CN114103710A - Self-adaptive charging system for electric automobile and working method thereof - Google Patents

Self-adaptive charging system for electric automobile and working method thereof Download PDF

Info

Publication number
CN114103710A
CN114103710A CN202111444506.8A CN202111444506A CN114103710A CN 114103710 A CN114103710 A CN 114103710A CN 202111444506 A CN202111444506 A CN 202111444506A CN 114103710 A CN114103710 A CN 114103710A
Authority
CN
China
Prior art keywords
charging
neural network
feature
adaptive
convolutional neural
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
CN202111444506.8A
Other languages
Chinese (zh)
Inventor
于隆山
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xiamen Renxin Industrial Co ltd
Original Assignee
Xiamen Renxin Industrial Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xiamen Renxin Industrial Co ltd filed Critical Xiamen Renxin Industrial Co ltd
Priority to CN202111444506.8A priority Critical patent/CN114103710A/en
Publication of CN114103710A publication Critical patent/CN114103710A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60L—PROPULSION OF ELECTRICALLY-PROPELLED VEHICLES; SUPPLYING ELECTRIC POWER FOR AUXILIARY EQUIPMENT OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRODYNAMIC BRAKE SYSTEMS FOR VEHICLES IN GENERAL; MAGNETIC SUSPENSION OR LEVITATION FOR VEHICLES; MONITORING OPERATING VARIABLES OF ELECTRICALLY-PROPELLED VEHICLES; ELECTRIC SAFETY DEVICES FOR ELECTRICALLY-PROPELLED VEHICLES
    • B60L53/00—Methods of charging batteries, specially adapted for electric vehicles; Charging stations or on-board charging equipment therefor; Exchange of energy storage elements in electric vehicles
    • B60L53/60—Monitoring or controlling charging stations
    • B60L53/62—Monitoring or controlling charging stations in response to charging parameters, e.g. current, voltage or electrical charge
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/60—Other road transportation technologies with climate change mitigation effect
    • Y02T10/70—Energy storage systems for electromobility, e.g. batteries
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/60—Other road transportation technologies with climate change mitigation effect
    • Y02T10/7072—Electromobility specific charging systems or methods for batteries, ultracapacitors, supercapacitors or double-layer capacitors
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
    • Y02T90/10—Technologies relating to charging of electric vehicles
    • Y02T90/12—Electric charging stations

Landscapes

  • Engineering & Computer Science (AREA)
  • Power Engineering (AREA)
  • Transportation (AREA)
  • Mechanical Engineering (AREA)
  • Charge And Discharge Circuits For Batteries Or The Like (AREA)
  • Electric Propulsion And Braking For Vehicles (AREA)

Abstract

The application relates to the field of intelligent charging of electric automobiles, and particularly discloses a self-adaptive charging system for an electric automobile and a working method of the self-adaptive charging system. It extracts high-dimensional features in the charging waveform image by parameter setting of convolution kernels for each layer in the convolutional neural network, while further facilitating training of the neural network with "controller" techniques of the neural network that search space and search state, and implements regression by parameter training of classifiers on feature values or combinations of feature values in the feature vectors to achieve classification of images by explicit expression of nonlinear features in the image containing harmonics, sudden changes, unsteady signals, and the like. By the mode, end-to-end adaptive adjustment of multiple electric vehicles charged simultaneously can be realized at one time, so that the system performance of the adaptive charging system for the electric vehicles is improved.

