CN117492398B - High-speed data acquisition system and acquisition method thereof - Google Patents
High-speed data acquisition system and acquisition method thereof Download PDFInfo
- Publication number
- CN117492398B CN117492398B CN202311529539.1A CN202311529539A CN117492398B CN 117492398 B CN117492398 B CN 117492398B CN 202311529539 A CN202311529539 A CN 202311529539A CN 117492398 B CN117492398 B CN 117492398B
- Authority
- CN
- China
- Prior art keywords
- signal
- training
- semantic
- feature map
- shallow
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000012545 processing Methods 0.000 claims abstract description 47
- 238000012549 training Methods 0.000 claims description 123
- 230000004927 fusion Effects 0.000 claims description 41
- 238000003062 neural network model Methods 0.000 claims description 34
- 238000004458 analytical method Methods 0.000 claims description 29
- 230000008054 signal transmission Effects 0.000 claims description 19
- 238000000605 extraction Methods 0.000 claims description 15
- 239000013598 vector Substances 0.000 claims description 12
- 238000012544 monitoring process Methods 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims 2
- 238000013527 convolutional neural network Methods 0.000 description 17
- 239000010410 layer Substances 0.000 description 16
- 230000006870 function Effects 0.000 description 13
- 238000010586 diagram Methods 0.000 description 8
- 238000011176 pooling Methods 0.000 description 7
- 230000000694 effects Effects 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000013136 deep learning model Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000003909 pattern recognition Methods 0.000 description 2
- 230000002787 reinforcement Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 230000008033 biological extinction Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 239000011229 interlayer Substances 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 230000000873 masking effect Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000002829 reductive effect Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
- G05B19/042—Programme control other than numerical control, i.e. in sequence controllers or logic controllers using digital processors
- G05B19/0423—Input/output
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/24—Pc safety
- G05B2219/24215—Scada supervisory control and data acquisition
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a high-speed data acquisition system and a high-speed data acquisition method, and relates to the field of high-speed data acquisition. The system comprises: the system comprises a first data acquisition board and a second data acquisition board, and a data processing board, wherein the first data acquisition board and the second data acquisition board are communicably connected to the data processing board, the first data acquisition board is used for receiving a first type signal, the second data acquisition board is used for receiving a second type signal, and the first type signal is different from the second type signal. In this way, the speed and effectiveness of data acquisition may be provided by analyzing the first type of signal to determine if its signal-to-noise ratio is above a predetermined threshold.
Description
Technical Field
The present application relates to the field of high-speed data acquisition, and more particularly, to a high-speed data acquisition system and an acquisition method thereof.
Background
High-speed data acquisition systems have important applications in many fields, such as scientific research, industrial monitoring, communication networks, etc., where such systems are required to be able to acquire and process different types of signals quickly and accurately and to provide high quality data.
Conventional high-speed data acquisition devices are typically extended with an x86 system architecture and a PCI, PCIE, etc. board card architecture. However, as the number of channels increases, the volume, weight, and power consumption of the device also increase, resulting in inconvenient deployment and inflexible use. In addition, the board card type framework is only suitable for laboratory environments, connection looseness is easy to cause under the external field environments of strong vibration and impact, and therefore reliability of data acquisition is affected. On the other hand, some small-volume and low-power-consumption data acquisition systems often use a single-chip microcomputer or a DSP as a main controller. Although the devices can realize basic acquisition functions, the device can only meet the limitation of single acquisition signal, low sampling frequency, low storage bandwidth and low data transmission rate, and can only meet the requirements of some low-speed acquisition application fields. Furthermore, conventional data acquisition devices typically have only one signal acquisition function per channel. If different types of signals need to be acquired, different acquisition boards or adapter plates need to be configured, so that the complexity of the system is increased, and meanwhile, the reliability of the system is reduced.
Accordingly, an optimized high-speed data acquisition system is desired.
Disclosure of Invention
The present application has been made to solve the above-mentioned technical problems. Embodiments of the present application provide a high-speed data acquisition system and method that can provide speed and effectiveness of data acquisition by analyzing a first type of signal to determine if its signal-to-noise ratio is above a predetermined threshold.
According to one aspect of the present application, there is provided a high-speed data acquisition system comprising: the system comprises a first data acquisition board and a second data acquisition board, and a data processing board, wherein the first data acquisition board and the second data acquisition board are communicably connected to the data processing board, the first data acquisition board is used for receiving a first type signal, the second data acquisition board is used for receiving a second type signal, and the first type signal is different from the second type signal.
According to another aspect of the present application, there is provided a high-speed data acquisition method comprising:
Communicatively connecting the first data acquisition board and the second data acquisition board to the data processing board;
Receiving a first type signal by the first data acquisition board; and
Receiving, by the second data acquisition board, a second type of signal, the first type of signal being different from the second type of signal;
Wherein, first data acquisition board includes: a signal-to-noise ratio analysis module;
wherein, the signal to noise ratio analysis module includes:
the signal acquisition unit is used for acquiring a first type signal acquired by the first data acquisition board;
The signal shallow feature analysis unit is used for carrying out feature extraction on the first type signal through a shallow feature extractor based on a first deep neural network model so as to obtain a signal shallow feature map;
the signal deep semantic feature capturing unit is used for extracting features of the signal shallow feature map through a semantic feature extractor based on a second deep neural network model so as to obtain a signal semantic feature map;
The signal multi-scale feature fusion unit is used for fusing the signal shallow feature map and the signal semantic feature map to obtain semantic mask enhanced signal shallow features; and
And the signal transmission judging unit is used for determining whether to transmit the first type signal to the data processing board based on the semantic mask enhanced signal shallow layer characteristics.
