CN110519233B - Satellite-borne sensor network data compression method based on artificial intelligence - Google Patents
Satellite-borne sensor network data compression method based on artificial intelligence Download PDFInfo
- Publication number
- CN110519233B CN110519233B CN201910704493.XA CN201910704493A CN110519233B CN 110519233 B CN110519233 B CN 110519233B CN 201910704493 A CN201910704493 A CN 201910704493A CN 110519233 B CN110519233 B CN 110519233B
- Authority
- CN
- China
- Prior art keywords
- data
- network
- time
- sensor
- training
- 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
Images
Classifications
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L69/00—Network arrangements, protocols or services independent of the application payload and not provided for in the other groups of this subclass
- H04L69/04—Protocols for data compression, e.g. ROHC
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W84/00—Network topologies
- H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks
Abstract
The invention provides an artificial intelligence-based satellite-borne sensor network data compression method, which comprises the following steps: acquiring sensor data at a plurality of terminal sensing nodes of a satellite-borne sensing network within a preset time period; preprocessing sensor data to eliminate abnormal data, completing missing data by a two-dimensional linear interpolation method, and mapping acquired data to a [ -1, 1] interval in a normalized mode; constructing an A-CCR network; extracting a data time sequence of one category from the preprocessed sensor data, dividing the data time sequence into m sections and randomly disordering the data time sequence, and performing loop iterative training on the A-CCR network according to the disordering sequence; extracting data time sequences of various categories from the preprocessed sensor data, dividing the data time sequences into n sections and randomly disordering the data time sequences, and performing loop iterative training on the A-CCR network initial model according to the disordering sequence; and inputting the satellite-borne sensor network data to be processed into an A-CCR network optimization model for compression processing.
Description
Technical Field
The invention relates to the technical field of aerospace and artificial intelligence, in particular to a satellite-borne sensor network data compression method based on artificial intelligence.
Background
In recent years, the development level of Wireless Sensor Network (WSN) technology has been advanced under the push of global information formation tide. The WSN has the characteristics of simple deployment, automatic data acquisition, real-time processing, ad hoc network multi-hop wireless communication, good adaptability to severe environments and the like, and is considered to become an important technology for future aviation detection. The autonomous navigation control, energy and propulsion, measurement and control communication, descending and landing entering, novel data acquisition equipment and the like are key technologies which need to be broken through and mastered urgently, and the data acquisition technology for scattering wireless sensing network nodes on the surface of a planet through a spacecraft to carry out environmental information is also one of research hotspots of the aerospace technology at the present stage in China. At present, the application of WSN in the military aerospace field is actively explored by many military and scientific strong countries in the world, and many universities and government organizations in europe and the united states, such as the u.s.department of defense advanced research project, the european space technology center, the u.s.national aerospace agency, the united states consortium of scientists, the u.s.department of defense, etc., have invested a lot of manpower and material resources for research. Under the vigorous development situation of aerospace industry in recent years, the application research requirements of WSN in the fields of aerospace, military and the like are more and more urgent. The research on WSN with better reliability, effectiveness and real-time performance on a spacecraft becomes one of the very important key technologies in the development of military affairs and aerospace career in China. In most cases, the sensor node on-board radio transceiver is a major cause of energy consumption. The energy problem has been a bottleneck limiting the widespread use of wireless sensor networks. Therefore, it is crucial for the WSN how to design a data compression scheme that saves power consumption and eliminates redundancy.
Data compression can effectively save energy consumption of communication and storage by reducing the data volume of the WSN. For commonly used data compression algorithms that focus on time series based de-approximation data, converting data samples into a set of coefficients to simplify the data representation, the performance of the compression algorithm depends on the number of coefficients needed to encode the input data, increasing computational power consumption as the coefficients increase. The Compressed Sensing (CS) method proposed by Donoho provides a new direction for data compression in wireless sensor networks. When the raw data is sparse on a basis, the CS method can recover a large amount of raw data using fewer measurements. The CS can greatly reduce the system cost due to the use of sparse binary matrices. CS requires that the signal be sparse or compressible at some level, otherwise the signal cannot be reconstructed. There are many signal recovery algorithms for fast reconstruction and reliable accuracy, such as base-tracking (BP), Orthogonal Matching Pursuit (OMP) and segmented OMP (stomp). BP has high computational complexity and cannot be used for large-scale applications. OMP and StOMP use a bottom-up approach in signal recovery, which is far less complex than BP. However, they require more measurements and lack recovery guarantees.