Description

Self-adaptive charging system for electric automobile and working method thereof
Technical Field
The present invention relates to the field of intelligent charging of electric vehicles, and more particularly, to an adaptive charging system for electric vehicles and a method for operating the same.
Background
Because the storage battery of the electric automobile, the distribution transformer, the charger and other devices have a large number of nonlinear elements, voltage and current signals in the charging process of the electric automobile both contain multiple harmonic wave (ripple wave) components which are integral multiples of the fundamental frequency and a large number of impact unstable-state waves.
Particularly, in the fast charging mode of the electric vehicle, which adopts high-power direct current charging, the distribution of charging loads in time and space is very random, and excessively intensive large-scale centralized charging can cause instantaneous overload, and sudden changes of voltage or current signals can often occur in the charging process, and the charging harmonics (ripples), unsteady waves and sudden changes of signals can cause great damage to electric equipment.
Therefore, an adaptive charging system for an electric vehicle is desired in order to adaptively adjust charging of the electric vehicle according to actual conditions, to improve charging performance of the adaptive charging system for the electric vehicle, and to protect a power system of the electric vehicle from damage.
Disclosure of Invention
The present application is proposed to solve the above-mentioned technical problems. Embodiments of the present application provide an adaptive charging system for electric vehicles and an operating method thereof, which extracts high-dimensional features in a charging waveform image through parameter setting of convolution kernels for each layer in a convolutional neural network, while further facilitating training of the neural network using a search space and a "controller" technique of the neural network of a search state, and implements regression of feature values or combinations of feature values in the feature vectors through parameter training of a classifier to implement classification of images through explicit expression of nonlinear features including harmonics (ripples), sudden changes, unsteady signals, and the like in the images. By the mode, end-to-end adaptive adjustment of multiple electric vehicles charged simultaneously can be realized at one time, so that the system performance of the adaptive charging system for the electric vehicles is improved.
According to an aspect of the present application, there is provided an adaptive charging system for an electric vehicle, including:
the charging data acquisition unit is used for acquiring charging waveform images of charging currents of a plurality of electric vehicles which are charged simultaneously;
a high-dimensional feature extraction unit, configured to input a charging waveform image of each electric vehicle that is simultaneously charged into a first convolution neural network respectively to extract high-dimensional features in the charging waveform image through parameter settings of convolution kernels of layers in the first convolution neural network to obtain a charging feature vector corresponding to each charging waveform image;
a two-dimensional cascade unit configured to cascade the charging feature vectors corresponding to each of the charging waveform images in two dimensions to obtain a charging feature matrix;
a search space construction unit, configured to perform explicit spatial coding on the charging feature matrix using a second convolutional neural network to extract high-dimensional correlation information between the charging feature vectors, so as to obtain a correlation feature map;
the searching unit is used for taking a charging characteristic vector corresponding to a certain electric automobile as a searching state, taking the associated characteristic diagram as a searching space, and multiplying the charging characteristic vector and the associated characteristic diagram by a matrix to obtain a control characteristic vector;
the vector fusion unit is used for multiplying the charging characteristic vector corresponding to a certain electric automobile and the control characteristic vector according to position points to obtain a classification characteristic vector; and
and the adaptive charging result generating unit is used for enabling the classified characteristic vectors to pass through a classifier so as to obtain a classification result representing an adaptive adjustment result of the charging of the electric automobile.
In the above adaptive charging system for an electric vehicle, the first convolution neural network extracts a high-dimensional feature in the charging waveform image in the following formula to obtain the charging feature map;
wherein the formula is:
fi＝active(Ni×fi-1+Bi)
wherein f isi-1For the input of the first convolutional neural network of the i-th layer, fiFor the output of the first convolutional neural network of the ith layer, NiIs the convolution kernel of the ith layer first convolution neural network, and BiActive represents a nonlinear activation function for the bias vector of the first convolutional neural network of the ith layer.
In the above adaptive charging system for an electric vehicle, the last layer of the first convolutional neural network is used to perform global mean pooling on the charging feature map to obtain the charging feature vector.
In the adaptive charging system for an electric vehicle, the two-dimensional cascade unit is further configured to: arranging charging feature vectors corresponding to each of the charging waveform images in rows of a matrix to obtain the charging feature matrix.
In the above adaptive charging system for an electric vehicle, the second convolutional neural network explicitly spatially encodes the charging feature matrix to obtain the correlation feature map; wherein the formula is:
fi＝tanh(Ni×fi-1+Bi)
wherein f isi-1As input to the ith layer of the second convolutional neural network, fiIs the output of the ith layer of the second convolutional neural network, NiIs the convolution kernel of the ith layer of the second convolutional neural network, and BiFor the bias vector of the ith layer second convolutional neural network, tanh represents the nonlinear activation function.
In the adaptive charging system for an electric vehicle, the adaptive charging result generating unit is further configured to: inputting the classification feature vector into a Softmax classification function of the classifier to obtain probability values that the classification feature vector belongs to tags representing adaptive adjustment results of charging respectively; and determining the label of the charging self-adaptive adjustment result corresponding to the maximum probability value as the classification result.
In the above adaptive charging system for an electric vehicle, the label for indicating the result of the adaptive adjustment of the charging includes: increasing the charging current and decreasing the charging current.
According to another aspect of the present application, an operating method of an adaptive charging system for an electric vehicle includes:
acquiring charging waveform images of charging currents of a plurality of electric vehicles which are charged simultaneously;
inputting a charging waveform image of each electric automobile which is charged simultaneously into a first convolution neural network respectively to extract high-dimensional features in the charging waveform image through parameter setting of convolution kernels of layers in the first convolution neural network so as to obtain a charging feature vector corresponding to each charging waveform image;
two-dimensionally concatenating charging feature vectors corresponding to each of the charging waveform images to obtain a charging feature matrix;
performing explicit spatial coding on the charging feature matrix by using a second convolutional neural network to extract high-dimensional correlation information between the charging feature vectors so as to obtain a correlation feature map;
taking a charging characteristic vector corresponding to a certain electric automobile as a searching state and the associated characteristic diagram as a searching space, and carrying out matrix multiplication on the charging characteristic vector and the associated characteristic diagram to obtain a control characteristic vector;
multiplying the charging characteristic vector corresponding to a certain electric automobile and the control characteristic vector according to position points to obtain a classification characteristic vector; and
and passing the classified feature vector through a classifier to obtain a classification result representing an adaptive adjustment result of the charging of the electric automobile.
In the above operating method of the adaptive charging system for an electric vehicle, the first convolutional neural network extracts a high-dimensional feature in the charging waveform image with the following formula to obtain the charging feature map; wherein the formula is:
fi＝active(Ni×fi-1+Bi)
wherein f isi-1For the input of the first convolutional neural network of the i-th layer, fiFor the output of the first convolutional neural network of the ith layer, NiIs the convolution kernel of the ith layer first convolution neural network, and BiActive represents a nonlinear activation function for the bias vector of the first convolutional neural network of the ith layer.
