no sensor is deployed. That is to say, in the technical solution of the present application, the distance topological feature vector is further subjected to matrix multiplication with the gaussian correction matrix, so as to map the gaussian correction matrix to a high-dimensional feature space of the distance topological feature vector, so as to improve accuracy of subsequent decoding regression, and further obtain a regression feature vector. In this way, the regression feature vector can be further processed through decoding regression in a decoder to generate a decoded value representing the measured predicted value of the location X where no sensor is deployed.
Specifically, in the embodiment of the present application, a process of performing decoding regression on the regression feature vector by using a decoder to obtain a decoded value, where the decoded value is a measurement predicted value of a position X where a sensor is not deployed, includes: decoding the regression feature vector using a plurality of fully-connected layers of the decoder to obtain the decoded value using the following formula:
where X is the regression feature vector, Y is the decoded value, W is the weight matrix, B is the offset vector,
representing the matrix multiplication, h (-) is the activation function.
In summary, the internet-of-things-based indoor air fine particle measurement method according to the embodiment of the present application is elucidated, which excavates a topological feature between the plurality of sensors and an undeployed sensor and a high-dimensional correlation feature distribution between measurement values of the plurality of sensors through a convolutional neural network model, and performs high-dimensional correlation extraction on a distance feature between the sensors through an encoder model, so that a modified gaussian distribution map can be further constructed to perform gaussian renormalization on a global level on the measurement value feature, thereby obtaining a modified gaussian distribution reflecting a global gaussian drift of the measurement value feature, so as to improve accuracy of decoding regression. In this way, the position of the undeployed sensor can be accurately predicted to form an accurate fine-grained indoor air mass distribution.
Exemplary System
Fig. 4 illustrates a block diagram of an internet of things based indoor air fine particle measurement system according to an embodiment of the application. As shown in fig. 4, an internet of things-based indoor air fine particle measurement system 400 according to an embodiment of the present application includes: a topology matrix constructing unit 410, configured to construct a topology matrix between a plurality of sensors and a position X where no sensor is deployed, where a value of each position on a diagonal position of the topology matrix is a distance between each sensor and the position X, and a value of each position on a non-diagonal position of the topology matrix is a distance between two corresponding sensors; a first convolution unit 420, configured to pass the topology matrix obtained by the topology matrix construction unit 410 through a first convolution neural network to obtain a topology feature matrix; a measurement value acquisition unit 430 for acquiring measurement values of the plurality of sensors; a measurement value correlation matrix calculation unit 440 configured to arrange the measurement values of the plurality of sensors obtained by the measurement value obtaining unit 430 into measurement value vectors and multiply the measurement value vectors with the transposes of the measurement value vectors to obtain measurement value correlation matrices; a second convolution unit 450, configured to pass the measured value correlation matrix obtained by the measured value correlation matrix calculation unit 440 through a second convolution neural network to obtain a correlation feature matrix; a fusion unit 460, configured to fuse the topological feature matrix obtained by the first convolution unit 420 and the associated feature matrix obtained by the second convolution unit 450 to obtain a topological associated feature matrix; an encoding unit 470, configured to arrange distances between the plurality of sensors and the position X into a distance vector and then obtain a distance feature vector through a sequence encoder including a one-dimensional convolutional layer and a full-link layer; a matrix multiplication unit 480, configured to perform matrix multiplication on the distance feature vector obtained by the encoding unit 470 as a query vector and the topology association feature matrix obtained by the fusion unit 460 to obtain a distance topology feature vector; a modified gaussian distribution map constructing unit 490, configured to construct a modified gaussian distribution map of the distance topological feature vector obtained by the matrix multiplying unit 480, where each position of the modified gaussian distribution map is a gaussian distribution, a mean vector of the modified gaussian distribution map is the distance topological feature vector, and a variance vector of the modified gaussian distribution map is generated based on the mean vector and an initial variance matrix constructed based on the mean vector; a gaussian discretization unit 500, configured to perform gaussian discretization on the gaussian distribution at each position of the modified gaussian distribution map obtained by the modified gaussian distribution map constructing unit 490 to convert the gaussian distribution at each position of the modified gaussian distribution map into a vector, and perform two-dimensional stitching on the vectors corresponding to each position of the modified gaussian distribution map to obtain a gaussian modification matrix; a regression feature vector generation unit 510, configured to perform matrix multiplication on the distance topology feature vector obtained by the matrix multiplication unit 480 and the gaussian correction matrix obtained by the gaussian discretization unit 500 to obtain a regression feature vector; and a decoding regression unit 520 configured to perform decoding regression on the regression feature vector obtained by the regression feature vector generation unit 510 using a decoder to obtain a decoded value, which is a measurement prediction value of the sensor-undeployed position X.