Deep Learning (DL) is an artificial intelligence technique with excellent mathematical fitting and deep feature learning capabilities. In recent years, Convolutional Neural Networks (CNNs) have shown remarkable capabilities in various fields, facilitating the widespread use of DL in various fields. The CNN can extract deeper and richer data hidden information through multi-layer iterative convolution. Large deep convolutional networks have also been applied to data compression, but most research is currently limited to the field of image compression. For wireless sensor networks, most people use RBMs or fully connected layers to learn compression and reconstruction networks separately, and few people merge compression and reconstruction networks together to achieve counterlearning. One important reason for the lack of research is that compression and reconstruction networks are not well correlated in training and are prone to situations where training learning is unstable. In conclusion, it is necessary to find an efficient deep learning model applied to an algorithm for data compression of a satellite-borne wireless sensor network.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a satellite-borne sensor network data compression method based on artificial intelligence to solve the technical defects, aiming at the technical problem that a WSN data compression method which saves energy consumption and eliminates redundancy is needed at present.
A satellite-borne sensor network data compression method based on artificial intelligence comprises the following steps:
the method comprises the following steps: in a preset period of time, collecting sensor data at a plurality of terminal sensing nodes of a satellite-borne sensing network, wherein the sensor data comprises a plurality of categories of sensor data;
step two: preprocessing sensor data to eliminate abnormal data, completing missing data by a two-dimensional linear interpolation method, and mapping acquired data to a [ -1, 1] interval in a normalized mode;
step three: the method comprises the steps of constructing an A-CCR network, wherein the A-CCR network comprises a coding part and a decoding part, the network coding part is constructed by combining a convolutional neural network and a Leaky _ ReLU activation layer, a judgment value is given while a compression result is output to judge whether input data is real data or reconstructed data, a Wassertein distance is used for optimizing an objective function so as to improve the supervised learning capacity of an encoder, and the network decoding part is constructed by combining the convolutional neural network and the ReLU activation layer, reconstructs compressed data and adds an MSE constraint term into the objective function;
step four: extracting a data time sequence of a category from the preprocessed sensor data, dividing the data time sequence of the category into m sections and randomly scrambling the sequence, performing circulating iterative training on the A-CCR network by using the data time sequence of the category of the m sections according to the scrambled sequence, obtaining an A-CCR network initial model after reaching a preset iteration time, scrambling the data time sequence of the category of the m sections again before training to the (m + 1) th time if the iteration time is more than m, and continuing training as training data after receiving the m sections of data time sequence of the previous scrambled sequence; if the iteration number is less than m, only the data time sequence of the segment number equal to the iteration number is used for training;
step five: extracting data time sequences of multiple categories from the preprocessed sensor data, dividing the data time sequences of the multiple categories into n sections and randomly scrambling the sequences, performing cyclic iterative training on the A-CCR network initial model by using the n sections of the data time sequences of the multiple categories according to the scrambled sequence, and obtaining an A-CCR network optimization model after preset iteration times are reached, wherein the iteration times are more than or equal to n; before training for the (n + 1) th time, disordering the sequence of the n data time sequences of multiple categories again, and continuing training as training data after receiving the n data time sequences of the previous disordering sequence;
step six: and inputting the satellite-borne sensor network data to be processed into an A-CCR network optimization model for compression processing.
Further, in the step one, the sensor node collects data every 31 seconds within a preset time.
Further, in the second step, the missing data is complemented by using the data points at the adjacent time of the same day and the data at the same time of the adjacent day by a two-dimensional linear interpolation method.
Further, in the step four, the cyclic iterative training of the A-CCR network is carried out, a coding part is trained for 1 time in each network overall iteration, the learning efficiency of two parts of the network is balanced in a decoding part trained for 3 times, and batch standardization is respectively used in the two parts of network training to ensure the stability of the network overall learning.
Further, the data time series of one category in step four is a temperature time series or a humidity time series or an illumination time series.
Further, the time series of data of the plurality of categories in the fifth step includes a temperature time series, a humidity time series and an illumination time series.