In the above operating method of the adaptive charging system for an electric vehicle, the last layer of the first convolutional neural network is used to perform global mean pooling on the charging feature map to obtain the charging feature vector.
In the above method for operating an adaptive charging system for an electric vehicle, two-dimensionally concatenating charging feature vectors corresponding to each of the charging waveform images to obtain a charging feature matrix, the method includes: arranging charging feature vectors corresponding to each of the charging waveform images in rows of a matrix to obtain the charging feature matrix.
In the above operating method of the adaptive charging system for an electric vehicle, the second convolutional neural network performs explicit spatial coding on the charging feature matrix to obtain the correlation feature map;
wherein the formula is:
fi＝tanh(Ni×fi-1+Bi)
wherein f isi-1As input to the ith layer of the second convolutional neural network, fiIs the output of the ith layer of the second convolutional neural network, NiIs the convolution kernel of the ith layer of the second convolutional neural network, and BiFor the bias vector of the ith layer second convolutional neural network, tanh represents the nonlinear activation function.
In the above method for operating an adaptive charging system for an electric vehicle, passing the classification feature vector through a classifier to obtain a classification result indicating an adaptive adjustment result of charging of the electric vehicle, the method includes: inputting the classification feature vector into a Softmax classification function of the classifier to obtain probability values that the classification feature vector belongs to tags representing adaptive adjustment results of charging respectively; and determining the label of the charging self-adaptive adjustment result corresponding to the maximum probability value as the classification result.
In the above method for operating an adaptive charging system for an electric vehicle, the tag indicating the result of the adaptive adjustment of the charging includes: increasing the charging current and decreasing the charging current.
Compared with the prior art, the adaptive charging system for the electric vehicle and the working method thereof extract high-dimensional features in a charging waveform image through parameter setting of convolution kernels of each layer in a convolution neural network, meanwhile, training of the neural network is further facilitated by utilizing a search space and a controller technology of the neural network in a search state, and characteristic values or combination of the characteristic values in characteristic vectors are subjected to regression through parameter training of a classifier so as to realize classification of the image through explicit expression of nonlinear features containing harmonic waves (ripples), sudden changes, unsteady signals and the like in the image. By the mode, end-to-end adaptive adjustment of multiple electric vehicles charged simultaneously can be realized at one time, so that the system performance of the adaptive charging system for the electric vehicles is improved.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally represent like parts or steps.
Fig. 1 is an application scenario diagram of an adaptive charging system for an electric vehicle according to an embodiment of the present application.
Fig. 2 is a block diagram of an adaptive charging system for an electric vehicle according to an embodiment of the present application.
Fig. 3 is a flowchart of an operating method of an adaptive charging system for an electric vehicle according to an embodiment of the present application.
Fig. 4 is a schematic architecture diagram of an operating method of an adaptive charging system for an electric vehicle according to an embodiment of the present application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.
Overview of a scene
As mentioned above, since the storage battery of the electric vehicle and the devices such as the distribution transformer and the charger have a large number of nonlinear elements, the voltage and current signals during the charging process of the electric vehicle both include multiple harmonic (ripple) components that are integral multiples of the fundamental frequency and a large number of impulsive unsteady-state waves.
Particularly, in the fast charging mode of the electric vehicle, which adopts high-power direct current charging, the distribution of charging loads in time and space is very random, and excessively intensive large-scale centralized charging can cause instantaneous overload, and sudden changes of voltage or current signals can often occur in the charging process, and the charging harmonics (ripples), unsteady waves and sudden changes of signals can cause great damage to electric equipment.
Therefore, an adaptive charging system for an electric vehicle is desired in order to adaptively adjust charging of the electric vehicle according to actual conditions, to improve charging performance of the adaptive charging system for the electric vehicle, and to protect a power system of the electric vehicle from damage.
Specifically, in the technical solution of the present application, in order to extract nonlinear features including harmonics (ripples), sudden changes, and unsteady signals in a charging waveform, a charging waveform image is input to a convolutional neural network to extract high-dimensional features in the image by parameter setting of convolutional kernels of respective layers in the convolutional neural network. Here, the output of the convolutional neural network is a high-dimensional feature vector, typically 256-or 512-dimensional, and the feature values or combinations of feature values in the feature vector can implicitly represent a high-dimensional representation of the features in the image.
That is, the eigenvalues or combination of eigenvalues in the eigenvectors, when regressed by the classifier, can achieve classification of the image by explicit expression of nonlinear features in the image that contain harmonics (ripples), abrupt and unsteady signals, etc., but these eigenvalues or combination of eigenvalues specifically correspond to what features are unintelligible because they have been mapped into a high-dimensional space. Therefore, regression needs to be achieved through parameter training of the classifier when the classifier is subsequently input.
Training of neural networks can be further facilitated based on recently emerging "controller" techniques of neural networks based on search space and search state. In the technical scheme of the application, because the electric vehicles are not charged individually, but in a specific charging system, a plurality of electric vehicles are charged simultaneously, and therefore, there is also a potential correlation between the charging waveforms of each electric vehicle.
Therefore, the charging waveform images of each of the plurality of electric vehicles that are simultaneously charged are respectively input to the first convolution neural network to obtain the charging feature vector corresponding to each of the charging waveform images. Then, the charging feature vectors are cascaded in two dimensions to obtain a charging feature matrix, and high-dimensional correlation information among the charging feature vectors is further extracted through a second convolutional neural network to obtain a correlation feature map. In this way, the charging feature vector corresponding to a certain electric vehicle is set as a search state, the correlation feature map is set as a search space, and the charging feature vector is multiplied by the correlation feature map to obtain a control feature vector, so that "control" of the first convolutional neural network, which is a sample generation network for generating the charging feature vector, can be realized.
Specifically, the charging feature vector and the control feature vector are point-multiplied to obtain a classification feature vector, and then a classifier is used to obtain a self-adaptive adjustment result of the charging of the electric vehicle, such as increasing or decreasing the charging current. In addition, by the scheme, end-to-end adaptive adjustment of multiple electric vehicles charged simultaneously can be realized at one time, so that the system performance of the adaptive charging system for the electric vehicles is improved.
Based on this, the present application proposes an adaptive charging system for an electric vehicle, comprising: the charging data acquisition unit is used for acquiring charging waveform images of charging currents of a plurality of electric vehicles which are charged simultaneously; a high-dimensional feature extraction unit, configured to input a charging waveform image of each electric vehicle that is simultaneously charged into a first convolution neural network respectively to extract high-dimensional features in the charging waveform image through parameter settings of convolution kernels of layers in the first convolution neural network to obtain a charging feature vector corresponding to each charging waveform image; a two-dimensional cascade unit configured to cascade the charging feature vectors corresponding to each of the charging waveform images in two dimensions to obtain a charging feature matrix; a search space construction unit, configured to perform explicit spatial coding on the charging feature matrix using a second convolutional neural network to extract high-dimensional correlation information between the charging feature vectors, so as to obtain a correlation feature map; the searching unit is used for taking a charging characteristic vector corresponding to a certain electric automobile as a searching state, taking the associated characteristic diagram as a searching space, and multiplying the charging characteristic vector and the associated characteristic diagram by a matrix to obtain a control characteristic vector; the vector fusion unit is used for multiplying the charging characteristic vector corresponding to a certain electric automobile and the control characteristic vector according to position points to obtain a classification characteristic vector; and an adaptive charging result generation unit for passing the classification feature vector through a classifier to obtain a classification result representing an adaptive adjustment result of the charging of the electric vehicle.