In one example, in the internet of things based indoor air fine particle measurement system 400, the first volume unit 420 is further configured to: each layer except the last layer of the first convolutional neural network carries out convolution processing, pooling processing and activation processing on input data so as to output a topological feature map from the last second layer of the first convolutional neural network, wherein the input of the first layer of the first convolutional neural network is the topological matrix; and the last layer of the first convolutional neural network performs convolution processing, global mean pooling along channel dimensions and activation processing on the topological feature map to generate the topological feature matrix.
In one example, in the internet of things based indoor air fine particle measurement system 400, the second convolution unit 450 is further configured to: performing convolution processing, pooling processing and activation processing on input data by each layer except the last layer of the second convolutional neural network to output a correlation feature map by the last second layer of the second convolutional neural network, wherein the input of the first layer of the second convolutional neural network is the measured value correlation matrix; the last layer of the second convolutional neural network performs convolution processing, global mean pooling along channel dimensions, and activation processing on the associated feature map to generate the associated feature matrix.
In one example, in the internet of things based indoor air fine particle measurement system 400, the encoding unit 470 is further configured to: fully concatenating the distance vector using a full concatenation of the sequence encoder to extract high-dimensional implicit features of feature values of respective positions in the distance vector in the following formula
Where X is a distance vector, Y is an output vector, W is a weight matrix, B is an offset vector,
represents a matrix multiplication: and performing one-dimensional convolutional encoding on the distance vector by using a one-dimensional convolutional layer of the sequence encoder according to the following formula to extract high-dimensional implicit associated features of the association between feature values of the positions in the distance vector, wherein the formula is as follows:
wherein, a is the width of the convolution kernel in the x direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, and w is the size of the convolution kernel.
In an example, in the internet of things based indoor air fine particle measurement system 400, the modified gaussian distribution map constructing unit 490 is further configured to: constructing a modified Gaussian distribution map of the distance topological feature vector X to N (mu, sigma) by the following formula, wherein the mean vector mu is the distance topological feature vector, and the variance vector sigma is an initial variance matrix of Gaussian distribution by first taking a variance matrix between every two eigenvalues of the mean vector as the variance matrix
Then, the mean vector mu and the initial variance matrix are calculated
The product of (c) is generated.
In one example, in the internet of things based indoor air fine particle measurement system 400, the decoding regression unit 520 is further configured to: decoding the regression feature vector using a plurality of fully-connected layers of the decoder to obtain the decoded value using the following formula:
where X is the regression feature vector, Y is the decoded value, W is the weight matrix, B is the offset vector,
representing the matrix multiplication, h (-) is the activation function.
Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the internet of things based indoor air fine particle measurement system 400 have been described in detail in the above description of the internet of things based indoor air fine particle measurement method with reference to fig. 1 to 3, and thus, a repeated description thereof will be omitted.
As described above, the internet of things-based indoor air fine particle measurement system 400 according to the embodiment of the present application may be implemented in various terminal devices, for example, a server of an internet of things-based indoor air fine particle measurement algorithm, and the like. In one example, the internet of things based indoor air fine particle measurement system 400 according to the embodiment of the present application may be integrated into a terminal device as one software module and/or hardware module. For example, the internet of things based indoor air fine particle measurement system 400 may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the internet of things-based indoor air fine particle measurement system 400 can also be one of the hardware modules of the terminal device.