Compared with the prior art, the invention has the advantages that:
a satellite-borne wireless sensor network data compression method with high compression rate and low reconstruction error is characterized in that an encoding and decoding countermeasure training network is constructed, a discriminator is added to an encoding part and used for distinguishing original data from reconstructed data, a Wassertein distance is used for optimizing an encoding objective function, an MSE constraint term is added to an objective function of the decoding part, countermeasure training of the whole network on preprocessed sensor data is achieved, and data reconstruction accuracy can be effectively improved. The coding and decoding network obtained by training on the temperature data set can be effectively transferred to the compression reconstruction of the humidity and illumination data, and has good reconstruction accuracy. The invention can effectively reduce the communication quantity of the satellite-borne wireless sensor network, saves the communication energy consumption, saves the storage energy consumption and reduces the communication blockage, and the coding and decoding confrontation training network designed by the invention can quickly and stably learn the characteristic distribution of the sensor data, has high compression rate, low reconstruction error and good sensor data mobility and robustness, and can effectively prolong the service life of the satellite-borne wireless sensor network.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a flow chart of a data compression method of an artificial intelligence-based satellite-borne sensor network according to the invention;
FIG. 2 is a temperature data preprocessing process in an embodiment of the present invention;
FIG. 3 is an A-CCR network structure in an embodiment of the present invention;
FIG. 4 is a reconstruction of an A-CCR network versus temperature data test sample in an embodiment of the present invention;
FIG. 5 shows migration reconstruction results of A-CCR network on humidity data test samples in an embodiment of the present invention;
FIG. 6 is a block diagram of the test results reconstructed using the data after the combined training according to the embodiment of the present invention.
Detailed Description
For a more clear understanding of the technical features, objects and effects of the present invention, embodiments of the present invention will now be described in detail with reference to the accompanying drawings.
A method for compressing data of a satellite-borne sensor network based on artificial intelligence, as shown in fig. 1, includes:
the method comprises the following steps: in a preset period of time, collecting sensor data at a plurality of terminal sensing nodes of a satellite-borne sensing network, wherein the sensor data comprises a plurality of categories of sensor data;
step two: preprocessing sensor data to eliminate abnormal data, completing missing data by a two-dimensional linear interpolation method, and mapping acquired data to a [ -1, 1] interval in a normalized mode;
step three: constructing an A-CCR network, wherein the A-CCR network comprises an encoding part and a decoding part;
step four: extracting a data time sequence of a category from the preprocessed sensor data, dividing the data time sequence of the category into m sections and randomly scrambling the sequence, performing circulating iterative training on the A-CCR network by using the data time sequence of the category of the m sections according to the scrambled sequence, obtaining an A-CCR network initial model after reaching a preset iteration time, scrambling the data time sequence of the category of the m sections again before training to the (m + 1) th time if the iteration time is more than m, and continuing training as training data after receiving the m sections of data time sequence of the previous scrambled sequence; if the iteration number is less than m, only the data time sequence of the segment number equal to the iteration number is used for training;
step five: extracting data time sequences of multiple categories from the preprocessed sensor data, dividing the data time sequences of the multiple categories into n sections and randomly scrambling the sequences, performing cyclic iterative training on the A-CCR network initial model by using the n sections of the data time sequences of the multiple categories according to the scrambled sequence, and obtaining an A-CCR network optimization model after preset iteration times are reached, wherein the iteration times are more than or equal to n; before training for the (n + 1) th time, disordering the sequence of the n data time sequences of multiple categories again, and continuing training as training data after receiving the n data time sequences of the previous disordering sequence;
step six: and inputting the satellite-borne sensor network data to be processed into an A-CCR network optimization model for compression processing.
According to the above steps, the acquired sensor data is first preprocessed. The sensor data is obtained by placing 54 sensor nodes in a laboratory from the wireless sensor network research team of the university of california from 2 month and 28 days to 4 month and 5 days in 2004 to acquire environmental data such as temperature, humidity, illumination and the like once every 31 seconds, and acquiring data of one month in total. Due to node failure, data loss or abnormal conditions can occur to some sensors at some time. Since the WSN is known to be in a laboratory, abnormal data and missing data exceeding the range of actual environmental data are complemented by a two-dimensional linear interpolation method using data points at the same time of the same day and data at the same time of the adjacent day. In order to reduce the influence of the compression reconstruction network model caused by the overlarge input data range, the original data is normalized to the range of [ -1, 1] by using a mean value normalization method, and the convergence rate of the network model learning is accelerated. The data collected by the sensor network is preprocessed through the above steps, and the data of the temperature training set is shown in fig. 2.