Fig. 1 illustrates an application scenario diagram of an adaptive charging system for an electric vehicle according to an embodiment of the present application. As shown in fig. 1, in this application scenario, first, charging waveform images of charging currents of a plurality of electric vehicles (e.g., B1-Bn as illustrated in fig. 1) that are simultaneously charged are acquired in a parking lot in which a plurality of electric vehicle charging devices (e.g., T1-Tn as illustrated in fig. 1) are provided and are charging the plurality of electric vehicles. Then, the obtained charging waveform images of the charging currents of the plurality of electric vehicles are input into a server (for example, a cloud server S as illustrated in fig. 1) deployed with an adaptive charging algorithm for electric vehicles, wherein the server can process the charging waveform images of the charging currents of the plurality of electric vehicles with the adaptive charging algorithm for electric vehicles to generate a classification result representing an adaptive adjustment result of charging of the electric vehicles.
Having described the general principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.
Exemplary System
Fig. 2 illustrates a block diagram of an adaptive charging system for an electric vehicle according to an embodiment of the present application. As shown in fig. 2, an adaptive charging system 200 for an electric vehicle according to an embodiment of the present application includes: a charging data acquisition unit 210 for acquiring charging waveform images of charging currents of a plurality of electric vehicles charged simultaneously; a high-dimensional feature extraction unit 220, configured to input charging waveform images of each electric vehicle that is simultaneously charged into a first convolution neural network respectively to extract high-dimensional features in the charging waveform images through parameter settings of convolution kernels of layers in the first convolution neural network to obtain charging feature vectors corresponding to each charging waveform image; a two-dimensional concatenation unit 230 configured to two-dimensionally concatenate the charging feature vectors corresponding to each of the charging waveform images to obtain a charging feature matrix; a search space construction unit 240, configured to perform explicit spatial coding on the charging feature matrix using a second convolutional neural network to extract high-dimensional correlation information between the charging feature vectors, so as to obtain a correlation feature map; a searching unit 250, configured to take a charging feature vector corresponding to a certain electric vehicle as a search state and take the associated feature map as a search space, and perform matrix multiplication on the charging feature vector and the associated feature map to obtain a control feature vector; a vector fusion unit 260, configured to multiply the charging feature vector and the control feature vector corresponding to a certain electric vehicle by a position point to obtain a classification feature vector; and an adaptive charging result generating unit 270 for passing the classification feature vector through a classifier to obtain a classification result representing an adaptive adjustment result of charging of the electric vehicle.
Specifically, in the embodiment of the present application, the charging data acquiring unit 210 and the high-dimensional feature extracting unit 220 are configured to acquire charging waveform images of charging currents of a plurality of electric vehicles that are simultaneously charged, and input the charging waveform images of each electric vehicle that are simultaneously charged into a first convolutional neural network respectively to extract high-dimensional features in the charging waveform images through parameter setting of convolution kernels of layers in the first convolutional neural network so as to acquire a charging feature vector corresponding to each charging waveform image. As mentioned above, since there are a large number of nonlinear elements in the storage battery of the electric vehicle, the distribution transformer, the charger, and other devices, the voltage and current signals in the charging process of the electric vehicle each include multiple harmonic (ripple) components that are integral multiples of the fundamental frequency and a large number of impulsive unsteady state waves, and it is considered that in practical application scenarios, usually, the electric vehicles are not charged individually, but in a specific charging system, the electric vehicles are charged simultaneously, and therefore, there is also a potential correlation between the charging waveforms of each electric vehicle.
Therefore, in the technical solution of the present application, in order to extract nonlinear characteristics including harmonics (ripples), sudden changes, unsteady signals, and the like in the charging waveform and correlation characteristics between the charging waveforms, first, charging waveform images of charging currents of a plurality of electric vehicles that are simultaneously charged need to be acquired in a parking lot; then, the charging waveform image is input into a first convolution neural network for processing, so as to extract high-dimensional features in the charging waveform image through parameter setting of convolution kernels of each layer in the first convolution neural network, and accordingly a charging feature vector corresponding to each charging waveform image is obtained. It is worth mentioning that here, the output of the first convolutional neural network is a high-dimensional feature vector, typically 256-dimensional or 512-dimensional, and the feature value or combination of feature values in the feature vector can implicitly express a high-dimensional representation of the feature in the image.
More specifically, in the embodiment of the present application, the first convolutional neural network extracts a high-dimensional feature in the charging waveform image in the following formula to obtain the charging feature map;
wherein the formula is:
fi＝active(Ni×fi-1+Bi)
wherein f isi-1Is the input of the ith layer first convolutional neural network,fifor the output of the first convolutional neural network of the ith layer, NiIs the convolution kernel of the ith layer first convolution neural network, and BiActive represents a nonlinear activation function for the bias vector of the first convolutional neural network of the ith layer. It should be noted that, here, the last layer of the first convolutional neural network is used to perform global mean pooling on the charging feature map to reduce the number of parameters and reduce overfitting, thereby obtaining the charging feature vector.
Specifically, in this embodiment of the present application, the two-dimensional concatenation unit 230 and the search space construction unit 240 are configured to two-dimensionally concatenate charging feature vectors corresponding to each of the charging waveform images to obtain a charging feature matrix, and perform explicit spatial coding on the charging feature matrix using a second convolutional neural network to extract high-dimensional correlation information between the charging feature vectors, so as to obtain a correlation feature map. It should be appreciated that training of neural networks based on the recently emerging "controller" technology of neural networks based on search space and search state may be further facilitated. Therefore, in the technical solution of the present application, after the charging feature vector corresponding to each of the charging waveform images is obtained as a search state, a search space needs to be generated. That is, first, the charging feature vectors corresponding to each of the charging waveform images are two-dimensionally concatenated to obtain a charging feature matrix, and in one specific example, the charging feature vectors corresponding to each of the charging waveform images may be arranged in rows of a matrix to obtain the charging feature matrix. Then, the charging feature matrix is further subjected to explicit spatial coding by using a second convolutional neural network so as to further extract high-dimensional correlation information among the charging feature vectors, thereby obtaining a correlation feature map serving as a search space.
More specifically, in the embodiment of the present application, the second convolutional neural network explicitly spatially encodes the charging feature matrix to obtain the associated feature map;
wherein the formula is:
fi＝tanh(Ni×fi-1+Bi)