Alternatively, in another example, the internet of things-based indoor air fine particle measurement system 400 and the terminal device may also be separate devices, and the internet of things-based indoor air fine particle measurement system 400 may be connected to the terminal device through a wired and/or wireless network and transmit the interaction information according to an agreed data format.
Exemplary computer program product and computer-readable storage Medium
In addition to the above-described methods and apparatus, embodiments of the present application may also be a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps in the functions in the internet of things based indoor air fine particle measurement method according to various embodiments of the present application described in the "exemplary methods" section of this specification, above.
The computer program product may be written with program code for performing the operations of embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server.
Furthermore, embodiments of the present application may also be a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, cause the processor to perform the steps in the internet of things based indoor air fine particle measurement method described in the "exemplary methods" section above in this specification.
The computer-readable storage medium may take any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may include, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, or device, or a combination of any of the above. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The foregoing describes the general principles of the present application in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present application are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present application. Furthermore, the foregoing disclosure of specific details is for the purpose of illustration and description and is not intended to be limiting, since the foregoing disclosure is not intended to be exhaustive or to limit the disclosure to the precise details disclosed.
The block diagrams of devices, apparatuses, systems referred to in this application are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by those skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably therewith. The words "or" and "as used herein mean, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".
It should also be noted that in the devices, apparatuses, and methods of the present application, the components or steps may be decomposed and/or recombined. These decompositions and/or recombinations are to be considered as equivalents of the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the application to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.
Claims (10)
1. An indoor air fine particle measurement method based on the Internet of things is characterized by comprising the following steps:
constructing a topological matrix between a plurality of sensors and a position X where no sensor is deployed, wherein the value of each position on the diagonal position of the topological matrix is the distance between each sensor and the position X, and the value of each position on the off-diagonal position of the topological matrix is the distance between two corresponding sensors; passing the topological matrix through a first convolutional neural network to obtain a topological feature matrix; acquiring measurement values of the plurality of sensors; arranging the measurement values of the plurality of sensors into a measurement value vector, and multiplying the measurement value vector by the transpose of the measurement value vector to obtain a measurement value incidence matrix; passing the measured value correlation matrix through a second convolutional neural network to obtain a correlation characteristic matrix; fusing the topological characteristic matrix and the associated characteristic matrix to obtain a topological associated characteristic matrix; arranging the distances between the plurality of sensors and the position X into distance vectors, and then obtaining distance characteristic vectors through a sequence encoder comprising a one-dimensional convolutional layer and a full-connection layer; taking the distance characteristic vector as a query vector to be subjected to matrix multiplication with the topology correlation characteristic matrix to obtain a distance topology characteristic vector; constructing a modified Gaussian distribution map of the distance topological feature vector, wherein each position of the modified Gaussian distribution map is a Gaussian distribution, a mean vector of the modified Gaussian distribution map is the distance topological feature vector, and a variance vector of the modified Gaussian distribution map is generated based on the mean vector and an initial variance matrix constructed based on the mean vector; performing Gaussian discretization on the Gaussian distribution of each position of the correction Gaussian distribution map to convert the Gaussian distribution of each position of the correction Gaussian distribution map into a vector, and performing two-dimensional splicing on the vectors corresponding to each position of the correction Gaussian distribution map to obtain a Gaussian correction matrix; performing matrix multiplication on the distance topological characteristic vector and the Gaussian correction matrix to obtain a regression characteristic vector; and performing decoding regression on the regression feature vector by using a decoder to obtain a decoded value, wherein the decoded value is a measurement predicted value of the position X of the undeployed sensor.
2. The internet of things-based indoor air fine particle measurement method of claim 1, wherein the passing the topological matrix through a first convolutional neural network to obtain a topological feature matrix comprises: each layer except the last layer of the first convolutional neural network carries out convolution processing, pooling processing and activation processing on input data so as to output a topological feature map from the last second layer of the first convolutional neural network, wherein the input of the first layer of the first convolutional neural network is the topological matrix; and the last layer of the first convolutional neural network performs convolution processing, global mean pooling along channel dimensions and activation processing on the topological feature map to generate the topological feature matrix.