Compared with the prior art, the high-speed data acquisition system and the acquisition method thereof provided by the application comprise the following steps: the system comprises a first data acquisition board and a second data acquisition board, and a data processing board, wherein the first data acquisition board and the second data acquisition board are communicably connected to the data processing board, the first data acquisition board is used for receiving a first type signal, the second data acquisition board is used for receiving a second type signal, and the first type signal is different from the second type signal. In this way, the speed and effectiveness of data acquisition may be provided by analyzing the first type of signal to determine if its signal-to-noise ratio is above a predetermined threshold.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the description of the embodiments will be briefly introduced below, the following drawings not being drawn to scale with respect to actual dimensions, emphasis instead being placed upon illustrating the gist of the present application.
FIG. 1 is a block diagram of a high-speed data acquisition system according to an embodiment of the present application.
Fig. 2 is a schematic block diagram of the signal-to-noise ratio analysis module in the high-speed data acquisition system according to an embodiment of the application.
Fig. 3 is a block diagram schematically illustrating the signal transmission judging unit in the high-speed data acquisition system according to an embodiment of the present application.
Fig. 4 is a flowchart of a high-speed data acquisition method according to an embodiment of the present application.
Fig. 5 is a schematic diagram of a system architecture of sub-step S120 of the high-speed data acquisition method according to an embodiment of the present application.
Fig. 6 is an application scenario diagram of a high-speed data acquisition system according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some, but not all embodiments of the application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are also within the scope of the application.
As used in the specification and in the claims, the terms "a," "an," "the," and/or "the" are not specific to a singular, but may include a plurality, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that the steps and elements are explicitly identified, and they do not constitute an exclusive list, as other steps or elements may be included in a method or apparatus.
Although the present application makes various references to certain modules in a system according to embodiments of the present application, any number of different modules may be used and run on a user terminal and/or server. The modules are merely illustrative, and different aspects of the systems and methods may use different modules.
A flowchart is used in the present application to describe the operations performed by a system according to embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in order precisely. Rather, the various steps may be processed in reverse order or simultaneously, as desired. Also, other operations may be added to or removed from these processes.
Hereinafter, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
Aiming at the technical problems, in the technical scheme of the application, a high-speed data acquisition system is provided. FIG. 1 is a block diagram of a high-speed data acquisition system according to an embodiment of the present application. As shown in fig. 1, a high-speed data acquisition system 100 according to an embodiment of the present application includes: a first data acquisition board 110 and a second data acquisition board 120, and a data processing board 130, wherein the first data acquisition board 110 and the second data acquisition board 120 are communicatively coupled to the data processing board 130, the first data acquisition board 110 is configured to receive a first type of signal and the second data acquisition board 120 is configured to receive a second type of signal, the first type of signal being different from the second type of signal.
Accordingly, consider that in a high-speed data acquisition system, the signal-to-noise ratio of a signal is an important indicator that represents the ratio of useful information to noise in the signal. The higher the signal-to-noise ratio, the stronger the useful information in the signal relative to the noise, and the higher the quality of the data acquisition. Therefore, improving the signal-to-noise ratio is critical to improving the performance of the data acquisition system. Based on the above, the technical idea of the application is to install a signal-to-noise ratio analysis module in the first data acquisition board. That is, after the first data acquisition board receives the first type signal, the signal to noise ratio analysis module analyzes the first type signal to determine whether the signal to noise ratio is higher than a preset threshold, if so, the first type signal is acquired, and if not, the first type signal is discarded, in this way, the speed and the effectiveness of data acquisition are provided. In this way, a more efficient and reliable data acquisition system is provided, which eliminates the limitations of traditional equipment, improves the accuracy and efficiency of data acquisition, and has better flexibility and adaptability.
Specifically, in the technical scheme of the application, first, the first type signal acquired by the first data acquisition board is acquired. It should be appreciated that during data acquisition, the first type of signal may contain a significant amount of noise and redundant information that may interfere with subsequent signal processing and analysis tasks. Therefore, in order to reduce the influence of noise and extract useful features, feature extraction of signals is required. Since convolutional neural networks are a deep learning model that is widely used for image processing and pattern recognition, it can effectively extract a characteristic representation of a signal. Based on this, in the technical solution of the present application, the first type signal is passed through a shallow feature extractor based on a first convolutional neural network model to obtain a signal shallow feature map, where the signal shallow feature map is a result obtained by performing preliminary feature extraction on the signal, and includes some basic feature information of the signal, such as edges, textures, and so on, so that the subsequent signal processing and analysis tasks can be helped to better understand and utilize the information of the signal.
It is then contemplated that in signal processing and analysis tasks, higher level semantic information of the signal may not be captured by virtue of shallow features alone. For a better understanding of the meaning and context of the signal, a deeper level of feature extraction is required. Therefore, in the technical scheme of the application, the signal shallow feature map is further processed through a semantic feature extractor based on a second convolutional neural network model to obtain a signal semantic feature map. It should be appreciated that by processing with the semantic feature extractor, more complex structures, patterns, and relationships in the signal can be captured, helping to understand the meaning of the signal more accurately.