wherein f isi-1As input to the ith layer of the second convolutional neural network, fiIs the output of the ith layer of the second convolutional neural network, NiIs the convolution kernel of the ith layer of the second convolutional neural network, and BiFor the bias vector of the ith layer second convolutional neural network, tanh represents the nonlinear activation function.
Specifically, in the embodiment of the present application, the searching unit 250 is configured to take a charging feature vector corresponding to a certain electric vehicle as a searching state and the associated feature map as a searching space, and perform matrix multiplication on the charging feature vector and the associated feature map to obtain a control feature vector. It should be understood that, in the technical solution of the present application, a "controller" technology of the neural network based on the search space and the search state is adopted, which can further facilitate training of the neural network to improve the accuracy of subsequent classification. Specifically, in the present invention, a charging feature vector corresponding to a certain electric vehicle is set as a search state, the correlation feature map is set as a search space, and a control feature vector is obtained by multiplying the charging feature vector by the correlation feature map, so that "control" of the first convolution neural network as a sample generation network for generating the charging feature vector can be realized.
Specifically, in the embodiment of the present application, the vector fusion unit 260 is configured to multiply the charging feature vector and the control feature vector corresponding to a certain electric vehicle by a position point to obtain a classification feature vector. It should be understood that the eigenvalues or combinations of eigenvalues in the eigenvectors, when regressed by the classifier, can be classified by explicit representation of the nonlinear features in the image, including harmonics (ripples), dips, unsteady signals, etc., but these eigenvalues or combinations of eigenvalues specifically correspond to what features are not understandable because they have been mapped into a high-dimensional space. Therefore, in the technical solution of the present application, regression needs to be implemented through parameter training of the classifier when the classifier is input. That is, the charging feature vector corresponding to a certain electric vehicle is multiplied by the obtained control feature vector by a position point to obtain a classification feature vector for classification.
Specifically, in the embodiment of the present application, the adaptive charging result generating unit 270 is configured to pass the classification feature vector through a classifier to obtain a classification result representing an adaptive adjustment result of charging of the electric vehicle. In a specific example, the adaptive charging result generating unit is further configured to, first, input the classification feature vector into a Softmax classification function of the classifier to obtain probability values that the classification feature vector belongs to each of the tags representing the adaptive adjustment result of charging, respectively. Specifically, the classification feature vector is input to a Softmax classification function of the classifier to obtain a first probability that the classification feature vector is respectively attributed to increasing charging current and a second probability that the classification feature vector is respectively attributed to decreasing charging current. Then, the label of the charging adaptive adjustment result corresponding to the maximum one of the probability values is determined as the classification result. That is, determining a result of the classification based on a comparison of the probability values, in particular, an increased charging current when the first probability is greater than the second probability; when the first probability is less than the second probability, the classification result is to reduce a charging current. It should be understood that, by the above scheme, end-to-end adaptive adjustment of the multiple electric vehicles charged simultaneously can be realized at one time, thereby improving the system performance of the adaptive charging system for the electric vehicles.
In summary, the adaptive charging system 200 for electric vehicles based on the embodiment of the present application is illustrated, which extracts high-dimensional features in a charging waveform image through parameter setting of convolution kernels for each layer in a convolutional neural network, while further facilitating training of the neural network by using a search space and a "controller" technique of the neural network of a search state, and implements regression by parameter training of a classifier on feature values or combinations of feature values in the feature vectors to implement classification of images through explicit expression of nonlinear features including harmonics (ripples), sudden changes, unsteady signals, and the like in the images. By the mode, end-to-end adaptive adjustment of multiple electric vehicles charged simultaneously can be realized at one time, so that the system performance of the adaptive charging system for the electric vehicles is improved.
As described above, the adaptive charging system 200 for an electric vehicle according to the embodiment of the present application may be implemented in various terminal devices, such as a server for an adaptive charging algorithm for an electric vehicle, and the like. In one example, the adaptive charging system 200 for an electric vehicle according to the embodiment of the present application may be integrated into a terminal device as one software module and/or hardware module. For example, the adaptive charging system 200 for an electric vehicle may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the adaptive charging system 200 for an electric vehicle may also be one of many hardware modules of the terminal device.
Alternatively, in another example, the adaptive charging system for electric vehicles 200 and the terminal device may be separate devices, and the adaptive charging system for electric vehicles 200 may be connected to the terminal device through a wired and/or wireless network and transmit the interactive information according to an agreed data format.
Exemplary method
Fig. 3 illustrates a flow chart of a method of operating an adaptive charging system for an electric vehicle. As shown in fig. 3, an operating method of an adaptive charging system for an electric vehicle according to an embodiment of the present application includes the steps of: s110, acquiring charging waveform images of charging currents of a plurality of electric vehicles which are charged simultaneously; s120, inputting the charging waveform image of each electric automobile charged simultaneously into a first convolution neural network respectively to extract high-dimensional features in the charging waveform image through parameter setting of convolution kernels of layers in the first convolution neural network so as to obtain a charging feature vector corresponding to each charging waveform image; s130, two-dimensionally cascading charging characteristic vectors corresponding to each charging waveform image to obtain a charging characteristic matrix; s140, performing explicit spatial coding on the charging feature matrix by using a second convolutional neural network to extract high-dimensional correlation information among the charging feature vectors so as to obtain a correlation feature map; s150, taking a charging characteristic vector corresponding to a certain electric automobile as a searching state, taking the correlation characteristic diagram as a searching space, and performing matrix multiplication on the charging characteristic vector and the correlation characteristic diagram to obtain a control characteristic vector; s160, multiplying the charging characteristic vector corresponding to a certain electric automobile and the control characteristic vector according to position points to obtain a classification characteristic vector; and S170, passing the classified feature vector through a classifier to obtain a classification result representing an adaptive adjustment result of the charging of the electric vehicle.
Fig. 4 illustrates an architecture diagram of an operating method of an adaptive charging system for an electric vehicle according to an embodiment of the present application. As shown in fig. 4, in the network architecture of the operating method of the adaptive charging system for electric vehicles, first, a charging waveform image (e.g., P as illustrated in fig. 4) of each electric vehicle that is obtained while being charged is respectively input to a first convolution neural network (e.g., CNN1 as illustrated in fig. 4) to extract high-dimensional features in the charging waveform image through parameter settings of convolution kernels of layers in the first convolution neural network to obtain a charging feature vector (e.g., VF1 as illustrated in fig. 4) corresponding to each charging waveform image; then, charging feature vectors corresponding to each of the charging waveform images are two-dimensionally concatenated to obtain a charging feature matrix (e.g., MF as illustrated in fig. 4); then, explicitly spatially encoding the charging feature matrix using a second convolutional neural network (e.g., CNN2 as illustrated in fig. 4) to extract high-dimensional correlation information between the respective charging feature vectors to obtain a correlation feature map (e.g., F as illustrated in fig. 4); then, taking a charging feature vector corresponding to a certain electric vehicle as a search state and the associated feature map as a search space, and performing matrix multiplication on the charging feature vector and the associated feature map to obtain a control feature vector (for example, VF2 as illustrated in fig. 4); then, the charging feature vector corresponding to a certain electric vehicle is multiplied by the control feature vector by a position point to obtain a classification feature vector (for example, VF3 as illustrated in fig. 4); and, finally, passing the classification feature vector through a classifier (e.g., circle S as illustrated in fig. 4) to obtain a classification result representing an adaptive adjustment result of the charging of the electric vehicle.