3. The internet of things-based indoor air fine particle measurement method of claim 2, wherein the passing the measurement value correlation matrix through a second convolutional neural network to obtain a correlation feature matrix comprises: performing convolution processing, pooling processing and activation processing on input data by each layer except the last layer of the second convolutional neural network to output a correlation feature map by the last second layer of the second convolutional neural network, wherein the input of the first layer of the second convolutional neural network is the measured value correlation matrix; the last layer of the second convolutional neural network performs convolution processing, global mean pooling along channel dimensions, and activation processing on the associated feature map to generate the associated feature matrix.
4. The internet of things-based indoor air fine particle measurement method of claim 3, wherein after the distances between the plurality of sensors and the position X are arranged into a distance vector, a distance feature vector is obtained through a sequence encoder comprising a one-dimensional convolutional layer and a full-link layer, and the method comprises the following steps: fully concatenate encoding the distance vector using a full concatenation of the sequence encoder to extract high-dimensional implicit features of feature values for each position in the distance vector using the formula
Where X is a distance vector, Y is an output vector, W is a weight matrix, B is an offset vector,
represents a matrix multiplication: andperforming one-dimensional convolutional encoding on the distance vector by using a one-dimensional convolutional layer of the sequence encoder according to the following formula so as to extract high-dimensional implicit associated features of association between feature values of all positions in the distance vector, wherein the formula is as follows:
wherein, a is the width of the convolution kernel in the x direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, and w is the size of the convolution kernel.
5. The internet of things-based indoor air fine particle measurement method of claim 4, wherein constructing the modified Gaussian distribution map of the distance topological feature vector comprises: constructing a modified Gaussian distribution map of the distance topological feature vector X to N (mu, sigma) by the following formula, wherein the mean vector mu is the distance topological feature vector, and the variance vector sigma is an initial variance matrix of Gaussian distribution by first taking a variance matrix between every two eigenvalues of the mean vector as the variance matrix
Then, the mean vector mu and the initial variance matrix are calculated
Product of (2) and (4) generating.
6. The internet of things-based indoor air fine particle measurement method of claim 5, wherein performing decoding regression on the regression feature vector using a decoder to obtain a decoded value, which is a measurement prediction value of a position X of a non-deployed sensor, comprises: decoding the regression feature vector using a plurality of fully-connected layers of the decoder to obtain the decoded value using the following formula:
where X is the regression feature vector, Y is the decoded value, W is the weight matrix, B is the offset vector,
representing the matrix multiplication, h (-) is the activation function.
7. The utility model provides an indoor air fine particle measurement system based on thing networking which characterized in that includes:
the topological matrix construction unit is used for constructing a topological matrix between a plurality of sensors and a position X where the sensors are not deployed, wherein the value of each position on the diagonal position of the topological matrix is the distance between each sensor and the position X, and the value of each position on the off-diagonal position of the topological matrix is the distance between two corresponding sensors; the first convolution unit is used for enabling the topology matrix obtained by the topology matrix building unit to pass through a first convolution neural network so as to obtain a topology characteristic matrix; a measurement value acquisition unit for acquiring measurement values of the plurality of sensors; a measurement value correlation matrix calculation unit configured to arrange the measurement values of the plurality of sensors obtained by the measurement value obtaining unit into measurement value vectors and multiply the measurement value vectors with the transpose of the measurement value vectors to obtain a measurement value correlation matrix; the second convolution unit is used for enabling the measured value incidence matrix obtained by the measured value incidence matrix calculation unit to pass through a second convolution neural network so as to obtain an incidence characteristic matrix; a fusion unit, configured to fuse the topological feature matrix obtained by the first convolution unit and the