Further, as the signal shallow feature map and the signal semantic feature map capture different levels of feature information of the first type signal respectively. By fusing the two layers of features, the advantages of the features can be comprehensively utilized, and the expression capability and the discrimination of the features are improved, so that the signal-to-noise ratio detection and judgment of signals can be more accurately carried out. Based on the above, in the technical scheme of the application, a residual information enhancement fusion module is further used for fusing the signal shallow feature map and the signal semantic feature map to obtain a semantic mask enhancement signal shallow feature map. It should be appreciated that the residual connection is a jump connection that allows the network to learn more easily the residual feature information in the shallow signal feature map and the signal semantic feature map, i.e. the differences between the two feature maps. Through residual connection, the signal semantic feature map can be utilized to carry out semantic mask reinforcement on the signal shallow feature map so as to enhance the semantic representation capability of the signal shallow feature map, that is, the information enhancement of the signal semantic feature map can be fused into the signal shallow feature map by introducing the residual connection mode, so that the signal shallow feature map focuses more on and emphasizes mask reinforcement shallow feature information related to the first type of signal.
And then, the semantic mask enhanced signal shallow feature map is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the signal-to-noise ratio of the first type signal exceeds a preset threshold value. That is, the signal-to-noise ratio of the first type signal is compared with a predetermined threshold to determine whether the signal-to-noise ratio is above the predetermined threshold by masking the enhanced signal shallow feature information with the semantic features of the first type signal for classification. In particular, transmitting the first type of signal to the data processing board in response to the signal-to-noise ratio of the first type of signal being above the predetermined threshold; discarding the first type signal in response to the signal-to-noise ratio of the first type signal being below the predetermined threshold. In this way, the speed and the effectiveness of data acquisition can be improved, the limitations of traditional equipment are abandoned, and a more efficient and reliable data acquisition system is provided.
Accordingly, the first data acquisition board 110 includes: a signal-to-noise ratio analysis module 111; as shown in fig. 2, the signal-to-noise ratio analysis module 111 includes: a signal acquisition unit 1111 for acquiring a first type signal acquired by the first data acquisition board; a signal shallow feature analysis unit 1112, configured to perform feature extraction on the first type signal by using a shallow feature extractor based on the first deep neural network model to obtain a signal shallow feature map; a signal deep semantic feature capturing unit 1113, configured to perform feature extraction on the signal shallow feature map by using a semantic feature extractor based on a second deep neural network model to obtain a signal semantic feature map; a signal multi-scale feature fusion unit 1114, configured to fuse the signal shallow feature map and the signal semantic feature map to obtain a semantic mask enhanced signal shallow feature; and a signal transmission judging unit 1115, configured to determine whether to transmit the first type signal to the data processing board based on the semantic mask enhanced signal shallow feature.
It should be appreciated that the function of the signal acquisition unit 1111 is to acquire signal data from external acquisition devices or sensors for subsequent processing and analysis. The function of the signal shallow feature analysis unit 1112 is to extract low-level features in the signal, such as edges, textures, etc., for subsequent deep feature extraction and analysis. The function of the signal deep semantic feature capturing unit 1113 is to extract high-level semantic information in the signal, such as objects, structures, etc., for deeper analysis and understanding of the signal. The function of the signal multi-scale feature fusion unit 1114 is to fuse low-level features with high-level semantic features to improve signal understanding and analysis capabilities. The signal transmission judging unit 1115 is used for judging whether the signal needs to be transmitted to the data processing board for further processing and analysis according to the signal characteristics obtained through analysis. Each unit plays different roles and functions in the signal-to-noise ratio analysis module, and the tasks of signal acquisition, feature extraction, feature fusion, transmission judgment and the like are completed together, so that the analysis and processing capacity of the signals is improved.
The first deep neural network model is a first convolutional neural network model, and the second deep neural network model is a second convolutional neural network model. It is worth mentioning that convolutional neural network (Convolutional Neural Network, CNN) is a deep learning model, especially suitable for processing data with a grid structure, such as images and audio. The core idea of the convolutional neural network is to extract and learn the characteristics of data through components such as a convolutional layer, a pooling layer, a fully connected layer and the like. These hierarchies enable convolutional neural networks to effectively capture local relationships and spatial structures in the input data. Specifically, the convolution layer extracts different features of the input data by convolving the input data with a set of learnable filters (also referred to as convolution kernels). The convolution operation can be regarded as a locally perceived operation that is computed on the input data by means of a sliding filter, resulting in a feature map. The pooling layer is used for reducing the size of the feature map and reducing the data dimension so as to reduce the calculation amount of the model. Common pooling operations include maximum pooling (selecting the maximum value in a local area) and average pooling (calculating the average value of a local area). The fully connected layer is used to connect the outputs of the convolutional and pooling layers to the final output layer for classification, regression, or other tasks. Neurons in the fully connected layer are connected to all neurons in the previous layer, and the inputs are linearly transformed and non-linearly activated by learning weights and biases. The training of convolutional neural networks is achieved by back-propagation algorithms and gradient descent optimization, which enable automatic learning of the feature representation of the data and end-to-end training on large-scale data sets. In summary, convolutional neural networks are deep learning models based on hierarchical structures such as convolution, pooling, full connection, and the like, are specially used for processing data with a grid structure, and can achieve excellent performance in computer vision and pattern recognition tasks through feature representation of the learning data.