More specifically, in steps S110 and S120, charging waveform images of charging currents of a plurality of electric vehicles that are simultaneously charged are acquired, and the charging waveform images of each electric vehicle that is simultaneously charged are respectively input to a first convolution neural network to extract high-dimensional features in the charging waveform images through parameter settings of convolution kernels of respective layers in the first convolution neural network to obtain a charging feature vector corresponding to each of the charging waveform images. It should be understood that, in the technical solution of the present application, in order to extract nonlinear characteristics including harmonic waves (ripples), sudden changes, unsteady signals, and the like in the charging waveform and correlation characteristics between the charging waveforms, first, charging waveform images of charging currents of a plurality of electric vehicles that are simultaneously charged need to be acquired in a parking lot; then, the charging waveform image is input into a first convolution neural network for processing, so as to extract high-dimensional features in the charging waveform image through parameter setting of convolution kernels of each layer in the first convolution neural network, and accordingly a charging feature vector corresponding to each charging waveform image is obtained. It is worth mentioning that here, the output of the first convolutional neural network is a high-dimensional feature vector, typically 256-dimensional or 512-dimensional, and the feature value or combination of feature values in the feature vector can implicitly express a high-dimensional representation of the feature in the image.
More specifically, in steps S130 and S140, the charging feature vectors corresponding to each of the charging waveform images are two-dimensionally concatenated to obtain a charging feature matrix, and the charging feature matrix is explicitly space-coded using a second convolutional neural network to extract high-dimensional correlation information between the respective charging feature vectors to obtain a correlation feature map. It should be appreciated that training of neural networks based on the recently emerging "controller" technology of neural networks based on search space and search state may be further facilitated. Therefore, in the technical solution of the present application, after the charging feature vector corresponding to each of the charging waveform images is obtained as a search state, a search space needs to be generated. That is, first, the charging feature vectors corresponding to each of the charging waveform images are two-dimensionally concatenated to obtain a charging feature matrix, and in one specific example, the charging feature vectors corresponding to each of the charging waveform images may be arranged in rows of a matrix to obtain the charging feature matrix. Then, the charging feature matrix is further subjected to explicit spatial coding by using a second convolutional neural network so as to further extract high-dimensional correlation information among the charging feature vectors, thereby obtaining a correlation feature map serving as a search space.
More specifically, in step S150, a control feature vector is obtained by matrix-multiplying a charging feature vector corresponding to a certain electric vehicle as a search state and the associated feature map as a search space. That is, a charging feature vector corresponding to a certain electric vehicle is set as a search state, the correlation feature map is set as a search space, and a control feature vector is obtained by multiplying the charging feature vector by the correlation feature map, so as to realize "control" of the first convolutional neural network as a sample generation network for generating the charging feature vector.
More specifically, in step S160, the charging feature vector and the control feature vector corresponding to a certain electric vehicle are multiplied by a location point to obtain a classification feature vector. It should be understood that the eigenvalues or combinations of eigenvalues in the eigenvectors, when regressed by the classifier, can be classified by explicit representation of the nonlinear features in the image, including harmonics (ripples), dips, unsteady signals, etc., but these eigenvalues or combinations of eigenvalues specifically correspond to what features are not understandable because they have been mapped into a high-dimensional space. Therefore, in the technical solution of the present application, regression needs to be implemented through parameter training of the classifier when the classifier is input. That is, the charging feature vector corresponding to a certain electric vehicle is multiplied by the obtained control feature vector by a position point to obtain a classification feature vector for classification.
More specifically, in step S170, the classified feature vectors are passed through a classifier to obtain a classification result representing an adaptive adjustment result of the charging of the electric vehicle. That is, first, the classification feature vector is input to the Softmax classification function of the classifier to obtain probability values that the classification feature vector is respectively attributed to tags representing adaptive adjustment results of charging. Specifically, the classification feature vector is input to a Softmax classification function of the classifier to obtain a first probability that the classification feature vector is respectively attributed to increasing charging current and a second probability that the classification feature vector is respectively attributed to decreasing charging current. Then, the label of the charging adaptive adjustment result corresponding to the maximum one of the probability values is determined as the classification result. That is, determining a result of the classification based on a comparison of the probability values, in particular, an increased charging current when the first probability is greater than the second probability; when the first probability is less than the second probability, the classification result is to reduce a charging current. It should be understood that, by the above scheme, end-to-end adaptive adjustment of the multiple electric vehicles charged simultaneously can be realized at one time, thereby improving the system performance of the adaptive charging system for the electric vehicles.
In summary, an operating method of the adaptive charging system for electric vehicles based on the embodiments of the present application is illustrated, which extracts high-dimensional features in a charging waveform image through parameter setting of convolution kernels for each layer in a convolutional neural network, while further facilitating training of the neural network by using a search space and a "controller" technique of the neural network in a search state, and implements regression of feature values or combinations of feature values in the feature vectors through parameter training of a classifier to implement classification of images through explicit expression of nonlinear features including harmonics (ripples), sudden changes, unsteady signals, and the like in the images. By the mode, end-to-end adaptive adjustment of multiple electric vehicles charged simultaneously can be realized at one time, so that the system performance of the adaptive charging system for the electric vehicles is improved.
The foregoing describes the general principles of the present application in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present application are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present application. Furthermore, the foregoing disclosure of specific details is for the purpose of illustration and description and is not intended to be limiting, since the foregoing disclosure is not intended to be exhaustive or to limit the disclosure to the precise details disclosed.
The block diagrams of devices, apparatuses, systems referred to in this application are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by those skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably therewith. The words "or" and "as used herein mean, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".
It should also be noted that in the devices, apparatuses, and methods of the present application, the components or steps may be decomposed and/or recombined. These decompositions and/or recombinations are to be considered as equivalents of the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the application to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.