associated feature matrix obtained by the second convolution unit to obtain a topological associated feature matrix; the encoding unit is used for arranging the distances between the sensors and the position X into distance vectors and then obtaining distance characteristic vectors through a sequence encoder comprising a one-dimensional convolutional layer and a full-connection layer; the matrix multiplication unit is used for performing matrix multiplication on the distance characteristic vector obtained by the coding unit as a query vector and the topology association characteristic matrix obtained by the fusion unit to obtain a distance topology characteristic vector; a modified gaussian distribution map constructing unit, configured to construct a modified gaussian distribution map of the distance topological feature vector obtained by the matrix multiplication unit, where each position of the modified gaussian distribution map is a gaussian distribution, a mean vector of the modified gaussian distribution map is the distance topological feature vector, and a variance vector of the modified gaussian distribution map is generated based on the mean vector and an initial variance matrix constructed based on the mean vector; the Gaussian discretization unit is used for carrying out Gaussian discretization on the Gaussian distribution of each position of the correction Gaussian distribution map obtained by the correction Gaussian distribution map construction unit so as to convert the Gaussian distribution of each position of the correction Gaussian distribution map into a vector, and carrying out two-dimensional splicing on the vectors corresponding to each position of the correction Gaussian distribution map so as to obtain a Gaussian correction matrix; a regression feature vector generation unit, configured to perform matrix multiplication on the distance topology feature vector obtained by the matrix multiplication unit and the gaussian correction matrix obtained by the gaussian discretization unit to obtain a regression feature vector; and a decoding regression unit configured to perform decoding regression on the regression feature vector obtained by the regression feature vector generation unit using a decoder to obtain a decoded value, which is a measurement prediction value of a position X where no sensor is deployed.
8. The internet of things based indoor air fine particle measurement system of claim 7, wherein the first convolution unit is further configured to: each layer except the last layer of the first convolutional neural network carries out convolution processing, pooling processing and activation processing on input data so as to output a topological feature map from the last second layer of the first convolutional neural network, wherein the input of the first layer of the first convolutional neural network is the topological matrix; and the last layer of the first convolutional neural network performs convolution processing, global mean pooling along channel dimensions and activation processing on the topological feature map to generate the topological feature matrix.
9. The internet of things based indoor air fine particle measurement system of claim 7, wherein the second convolution unit is further configured to: performing convolution processing, pooling processing and activation processing on input data by each layer except the last layer of the second convolutional neural network to output a correlation feature map by the last second layer of the second convolutional neural network, wherein the input of the first layer of the second convolutional neural network is the measured value correlation matrix; the last layer of the second convolutional neural network performs convolution processing, global mean pooling along channel dimensions, and activation processing on the associated feature map to generate the associated feature matrix.
10. The internet of things based indoor air fine particle measurement system of claim 7, wherein the encoding unit is further configured to: fully concatenating the distance vector using a full concatenation of the sequence encoder to extract high-dimensional implicit features of feature values of respective positions in the distance vector in the following formula
Where X is the distance vector, Y is the output vector, W is the weight matrix, B is the offset vector,
represents matrix multiplication: and
performing one-dimensional convolutional encoding on the distance vector by using a one-dimensional convolutional layer of the sequence encoder according to the following formula to extract high-dimensional implicit association features of association between feature values of all positions in the distance vector, wherein the formula is as follows:
wherein, a is the width of the convolution kernel in the x direction, F is the parameter vector of the convolution kernel, G is the local vector matrix operated with the convolution kernel function, and w is the size of the convolution kernel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210763054.8A CN115144312A (en) | 2022-06-29 | 2022-06-29 | Indoor air fine particle measuring method and system based on Internet of things |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210763054.8A CN115144312A (en) | 2022-06-29 | 2022-06-29 | Indoor air fine particle measuring method and system based on Internet of things |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115144312A true CN115144312A (en) | 2022-10-04 |
Family
ID=83410148
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210763054.8A Withdrawn CN115144312A (en) | 2022-06-29 | 2022-06-29 | Indoor air fine particle measuring method and system based on Internet of things |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115144312A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115827257A (en) * | 2023-02-20 | 2023-03-21 | 腾云创威信息科技(威海)有限公司 | CPU capacity prediction method and system for processor system |
| CN116125133A (en) * | 2023-02-16 | 2023-05-16 | 南京博纳威电子科技有限公司 | Non-contact type current and voltage integrated measurement on-line monitoring method and system |
| CN117054968A (en) * | 2023-08-19 | 2023-11-14 | 杭州优航信息技术有限公司 | Sound source positioning system and method based on linear array microphone |
-
2022
- 2022-06-29 CN CN202210763054.8A patent/CN115144312A/en not_active Withdrawn
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116125133A (en) * | 2023-02-16 | 2023-05-16 | 南京博纳威电子科技有限公司 | Non-contact type current and voltage integrated measurement on-line monitoring method and system |
| CN116125133B (en) * | 2023-02-16 | 2023-10-20 | 南京博纳威电子科技有限公司 | Non-contact type current and voltage integrated measurement on-line monitoring method and system |
| CN115827257A (en) * | 2023-02-20 | 2023-03-21 | 腾云创威信息科技(威海)有限公司 | CPU capacity prediction method and system for processor system |
| CN115827257B (en) * | 2023-02-20 | 2023-05-05 | 腾云创威信息科技(威海)有限公司 | CPU capacity prediction method and system for processor system |
| CN117054968A (en) * | 2023-08-19 | 2023-11-14 | 杭州优航信息技术有限公司 | Sound source positioning system and method based on linear array microphone |
| CN117054968B (en) * | 2023-08-19 | 2024-03-12 | 杭州优航信息技术有限公司 | Sound source positioning system and method based on linear array microphone |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115144312A (en) | Indoor air fine particle measuring method and system based on Internet of things | |
| Guo et al. | Data‐driven flood emulation: Speeding up urban flood predictions by deep convolutional neural networks | |
| Bucak et al. | Incremental subspace learning via non-negative matrix factorization | |
| CN115356434A (en) | Gas monitoring system and monitoring method for hexafluorobutadiene storage place | |
| CN115392369A (en) | Indoor air quality prediction system based on Internet of things and prediction method thereof | |
| CN114724386B (en) | Short-time traffic flow prediction method and system under intelligent traffic and electronic equipment | |
| CN114900126B (en) | Grounding test equipment and grounding test method for solar cell module | |
| CN115171317A (en) | Internet of things smart home method and system and electronic equipment | |
| CN115169463A (en) | Indoor air quality monitoring system based on Internet of things and monitoring method thereof | |
| CN107563409A (en) | A kind of description method based on area image feature concern network with arest neighbors sequence | |
| CN115471216B (en) | Data management method of intelligent laboratory management platform | |
| CN115424204A (en) | Pedestrian detection method and system based on information fusion | |
| CN114065693A (en) | Method and system for optimizing layout of super-large-scale integrated circuit structure and electronic equipment | |
| CN112801063A (en) | Neural network system and image crowd counting method based on neural network system | |
| CN116091414A (en) | Cardiovascular image recognition method and system based on deep learning | |
| CN113837302B (en) | City safety management monitoring method and system based on Internet of things and electronic equipment | |
| CN114566190A (en) | Early warning method and system based on voice characteristics and electronic equipment thereof | |
| CN114821299A (en) | Remote sensing image change detection method | |
| CN114821169A (en) | Method-level non-intrusive call link tracking method under micro-service architecture | |
| CN114611795A (en) | Water level linkage-based water level prediction method and system | |
| CN117542122A (en) | Human body pose estimation and three-dimensional reconstruction method, network training method and device | |
| CN115661758A (en) | Public place crowd density monitoring method and system based on artificial intelligence | |
| CN116247824B (en) | Control method and system for power equipment | |
| CN112800217A (en) | Vector relevance matrix-based intelligent assessment method for government affair transaction processing report | |
| CN116467485A (en) | Video image retrieval construction system and method thereof |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20221004 |
|
| WW01 | Invention patent application withdrawn after publication |