A coding and decoding antithetical training network A-CCR is constructed, a coding part D is used for compressing input data to obtain compressed data z and judging whether the input data is original data or decoded data, and a decoding part G is used for reconstructing the compressed data z. And respectively calculating two objective functions of loss _ D and loss _ G according to the judgment result output by the coding part, updating the learning rate during model training by using an ADAM (adaptive analysis and analysis) algorithm, and finely adjusting the model weight parameters by using a BP (back propagation) algorithm to learn the hidden mathematical characteristics of the original sensor data. The overall structure of the network is shown in fig. 3:
the specific parameters of the A-CCR network coding and decoding part network are shown in Table 1:
TABLE 1A-CCR network codec network parameters
The goal of the encoder is to have its output close to the distribution of the real tags, so the goal of D is to reduce the cross entropy, i.e.:
thus, an objective function for D is obtained, normalized to:
the aim of the decoder is to make the encoder think that the data reconstructed by the decoder is real data as much as possible, even if the cross entropy is as large as possible, in order to make the data reconstructed by the decoder as close to the original data as possible, an MSE constraint term is added to the loss function of the decoder, and the MSE constraint term is:
when an A-CCR network is trained, the preprocessed sensor network data is used as a training set of the network, a BP algorithm is used for training the network, and the algorithm flow is as follows:
in the algorithm, x is an input training set, z is a compression result output by an encoding network, Loss _ D is an objective function of a network encoding part, and Loss _ G is an objective function of a network decoding part. After the A-CCR network is subjected to the anti-training by the algorithm, the data of the sensor network has good compression reconstruction performance and good migration capability.
Example 1: reconstruction result of A-CCR network on test sample after training on temperature data
During A-CCR network training, the number of training iterations is 2500, 600000 temperature data in sensor network data are taken as a training set in training data, and after the network is trained, 12000 data are taken from the rest temperature data to test the network performance. Example results as shown in fig. 4, the 'x' point is a reconstruction result of the network, and the 'o' point is real data.
The embodiment result shows that the A-CCR network has higher reconstruction precision, and the reconstructed data is very close to the trend and the numerical value of the original data. Sampling points are input into the network in a small batch mode for training, and feature extraction is carried out through the coding network, so that data compression can be effectively realized. For all sample points in the test set, the maximum reconstruction error of the a-CCR network was 2.2502 ℃, the minimum reconstruction error was less than 0.0005 ℃, and the average reconstruction error was 0.631 ℃. For these 12000 test samples, the reconstruction error value for the 70% sample was less than 0.2 ℃. The embodiment proves that the network can still have high compression reconstruction precision under the condition of 24 times compression rate on temperature data.
Example 2: migration reconstruction embodiment of humidity data by using algorithm
In order to verify the generalization performance of the a-CCR network and the correlation between different sensor data, the temperature data is trained to obtain the network parameters, and then the network parameters are used to test the humidity data in the wireless sensor network, as shown in fig. 5, the example result indicates that the 'x' point is the reconstruction result of the network, and the 'o' point is the real data.
In the embodiment results, the average reconstruction error of the a-CCR network in the migration compression reconstruction test of the humidity data is 1.351, which shows that the hidden mathematical features learned by the network have commonality for different types of data of the same node, and have certain mobility between different data, so that the humidity data can be effectively subjected to compression reconstruction, and meanwhile, the reconstruction result has very high reconstruction accuracy in a small data part and relatively large error in a large data part, and also shows the difference between the temperature data and the humidity data.
Example 3: reconstruction embodiments using multiple data combination training
And in consideration of the correlation among different sensor data, combining temperature, humidity and illumination data to be used as a training set, retraining the network trained by using the temperature data, learning the correlation among different sensor data, and improving the compression reconstruction precision of the A-CCR network. Example results as shown in fig. 6, the 'x' point is a reconstruction result of the network, and the 'o' point is real data.
In the embodiment results, the average reconstruction error in the compression reconstruction test of the temperature data after the retraining of the A-CCR network is 0.367 ℃, the average reconstruction error in the compression reconstruction test of the humidity data is 0.556, the average reconstruction error in the compression reconstruction test of the illumination data is 43.823Lux, the compression reconstruction precision of the temperature data is obviously improved, and the correlation among different types of sensor data can effectively improve the compression reconstruction precision of the network.