Claims (10)

1. An adaptive charging system for an electric vehicle, comprising:
the charging data acquisition unit is used for acquiring charging waveform images of charging currents of a plurality of electric vehicles which are charged simultaneously;
a high-dimensional feature extraction unit, configured to input a charging waveform image of each electric vehicle that is simultaneously charged into a first convolution neural network respectively to extract high-dimensional features in the charging waveform image through parameter settings of convolution kernels of layers in the first convolution neural network to obtain a charging feature vector corresponding to each charging waveform image;
a two-dimensional cascade unit configured to cascade the charging feature vectors corresponding to each of the charging waveform images in two dimensions to obtain a charging feature matrix;
a search space construction unit, configured to perform explicit spatial coding on the charging feature matrix using a second convolutional neural network to extract high-dimensional correlation information between the charging feature vectors, so as to obtain a correlation feature map;
the searching unit is used for taking a charging characteristic vector corresponding to a certain electric automobile as a searching state, taking the associated characteristic diagram as a searching space, and multiplying the charging characteristic vector and the associated characteristic diagram by a matrix to obtain a control characteristic vector;
the vector fusion unit is used for multiplying the charging characteristic vector corresponding to a certain electric automobile and the control characteristic vector according to position points to obtain a classification characteristic vector; and
and the adaptive charging result generating unit is used for enabling the classified characteristic vectors to pass through a classifier so as to obtain a classification result representing an adaptive adjustment result of the charging of the electric automobile.
2. The adaptive charging system for electric vehicles according to claim 1, wherein the first convolutional neural network extracts a high-dimensional feature in the charging waveform image in the following formula to obtain the charging feature map;
wherein the formula is:
fi＝active(Ni×fi-1+Bi)
wherein f isi-1For the input of the first convolutional neural network of the i-th layer, fiFor the output of the first convolutional neural network of the ith layer, NiIs the convolution kernel of the ith layer first convolution neural network, and BiActive represents a nonlinear activation function for the bias vector of the first convolutional neural network of the ith layer.
3. The adaptive charging system for electric vehicles according to claim 2, wherein the last layer of the first convolutional neural network is used for global mean pooling processing of the charging feature map to obtain the charging feature vector.
4. The adaptive charging system for electric vehicles according to claim 3, wherein the two-dimensional cascade unit is further configured to arrange the charging feature vectors corresponding to each of the charging waveform images in rows of a matrix to obtain the charging feature matrix.
5. The adaptive charging system for electric vehicles of claim 4 wherein the second convolutional neural network explicitly spatially encodes the charging signature matrix to obtain the correlation signature;
wherein the formula is:
fi＝tanh(Ni×fi-1+Bi)
wherein f isi-1As input to the ith layer of the second convolutional neural network, fiIs the output of the ith layer of the second convolutional neural network, NiIs the convolution kernel of the ith layer of the second convolutional neural network, and BiFor the bias vector of the ith layer second convolutional neural network, tanh represents the nonlinear activation function.
6. The adaptive charging system for electric vehicles according to claim 5, wherein the adaptive charging result generating unit is further configured to input the classification feature vector into a Softmax classification function of the classifier to obtain probability values that the classification feature vector is respectively attributed to tags each representing an adaptive adjustment result of charging; and determining the label of the charging self-adaptive adjustment result corresponding to the maximum probability value as the classification result.
7. The adaptive charging system for electric vehicles according to claim 6, wherein the label for indicating the adaptive adjustment result of charging includes: increasing the charging current and decreasing the charging current.
8. An operating method of an adaptive charging system for an electric vehicle, comprising:
acquiring charging waveform images of charging currents of a plurality of electric vehicles which are charged simultaneously;
inputting a charging waveform image of each electric automobile which is charged simultaneously into a first convolution neural network respectively to extract high-dimensional features in the charging waveform image through parameter setting of convolution kernels of layers in the first convolution neural network so as to obtain a charging feature vector corresponding to each charging waveform image;
two-dimensionally concatenating charging feature vectors corresponding to each of the charging waveform images to obtain a charging feature matrix;
performing explicit spatial coding on the charging feature matrix by using a second convolutional neural network to extract high-dimensional correlation information between the charging feature vectors so as to obtain a correlation feature map;
taking a charging characteristic vector corresponding to a certain electric automobile as a searching state and the associated characteristic diagram as a searching space, and carrying out matrix multiplication on the charging characteristic vector and the associated characteristic diagram to obtain a control characteristic vector;
multiplying the charging characteristic vector corresponding to a certain electric automobile and the control characteristic vector according to position points to obtain a classification characteristic vector; and
and passing the classified feature vector through a classifier to obtain a classification result representing an adaptive adjustment result of the charging of the electric automobile.
9. The operating method of the adaptive charging system for electric vehicles according to claim 8, wherein the first convolutional neural network extracts a high-dimensional feature in the charging waveform image in the following formula to obtain the charging feature map;
wherein the formula is:
fi＝active(Ni×fi-1+Bi)
wherein f isi-1For the input of the first convolutional neural network of the i-th layer, fiFor the output of the first convolutional neural network of the ith layer, NiIs the convolution kernel of the ith layer first convolution neural network, and BiActive represents a nonlinear activation function for the bias vector of the first convolutional neural network of the ith layer.
10. The method of claim 8, wherein a last layer of the first convolutional neural network is used for performing global mean pooling on the charging feature map to obtain the charging feature vector.
CN202111444506.8A 2021-11-30 2021-11-30 Self-adaptive charging system for electric automobile and working method thereof Withdrawn CN114103710A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111444506.8A CN114103710A (en) 2021-11-30 2021-11-30 Self-adaptive charging system for electric automobile and working method thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111444506.8A CN114103710A (en) 2021-11-30 2021-11-30 Self-adaptive charging system for electric automobile and working method thereof

Publications (1)

Publication Number Publication Date
CN114103710A true CN114103710A (en) 2022-03-01

Family

ID=80368701

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111444506.8A Withdrawn CN114103710A (en) 2021-11-30 2021-11-30 Self-adaptive charging system for electric automobile and working method thereof

Country Status (1)

Country Link
CN (1) CN114103710A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114871486A (en) * 2022-06-21 2022-08-09 大畏机床(江苏)有限公司 Double housing planer and processing control method thereof
CN115514343A (en) * 2022-05-13 2022-12-23 浙江腾腾电气有限公司 Power grid waveform filtering system and filtering method thereof
CN115547466A (en) * 2022-09-05 2022-12-30 西安启玥华辰软件咨询开发有限公司 Medical institution registration and review system and method based on big data

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115514343A (en) * 2022-05-13 2022-12-23 浙江腾腾电气有限公司 Power grid waveform filtering system and filtering method thereof
CN115514343B (en) * 2022-05-13 2023-08-11 浙江腾腾电气有限公司 Power grid waveform filtering system and filtering method thereof
CN114871486A (en) * 2022-06-21 2022-08-09 大畏机床(江苏)有限公司 Double housing planer and processing control method thereof
CN114871486B (en) * 2022-06-21 2023-05-05 大畏机床(江苏)有限公司 Planer and processing control method thereof
CN115547466A (en) * 2022-09-05 2022-12-30 西安启玥华辰软件咨询开发有限公司 Medical institution registration and review system and method based on big data
CN115547466B (en) * 2022-09-05 2023-06-16 西安启玥华辰软件咨询开发有限公司 Medical institution registration and review system and method based on big data

Similar Documents

Publication Publication Date Title
CN114103710A (en) Self-adaptive charging system for electric automobile and working method thereof
CN109964250B (en) Method and system for analyzing images in convolutional neural networks
Zhu et al. Improving Deep Neural Network Sparsity through Decorrelation Regularization.
CN113985294B (en) Method and device for estimating remaining life of battery
CN115235612B (en) Intelligent fault diagnosis system and method for servo motor
CN115514343B (en) Power grid waveform filtering system and filtering method thereof
Liu et al. RB-Net: Training highly accurate and efficient binary neural networks with reshaped point-wise convolution and balanced activation
CN110378320A (en) The common cycle of multiple signals determines method, apparatus and readable storage medium storing program for executing
CN115081332B (en) Working condition sensitivity analysis and data processing method and device for parameter identification
CN114627369B (en) Environment monitoring system, method and computer device thereof
Chen et al. Uavnet: an efficient obstacel detection model for uav with autonomous flight
CN113538037A (en) Method, system, equipment and storage medium for monitoring charging event of battery car
Tikariha et al. Classification of faults in an IEEE 30 bus transmission system using fully convolutional network
CN115617636A (en) Distributed performance test system
Chen et al. Overload: Latency attacks on object detection for edge devices
CN115311509A (en) Power system transient stability evaluation method and system based on imaging data driving
Harandi et al. Face recognition using reinforcement learning
Zhang et al. Low probability of intercept radar signal recognition by staked autoencoder and SVM
CN115311527A (en) Neural network image identification method based on pulse statistics and electronic equipment
CN116975742A (en) Partial discharge pattern recognition method, apparatus, device, and storage medium
Im Choi et al. Visual-Saliency-Guided Channel Pruning for Deep Visual Detectors in Autonomous Driving
CN113708434A (en) Lithium battery charging and discharging intelligent power supply management system and operation method thereof
Um et al. A novel antenna pattern design using generative adversarial network
He et al. Partial discharge pattern recognition algorithm based on sparse self-coding and extreme learning machine
Yang et al. Power system transient stability assessment method based on convolutional neural network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WW01 Invention patent application withdrawn after publication
WW01 Invention patent application withdrawn after publication

Application publication date: 20220301