Wherein the signal multi-scale feature fusion unit 1114 is configured to: and fusing the signal shallow feature map and the signal semantic feature map by using a residual information enhancement fusion module to obtain a semantic mask enhanced signal shallow feature map as the semantic mask enhanced signal shallow feature. It is worth mentioning that the residual information enhancement fusion module is a method for signal multi-scale feature fusion, and aims to improve the effect and performance of feature fusion. The function of the module is to fuse the shallow feature map of the signal with the semantic feature map by introducing residual connection, so as to obtain the signal shallow feature map with enhanced semantic mask as a part of the enhanced signal of the semantic mask. Residual connection is a cross-level connection mode, and residual information is transferred to subsequent levels by adding a feature map of a previous layer to a feature map of a subsequent layer. This way of connection can help the network learn better about the feature representation and alleviate the problem of gradient extinction. In the signal multi-scale feature fusion unit, a residual information enhancement fusion module fuses a shallow feature map and a semantic feature map of a signal. The specific steps may be to add the two feature maps element by element to obtain a fused feature map. The advantage of this is that the detail information of shallow features can be preserved, and the advanced semantic information of semantic features is combined, so that richer and more accurate feature representation is obtained. The semantic mask enhanced signal shallow feature map is a result obtained by a residual information enhanced fusion module. The method integrates shallow features and semantic features of signals, and has better semantic understanding and feature expression capability. Such feature maps may be used for subsequent signal analysis and processing tasks such as object detection, image segmentation, etc. Therefore, the residual information enhancement fusion module plays a key role in signal multi-scale feature fusion, and feature information of different levels is effectively fused by introducing residual connection, so that the effect and performance of feature fusion are improved, and the signal shallow features with more semantic understanding capability are obtained.
As shown in fig. 3, the signal transmission determining unit 1115 includes: a signal-to-noise ratio threshold monitoring subunit 11151, configured to pass the semantic mask enhanced signal shallow feature map through a classifier to obtain a classification result, where the classification result is used to indicate whether the signal-to-noise ratio of the first type signal exceeds a predetermined threshold; and a signal transmission sub-unit 11152 for determining whether to transmit the first type signal to the data processing board based on the classification result. It should be understood that the signal transmission determining unit 1115 is a module for determining signal transmission, and includes a signal-to-noise ratio threshold monitoring subunit 11151 and a signal transmission subunit 11152. The snr threshold monitoring subunit 11151 is configured to classify the shallow feature map of the semantic mask enhanced signal by a classifier to obtain a classification result, where the classification result is used to indicate whether the snr of the first type signal exceeds a predetermined threshold. The signal-to-noise ratio is an index for measuring the ratio between useful information and noise in the signal and is used for evaluating the quality of the signal, and the signal is classified by a classifier, so that whether the quality of the signal meets the preset signal-to-noise ratio threshold requirement can be judged. The signal transmission subunit 11152 is configured to determine whether to transmit the first type of signal to the data processing board based on the classification result of the signal-to-noise ratio threshold monitoring subunit, and according to the classification result, if the signal-to-noise ratio exceeds a predetermined threshold, it indicates that the signal quality is good, and may transmit the signal to the data processing board for subsequent processing, otherwise, if the signal-to-noise ratio is below the predetermined threshold, it indicates that the signal quality is poor, there may be more noise interference, and may choose not to transmit the signal or take other processing modes. In summary, the signal transmission determining unit 1115 is configured to determine whether to transmit the first type signal to the data processing board according to the quality of the semantic mask enhanced signal shallow feature map. And evaluating and classifying the signal quality through the signal-to-noise ratio threshold monitoring subunit, and then determining whether to transmit the signal or not by the signal transmission subunit according to the classification result. In this way, signal transmission and processing can be selectively performed according to the quality of the signal, so as to improve the efficiency and performance of the system.
Further, in the technical scheme of the application, the high-speed data acquisition system further comprises a training module for training the shallow feature extractor based on the first convolutional neural network model, the semantic feature extractor based on the second convolutional neural network model, the residual information enhancement fusion module and the classifier. It should be appreciated that the training module plays an important role in a high-speed data acquisition system. It is used to train shallow feature extractors, semantic feature extractors, residual information enhancement fusion modules, and classifiers to enable them to efficiently process and analyze signal data. The training module has the following main functions: 1. training a shallow feature extractor and a semantic feature extractor: the shallow feature extractor and the semantic feature extractor are components based on a convolutional neural network model and are used for extracting shallow features and semantic features of signal data. The training module enables the feature extractor to learn a feature representation method suitable for the signal data by training the feature extractor. Through a large amount of training data and corresponding labels, the training module can optimize network parameters so that the feature extractor can accurately capture useful features in the signal data. 2. Training residual information enhancement fusion module: the residual information enhancement fusion module is a key component for signal multi-scale feature fusion. The training module can effectively fuse shallow features and semantic features of signals by training the residual information enhancement fusion module, and the effect and performance of feature fusion are improved. Through the training module, parameters of the residual information enhancement fusion module can be optimized, so that the correlation and importance among the features can be better captured. 3. Training a classifier: the classifier is used for classifying the semantic mask enhanced signal shallow feature map to determine whether the quality of the signal exceeds a predetermined signal-to-noise threshold. The training module trains the classifier to accurately classify the feature according to the features of the feature map. By providing a large number of training samples and corresponding labels, the training module can optimize the parameters of the classifier so that the classifier can accurately classify and judge the signals. Through the training process of the training module, the shallow feature extractor, the semantic feature extractor, the residual information enhancement fusion module and the classifier can have good feature extraction and classification capabilities, so that the processing and analysis capabilities of the high-speed data acquisition system on signals are improved.
Wherein, in one example, the training module comprises: the training signal acquisition unit is used for acquiring training data, wherein the training data comprises a first type of training signal acquired by the first data acquisition board and a true value of whether the signal-to-noise ratio of the first type of training signal exceeds a preset threshold value; the training signal shallow feature analysis unit is used for carrying out feature extraction on the first type training signal through a shallow feature extractor based on a first deep neural network model so as to obtain a training signal shallow feature map; the training signal deep semantic feature capturing unit is used for extracting features of the training signal shallow feature map through a semantic feature extractor based on a second deep neural network model so as to obtain a training signal semantic feature map; the training signal multi-scale feature fusion unit is used for fusing the training signal shallow feature map and the training signal semantic feature map by using a residual information enhancement fusion module so as to obtain a training semantic mask enhancement signal shallow feature map; the training optimization unit is used for optimizing the training semantic mask enhanced signal shallow feature map to obtain an optimized training semantic mask enhanced signal shallow feature map; the training loss unit is used for enabling the shallow feature map of the optimized training semantic mask enhanced signal to pass through a classifier to obtain a classification loss function value; and the training signal transmission judging unit is used for training the shallow feature extractor based on the first convolutional neural network model, the semantic feature extractor based on the second convolutional neural network model, the residual information enhancement fusion module and the classifier based on the classification loss function value.
In particular, in the technical scheme of the application, the training signal shallow feature map and the training signal semantic feature map respectively express shallow image semantic features and deep image semantic features of the signal waveform image of the first type training signal, and under the condition of fusion by using a residual information enhancement fusion module, the overall image semantic feature expression effect of the training semantic mask enhancement signal shallow feature map is improved based on residual semantic feature distribution between the shallow image semantic features and the deep image semantic features, but new image semantic feature expression dimensions are introduced, namely, interlayer residual error expression dimensions are further introduced outside the shallow expression dimensions and the deep expression dimensions of the image semantic features, so that the training semantic mask enhancement signal shallow feature map is subjected to multi-dimensional sparsification of the image semantic features overall, and the training semantic mask enhancement signal shallow feature map has poor convergence of probability density distribution of class regression probability when subjected to classification regression through a classifier, and influences the accuracy of classification regression effect and classification result. Therefore, preferably, the training semantic mask enhanced signal shallow feature vector obtained after the training semantic mask enhanced signal shallow feature vector is unfolded is optimized when the training semantic mask enhanced signal shallow feature map is classified by a classifier.
Accordingly, in one example, the training optimization unit is further configured to: optimizing the training semantic mask enhanced signal shallow feature map by using the following optimization formula to obtain the optimized training semantic mask enhanced signal shallow feature map; wherein, the optimization formula is:
Wherein V is the training semantic mask enhanced signal shallow feature vector obtained after the training semantic mask enhanced signal shallow feature map is unfolded, V i and V j are the ith and jth feature values of the training semantic mask enhanced signal shallow feature vector, and The method is characterized in that the method is an overall feature mean value of the training semantic mask enhancement signal shallow feature vector, exp { · } represents an exponential operation of a numerical value, the exponential operation of the numerical value represents a natural exponential function value which takes the numerical value as a power, v' i is an ith feature value of the optimized training semantic mask enhancement signal shallow feature vector obtained after the development of the optimized training semantic mask enhancement signal shallow feature map.
Specifically, aiming at the fact that the local probability density of probability density distribution in a probability space caused by sparse distribution of the training semantic mask enhancement signal shallow feature vector in a high-dimensional feature space is not matched, the regularized global self-consistent class coding is used for simulating the global self-consistent relation of class coding behaviors of the high-dimensional feature manifold of the training semantic mask enhancement signal shallow feature vector in the probability space so as to adjust the error landscape of the feature manifold in a high-dimensional open space domain, and the self-consistent matching type class coding of the high-dimensional feature manifold of the training semantic mask enhancement signal shallow feature vector to the explicit probability space is realized, so that the convergence of the probability density distribution of the class regression probability of the training semantic mask enhancement signal shallow feature vector is improved, and the classification regression effect and classification result accuracy are improved. Therefore, the signal transmission processing judgment can be performed based on the signal analysis and the signal-to-noise ratio comparison, a more efficient and reliable data acquisition system is provided in such a way, the limitation of the traditional data acquisition equipment is abandoned, the accuracy and the efficiency of data acquisition are improved, and the system has better flexibility and adaptability.
Further, the training loss unit is configured to: processing the optimized training semantic mask enhanced signal shallow feature map by using the classifier according to the following training classification formula to obtain a training classification result; wherein, training classification formula is: softmax { (W n,Bn):…:(W1,B1) |project (F) }; project (F) represents projecting the shallow feature map of the optimized training semantic mask enhanced signal into vectors, W 1 to W n are weight matrices of all the connection layers, and B 1 to B n represent bias matrices of all the connection layers; and calculating a cross entropy value between the training classification result and the true value as the classification loss function value.
In summary, a high-speed data acquisition system 100 in accordance with embodiments of the present application is illustrated that may improve the accuracy and efficiency of data acquisition.
As described above, the high-speed data collection system 100 according to the embodiment of the present application may be implemented in various terminal devices, for example, a server or the like having a high-speed data collection algorithm according to the embodiment of the present application. In one example, the high-speed data acquisition system 100 according to embodiments of the present application may be integrated into the terminal device as a software module and/or hardware module. For example, the high-speed data acquisition system 100 according to the embodiment of the present application may be a software module in the operating system of the terminal device, or may be an application program developed for the terminal device; of course, the high-speed data acquisition system 100 according to the embodiment of the present application may also be one of a plurality of hardware modules of the terminal device.
Alternatively, in another example, the high-speed data acquisition system 100 according to an embodiment of the present application and the terminal device may be separate devices, and the high-speed data acquisition system 100 may be connected to the terminal device through a wired and/or wireless network and transmit interactive information in a contracted data format.
Fig. 4 is a flowchart of a high-speed data acquisition method according to an embodiment of the present application. As shown in fig. 4, a high-speed data acquisition method according to an embodiment of the present application includes: s110, the first data acquisition board and the second data acquisition board are connected to the data processing board in a communication mode; s120, receiving a first type signal through the first data acquisition board; and S130, receiving a second type signal through the second data acquisition board, wherein the first type signal is different from the second type signal.
Fig. 5 is a schematic diagram of a system architecture of sub-step S120 of the high-speed data acquisition method according to an embodiment of the present application. As shown in fig. 5, in a specific example, in the above-described high-speed data acquisition method, receiving, by the first data acquisition board, a first type of signal includes: acquiring a first type signal acquired by the first data acquisition board; performing feature extraction on the first type signal by a shallow feature extractor based on a first deep neural network model to obtain a signal shallow feature map; extracting features of the signal shallow feature map through a semantic feature extractor based on a second deep neural network model to obtain a signal semantic feature map; fusing the signal shallow feature map and the signal semantic feature map to obtain semantic mask enhanced signal shallow features; and determining whether to transmit the first type of signal to the data processing board based on the semantic mask enhanced signal shallow features.
Here, it will be understood by those skilled in the art that the specific operations of the respective steps in the above-described high-speed data acquisition method have been described in detail in the above description of the high-speed data acquisition system 100 with reference to fig. 1 to 3, and thus, repetitive descriptions thereof will be omitted.
Fig. 6 is an application scenario diagram of a high-speed data acquisition system according to an embodiment of the present application. As shown in fig. 6, in this application scenario, first, a first type signal (e.g., D illustrated in fig. 6) acquired by a first data acquisition board is acquired, and then the first type signal is input into a server (e.g., S illustrated in fig. 6) deployed with a high-speed data acquisition algorithm, wherein the server is capable of processing the first type signal using the high-speed data acquisition algorithm to obtain a classification result indicating whether a signal-to-noise ratio of the first type signal exceeds a predetermined threshold, and then, based on the classification result, it is determined whether to transmit the first type signal to the data processing board.
The application uses specific words to describe embodiments of the application. Reference to "a first/second embodiment," "an embodiment," and/or "some embodiments" means that a particular feature, structure, or characteristic is associated with at least one embodiment of the application. Thus, it should be emphasized and should be appreciated that two or more references to "an embodiment" or "one embodiment" or "an alternative embodiment" in various positions in this specification are not necessarily referring to the same embodiment. Furthermore, certain features, structures, or characteristics of one or more embodiments of the application may be combined as suitable.
Furthermore, those skilled in the art will appreciate that the various aspects of the application are illustrated and described in the context of a number of patentable categories or circumstances, including any novel and useful procedures, machines, products, or materials, or any novel and useful modifications thereof. Accordingly, aspects of the application may be performed entirely by hardware, entirely by software (including firmware, resident software, micro-code, etc.) or by a combination of hardware and software. The above hardware or software may be referred to as a "data block," module, "" engine, "" unit, "" component, "or" system. Furthermore, aspects of the application may take the form of a computer product, comprising computer-readable program code, embodied in one or more computer-readable media.
Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
The foregoing is illustrative of the present application and is not to be construed as limiting thereof. Although a few exemplary embodiments of this application have been described, those skilled in the art will readily appreciate that many modifications are possible in the exemplary embodiments without materially departing from the novel teachings and advantages of this application. Accordingly, all such modifications are intended to be included within the scope of this application as defined in the following claims. It is to be understood that the foregoing is illustrative of the present application and is not to be construed as limited to the specific embodiments disclosed, and that modifications to the disclosed embodiments, as well as other embodiments, are intended to be included within the scope of the appended claims. The application is defined by the claims and their equivalents.
Claims (3)
1. A high-speed data acquisition system, comprising: a first data acquisition board and a second data acquisition board, and a data processing board, wherein the first data acquisition board and the second data acquisition board are communicatively connected to the data processing board, the first data acquisition board is configured to receive a first type of signal, the second data acquisition board is configured to receive a second type of signal, the first type of signal being different from the second type of signal;
Wherein, first data acquisition board includes: a signal-to-noise ratio analysis module;
wherein, the signal to noise ratio analysis module includes:
the signal acquisition unit is used for acquiring a first type signal acquired by the first data acquisition board;
The signal shallow feature analysis unit is used for carrying out feature extraction on the first type signal through a shallow feature extractor based on a first deep neural network model so as to obtain a signal shallow feature map;
the signal deep semantic feature capturing unit is used for extracting features of the signal shallow feature map through a semantic feature extractor based on a second deep neural network model so as to obtain a signal semantic feature map;
the signal multi-scale feature fusion unit is used for fusing the signal shallow feature map and the signal semantic feature map to obtain semantic mask enhanced signal shallow features;
And a signal transmission judging unit configured to determine whether to transmit the first type signal to the data processing board based on the semantic mask enhanced signal shallow feature;
the first depth neural network model is a first convolution neural network model, and the second depth neural network model is a second convolution neural network model;
Wherein, the signal multiscale feature fusion unit includes:
A residual information enhancement fusion module is used for fusing the signal shallow feature map and the signal semantic feature map to obtain a semantic mask enhancement signal shallow feature map as the semantic mask enhancement signal shallow feature;
wherein the signal transmission judging unit includes:
The signal-to-noise ratio threshold monitoring subunit is used for enabling the semantic mask enhanced signal shallow feature map to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the signal-to-noise ratio of the first type signal exceeds a preset threshold;
and a signal transmission subunit configured to determine whether to transmit the first type signal to the data processing board based on the classification result;
the high-speed data acquisition system further comprises a training module for training the shallow feature extractor based on the first deep neural network model, the semantic feature extractor based on the second deep neural network model, the residual information enhancement fusion module and the classifier;
wherein, training module includes:
The training signal acquisition unit is used for acquiring training data, wherein the training data comprises a first type of training signal acquired by the first data acquisition board and a true value of whether the signal-to-noise ratio of the first type of training signal exceeds a preset threshold value;
The training signal shallow feature analysis unit is used for carrying out feature extraction on the first type training signal through a shallow feature extractor based on a first deep neural network model so as to obtain a training signal shallow feature map;
the training signal deep semantic feature capturing unit is used for extracting features of the training signal shallow feature map through a semantic feature extractor based on a second deep neural network model so as to obtain a training signal semantic feature map;
the training signal multi-scale feature fusion unit is used for fusing the training signal shallow feature map and the training signal semantic feature map by using a residual information enhancement fusion module so as to obtain a training semantic mask enhancement signal shallow feature map;
the training optimization unit is used for optimizing the training semantic mask enhanced signal shallow feature map to obtain an optimized training semantic mask enhanced signal shallow feature map;
the training loss unit is used for enabling the shallow feature map of the optimized training semantic mask enhanced signal to pass through a classifier to obtain a classification loss function value;
and the training signal transmission judging unit is used for training the shallow layer feature extractor based on the first depth neural network model, the semantic feature extractor based on the second depth neural network model, the residual information enhancement fusion module and the classifier based on the classification loss function value.
2. The high-speed data acquisition system of claim 1, wherein the training loss unit is configured to:
Processing the optimized training semantic mask enhanced signal shallow feature map by using the classifier according to the following training classification formula to obtain a training classification result; wherein, training classification formula is:
；
Wherein, Representing the projection of the optimization training semantic mask enhanced signal shallow feature map as a vector,/>To the point ofFor the weight matrix of each layer of full-connection layer,/>To/>Representing the bias matrix of each fully connected layer;
and calculating a cross entropy value between the training classification result and the true value as the classification loss function value.
3. A high-speed data acquisition method, comprising:
Communicatively connecting the first data acquisition board and the second data acquisition board to the data processing board;
Receiving a first type signal by the first data acquisition board;
and receiving, by the second data acquisition board, a second type of signal, the first type of signal being different from the second type of signal;
wherein receiving, by the first data acquisition board, a first type of signal comprises:
Acquiring a first type signal acquired by the first data acquisition board;
performing feature extraction on the first type signal by a shallow feature extractor based on a first deep neural network model to obtain a signal shallow feature map;
extracting features of the signal shallow feature map through a semantic feature extractor based on a second deep neural network model to obtain a signal semantic feature map;
Fusing the signal shallow feature map and the signal semantic feature map to obtain semantic mask enhanced signal shallow features;
and determining whether to transmit the first type of signal to the data processing board based on the semantic mask enhanced signal shallow features;
the first depth neural network model is a first convolution neural network model, and the second depth neural network model is a second convolution neural network model;
The method for merging the signal shallow feature map and the signal semantic feature map to obtain the semantic mask enhanced signal shallow feature comprises the following steps:
A residual information enhancement fusion module is used for fusing the signal shallow feature map and the signal semantic feature map to obtain a semantic mask enhancement signal shallow feature map as the semantic mask enhancement signal shallow feature;
wherein determining whether to transmit the first type of signal to the data processing board based on the semantic mask enhanced signal shallow features comprises:
Passing the semantic mask enhanced signal shallow feature map through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the signal-to-noise ratio of the first type signal exceeds a preset threshold;
and determining whether to transmit the first type signal to the data processing board based on the classification result;
The high-speed data acquisition method further comprises the following steps: the training module is used for training the shallow feature extractor based on the first deep neural network model, the semantic feature extractor based on the second deep neural network model, the residual information enhancement fusion module and the classifier;
wherein, training module includes:
Acquiring training data, wherein the training data comprises a first type training signal acquired by the first data acquisition board and a true value of whether the signal-to-noise ratio of the first type training signal exceeds a preset threshold value;
performing feature extraction on the first type training signals through a shallow feature extractor based on a first deep neural network model to obtain a training signal shallow feature map;
Extracting features of the training signal shallow feature map through a semantic feature extractor based on a second deep neural network model to obtain a training signal semantic feature map;
a residual information enhancement fusion module is used for fusing the training signal shallow feature map and the training signal semantic feature map to obtain a training semantic mask enhancement signal shallow feature map;
optimizing the training semantic mask enhanced signal shallow feature map to obtain an optimized training semantic mask enhanced signal shallow feature map;
The shallow feature map of the optimized training semantic mask enhanced signal passes through a classifier to obtain a classification loss function value;
and training the shallow feature extractor based on the first deep neural network model, the semantic feature extractor based on the second deep neural network model, the residual information enhancement fusion module and the classifier based on the classification loss function value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311529539.1A CN117492398B (en) | 2023-11-16 | 2023-11-16 | High-speed data acquisition system and acquisition method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311529539.1A CN117492398B (en) | 2023-11-16 | 2023-11-16 | High-speed data acquisition system and acquisition method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117492398A CN117492398A (en) | 2024-02-02 |
| CN117492398B true CN117492398B (en) | 2024-05-28 |
Family
ID=89676248
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311529539.1A Active CN117492398B (en) | 2023-11-16 | 2023-11-16 | High-speed data acquisition system and acquisition method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117492398B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN2867358Y (en) * | 2006-01-27 | 2007-02-07 | 中国科学院上海光学精密机械研究所 | Laser surveying dynamic range expanded device |
| CN113239954A (en) * | 2021-04-01 | 2021-08-10 | 河海大学 | Attention mechanism-based image semantic segmentation feature fusion method |
| CN114119993A (en) * | 2021-10-30 | 2022-03-01 | 南京理工大学 | Salient object detection method based on self-attention mechanism |
| CN116310563A (en) * | 2023-03-22 | 2023-06-23 | 广州微明信息科技有限公司 | Noble metal inventory management method and system |
| CN116343015A (en) * | 2023-04-10 | 2023-06-27 | 阜外华中心血管病医院 | Medical food water content measurement system based on artificial intelligence |
| CN116760491A (en) * | 2023-07-19 | 2023-09-15 | 杭州电子科技大学 | Signal-to-noise ratio estimation method based on deep learning |
-
2023
- 2023-11-16 CN CN202311529539.1A patent/CN117492398B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN2867358Y (en) * | 2006-01-27 | 2007-02-07 | 中国科学院上海光学精密机械研究所 | Laser surveying dynamic range expanded device |
| CN113239954A (en) * | 2021-04-01 | 2021-08-10 | 河海大学 | Attention mechanism-based image semantic segmentation feature fusion method |
| CN114119993A (en) * | 2021-10-30 | 2022-03-01 | 南京理工大学 | Salient object detection method based on self-attention mechanism |
| CN116310563A (en) * | 2023-03-22 | 2023-06-23 | 广州微明信息科技有限公司 | Noble metal inventory management method and system |
| CN116343015A (en) * | 2023-04-10 | 2023-06-27 | 阜外华中心血管病医院 | Medical food water content measurement system based on artificial intelligence |
| CN116760491A (en) * | 2023-07-19 | 2023-09-15 | 杭州电子科技大学 | Signal-to-noise ratio estimation method based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117492398A (en) | 2024-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108388927B (en) | Small sample polarization SAR terrain classification method based on deep convolution twin network | |
| CN110166484A (en) | A kind of industrial control system intrusion detection method based on LSTM-Attention network | |
| CN108229588B (en) | Machine learning identification method based on deep learning | |
| CN112149524B (en) | Radar signal sorting and identifying method, device, detector and storage medium | |
| CN110675395A (en) | Intelligent on-line monitoring method for power transmission line | |
| CN113139594B (en) | Self-adaptive detection method for airborne image unmanned aerial vehicle target | |
| CN111062423B (en) | Point cloud classification method of point cloud graph neural network based on self-adaptive feature fusion | |
| CN114896887B (en) | Frequency-using equipment radio frequency fingerprint identification method based on deep learning | |
| CN104751186A (en) | Iris image quality classification method based on BP (back propagation) network and wavelet transformation | |
| CN112288700A (en) | Rail defect detection method | |
| CN109446897B (en) | Scene recognition method and device based on image context information | |
| CN113963170A (en) | RGBD image saliency detection method based on interactive feature fusion | |
| CN103310235A (en) | Steganalysis method based on parameter identification and estimation | |
| CN112308148A (en) | Defect category identification and twin neural network training method, device and storage medium | |
| CN103177264B (en) | The image classification method that view-based access control model dictionary Global Topological is expressed | |
| CN111310719A (en) | Unknown radiation source individual identification and detection method | |
| CN112488043B (en) | Unmanned aerial vehicle target detection method based on edge intelligence | |
| CN108229692B (en) | Machine learning identification method based on dual contrast learning | |
| CN117034123B (en) | Fault monitoring system and method for fitness equipment | |
| CN117492398B (en) | High-speed data acquisition system and acquisition method thereof | |
| CN116523711A (en) | Education supervision system and method based on artificial intelligence | |
| CN116402777B (en) | Power equipment detection method and system based on machine vision | |
| CN108345943B (en) | Machine learning identification method based on embedded coding and contrast learning | |
| CN115143128B (en) | Fault diagnosis method and system for small-sized submersible electric pump | |
| CN108229693B (en) | Machine learning identification device and method based on comparison learning |
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 |