In summary, the invention provides a deep fault-tolerant compression method for a wireless sensor network with high compression rate and low reconstruction error, which includes the steps of performing two-dimensional linear interpolation on missing data and abnormal data by using data points at the same time of the same day and data at the same time of the adjacent day, constructing a coding and decoding countermeasure network, optimizing an objective function of a coding part by using Wassertein distance, adding an MSE constraint term into the objective function of the decoding part, and improving the supervised learning capability of a coder. And then, carrying out iterative training on the preprocessed sensor network data by using a countermeasure coding and decoding network to obtain network model parameters. The invention effectively reduces the communication quantity of the satellite-borne wireless sensor network, saves the communication energy consumption, saves the storage energy consumption and reduces the communication blockage.
While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (6)
1. A satellite-borne sensor network data compression method based on artificial intelligence is characterized by comprising the following steps:
the method comprises the following steps: in a preset period of time, collecting sensor data at a plurality of terminal sensing nodes of a satellite-borne sensing network, wherein the sensor data comprises a plurality of categories of sensor data;
step two: preprocessing sensor data to eliminate abnormal data, completing missing data by a two-dimensional linear interpolation method, and mapping acquired data to a [ -1, 1] interval in a normalized mode;
step three: the method comprises the steps of constructing an A-CCR network, wherein the A-CCR network comprises a coding part and a decoding part, the network coding part is constructed by combining a convolutional neural network and a Leaky _ ReLU activation layer, a judgment value is given while a compression result is output to judge whether input data is real data or reconstructed data, a Wassertein distance is used for optimizing an objective function so as to improve the supervised learning capacity of an encoder, and the network decoding part is constructed by combining the convolutional neural network and the ReLU activation layer, reconstructs compressed data and adds an MSE constraint term into the objective function;
step four: extracting a data time sequence of a category from the preprocessed sensor data, dividing the data time sequence of the category into m sections and randomly scrambling the sequence, performing circulating iterative training on the A-CCR network by using the data time sequence of the category of the m sections according to the scrambled sequence, obtaining an A-CCR network initial model after reaching a preset iteration time, scrambling the data time sequence of the category of the m sections again before training to the (m + 1) th time if the iteration time is more than m, and continuing training as training data after receiving the m sections of data time sequence of the previous scrambled sequence; if the iteration number is less than m, only the data time sequence of the segment number equal to the iteration number is used for training;
step five: extracting data time sequences of multiple categories from the preprocessed sensor data, dividing the data time sequences of the multiple categories into n sections and randomly scrambling the sequences, performing cyclic iterative training on the A-CCR network initial model by using the n sections of the data time sequences of the multiple categories according to the scrambled sequence, and obtaining an A-CCR network optimization model after preset iteration times are reached, wherein the iteration times are more than or equal to n; before training for the (n + 1) th time, disordering the sequence of the n data time sequences of multiple categories again, and continuing training as training data after receiving the n data time sequences of the previous disordering sequence;
step six: and inputting the satellite-borne sensor network data to be processed into an A-CCR network optimization model for compression processing.
2. The artificial intelligence based data compression method for the spaceborne sensor network is characterized in that in the step one, the sensor nodes collect data every 31 seconds within a preset time.
3. The artificial intelligence based data compression method for the spaceborne sensor network according to the claim 1, wherein in the second step, missing data is completed by using data points at the same time of the same day and data at the same time of the same day in a two-dimensional linear interpolation method.
4. The artificial intelligence based data compression method for the spaceborne sensor network is characterized in that in the step four, the A-CCR network carries out cyclic iteration training, a coding part is trained for 1 time in each network overall iteration, in a decoding part is trained for 3 times, the learning efficiency of the two parts of the network is balanced, and batch standardization is respectively used in the two parts of network training to ensure the stability of the network overall learning.
5. The artificial intelligence based data compression method for the space-borne sensor network according to claim 1, wherein the data time series in one of the four steps is a temperature time series, a humidity time series or an illumination time series.
6. The artificial intelligence based data compression method for the space-borne sensor network according to claim 1, wherein the multiple categories of data time series in the fifth step include a temperature time series, a humidity time series and an illumination time series.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910704493.XA CN110519233B (en) | 2019-07-31 | 2019-07-31 | Satellite-borne sensor network data compression method based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910704493.XA CN110519233B (en) | 2019-07-31 | 2019-07-31 | Satellite-borne sensor network data compression method based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110519233A CN110519233A (en) | 2019-11-29 |
| CN110519233B true CN110519233B (en) | 2021-07-20 |
Family
ID=68624283
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910704493.XA Active CN110519233B (en) | 2019-07-31 | 2019-07-31 | Satellite-borne sensor network data compression method based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110519233B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111641679B (en) * | 2020-04-30 | 2023-06-20 | 未来穿戴技术有限公司 | Data transmission method of wearable massager and electronic equipment |
| CN112153659B (en) * | 2020-08-21 | 2022-05-03 | 中国地质大学（武汉） | Efficient and energy-saving construction method of data compression model of satellite-borne wireless sensor network |
| CN113115247B (en) * | 2021-03-17 | 2022-07-12 | 苏州康迈德医疗科技有限公司 | Signal processing method, device and equipment |
| CN113804252A (en) * | 2021-09-10 | 2021-12-17 | 广州市吉华勘测股份有限公司 | High formwork supporting safety monitoring method, device, equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102202322A (en) * | 2011-04-08 | 2011-09-28 | 上海交通大学 | Method for anti-interference of wireless sensor network based on game theory |
| CN107403160A (en) * | 2017-07-28 | 2017-11-28 | 中国地质大学（武汉） | Image detecting method, equipment and its storage device in a kind of intelligent driving scene |
| CN109191402A (en) * | 2018-09-03 | 2019-01-11 | 武汉大学 | The image repair method and system of neural network are generated based on confrontation |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130031042A1 (en) * | 2011-07-27 | 2013-01-31 | Sintayehu Dehnie | Distributed assured network system (DANS) |
-
2019
- 2019-07-31 CN CN201910704493.XA patent/CN110519233B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102202322A (en) * | 2011-04-08 | 2011-09-28 | 上海交通大学 | Method for anti-interference of wireless sensor network based on game theory |
| CN107403160A (en) * | 2017-07-28 | 2017-11-28 | 中国地质大学（武汉） | Image detecting method, equipment and its storage device in a kind of intelligent driving scene |
| CN109191402A (en) * | 2018-09-03 | 2019-01-11 | 武汉大学 | The image repair method and system of neural network are generated based on confrontation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110519233A (en) | 2019-11-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110519233B (en) | Satellite-borne sensor network data compression method based on artificial intelligence | |
| Li et al. | An energy-efficient data collection scheme using denoising autoencoder in wireless sensor networks | |
| CN112924749B (en) | Unsupervised counterstudy electromagnetic spectrum abnormal signal detection method | |
| CN111784041B (en) | Wind power prediction method and system based on graph convolution neural network | |
| CN110874550A (en) | Data processing method, device, equipment and system | |
| CN113113030B (en) | High-dimensional damaged data wireless transmission method based on noise reduction self-encoder | |
| Bin et al. | Compressive data gathering with generative adversarial networks for wireless geophone networks | |
| CN110446173B (en) | Efficient and energy-saving satellite-borne wireless sensor network data compression method | |
| CN109376797B (en) | Network traffic classification method based on binary encoder and multi-hash table | |
| Zhang et al. | Goal-oriented communications for the IoT and application to data compression | |
| CN112819082B (en) | Satellite spectrum sensing data reconstruction method based on deep learning | |
| Leinonen et al. | Compressed sensing with applications in wireless networks | |
| CN110995396B (en) | Compression method of communication messages of electricity consumption information acquisition system based on hierarchical structure | |
| Alsheikh et al. | Toward a robust sparse data representation for wireless sensor networks | |
| CN116961805A (en) | Short wave prediction method based on space-time diagram neural network | |
| CN112712855B (en) | Joint training-based clustering method for gene microarray containing deletion value | |
| CN111224938A (en) | Wireless seismograph network compressed data transmission method | |
| CN111935766B (en) | Wireless network flow prediction method based on global spatial dependency | |
| Dong et al. | Innovative semantic communication system | |
| CN111967577A (en) | Energy internet scene generation method based on variational self-encoder | |
| CN116760674B (en) | Intelligent monitoring and identifying method and system for wireless signal modulation mode | |
| Huang et al. | Application of Steady State Data Compressed Sensing Based on LSTM and RNN in Rural Power Grid | |
| CN116361256B (en) | Data synchronization method and system based on log analysis | |
| CN116029340B (en) | Image and semantic information transmission method based on deep learning network | |
| CN117220826B (en) | Agricultural Internet of things perception data prediction method based on semantic communication |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |