CN115830591A - Gas diffusion layer component identification method, device, equipment and storage medium - Google Patents
Gas diffusion layer component identification method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN115830591A CN115830591A CN202111083066.8A CN202111083066A CN115830591A CN 115830591 A CN115830591 A CN 115830591A CN 202111083066 A CN202111083066 A CN 202111083066A CN 115830591 A CN115830591 A CN 115830591A
- Authority
- CN
- China
- Prior art keywords
- component
- gas diffusion
- diffusion layer
- layer
- region information
- 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.)
- Pending
Links
- 238000009792 diffusion process Methods 0.000 title claims abstract description 108
- 238000000034 method Methods 0.000 title claims abstract description 64
- 238000003860 storage Methods 0.000 title claims abstract description 22
- 229920000049 Carbon (fiber) Polymers 0.000 claims abstract description 24
- 239000004917 carbon fiber Substances 0.000 claims abstract description 24
- 239000003795 chemical substances by application Substances 0.000 claims abstract description 15
- 239000011230 binding agent Substances 0.000 claims abstract description 14
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical group C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims description 53
- 230000006870 function Effects 0.000 claims description 30
- 238000010801 machine learning Methods 0.000 claims description 23
- 238000012545 processing Methods 0.000 claims description 10
- 238000000513 principal component analysis Methods 0.000 claims description 8
- 238000009826 distribution Methods 0.000 abstract description 10
- 238000005516 engineering process Methods 0.000 abstract description 9
- 238000013473 artificial intelligence Methods 0.000 abstract description 6
- 239000007789 gas Substances 0.000 description 89
- 239000000446 fuel Substances 0.000 description 21
- 238000001878 scanning electron micrograph Methods 0.000 description 13
- 239000000853 adhesive Substances 0.000 description 9
- 230000001070 adhesive effect Effects 0.000 description 9
- 239000012528 membrane Substances 0.000 description 7
- 239000004810 polytetrafluoroethylene Substances 0.000 description 6
- 229920001343 polytetrafluoroethylene Polymers 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 230000002209 hydrophobic effect Effects 0.000 description 5
- -1 Polytetrafluoroethylene Polymers 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000004626 scanning electron microscopy Methods 0.000 description 4
- 239000004890 Hydrophobing Agent Substances 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 3
- 239000003054 catalyst Substances 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 239000000835 fiber Substances 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 239000005011 phenolic resin Substances 0.000 description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 3
- KXGFMDJXCMQABM-UHFFFAOYSA-N 2-methoxy-6-methylphenol Chemical compound [CH]OC1=CC=CC([CH])=C1O KXGFMDJXCMQABM-UHFFFAOYSA-N 0.000 description 2
- UFHFLCQGNIYNRP-UHFFFAOYSA-N Hydrogen Chemical compound [H][H] UFHFLCQGNIYNRP-UHFFFAOYSA-N 0.000 description 2
- NBIIXXVUZAFLBC-UHFFFAOYSA-N Phosphoric acid Chemical compound OP(O)(O)=O NBIIXXVUZAFLBC-UHFFFAOYSA-N 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 238000013434 data augmentation Methods 0.000 description 2
- 238000010894 electron beam technology Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 239000001257 hydrogen Substances 0.000 description 2
- 229910052739 hydrogen Inorganic materials 0.000 description 2
- 229920001568 phenolic resin Polymers 0.000 description 2
- 239000011148 porous material Substances 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000003325 tomography Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- BVKZGUZCCUSVTD-UHFFFAOYSA-L Carbonate Chemical compound [O-]C([O-])=O BVKZGUZCCUSVTD-UHFFFAOYSA-L 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 238000004887 air purification Methods 0.000 description 1
- 229910000147 aluminium phosphate Inorganic materials 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000007731 hot pressing Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 239000001301 oxygen Substances 0.000 description 1
- 229910052760 oxygen Inorganic materials 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 239000000376 reactant Substances 0.000 description 1
- 239000012495 reaction gas Substances 0.000 description 1
- 230000002940 repellent Effects 0.000 description 1
- 239000005871 repellent Substances 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000009987 spinning Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
- 239000013585 weight reducing agent Substances 0.000 description 1
Images
Classifications
-
- 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
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/30—Hydrogen technology
- Y02E60/50—Fuel cells
Landscapes
- Image Analysis (AREA)
Abstract
The application relates to a gas diffusion layer component identification method, a device, equipment and a storage medium, wherein the method comprises the following steps: acquiring a scanned image of the gas diffusion layer; performing component identification on the scanned image according to the component identification model to obtain component area information corresponding to the scanned image; the component region information includes at least one of carbon fiber region information, binder region information, and hydrophobizing agent region information. The application finishes the identification of each component in the gas diffusion layer through the artificial intelligence image recognition technology, so that the distribution condition of each component in the gas diffusion layer can be effectively distinguished, the intelligence is high, and the probability of artificial errors can be reduced.
Description
Technical Field
The present disclosure relates to the field of fuel cell technologies, and in particular, to a method, an apparatus, a device, and a storage medium for identifying components of a gas diffusion layer.
Background
With the shortage of fossil energy such as petroleum and the urgent requirement of environmental protection, new energy automobiles become research hotspots of manufacturers and research and development institutions of various automobiles in the world. The fuel cell technology is a clean energy conversion technology with a great application prospect, chemical energy in fuel is directly converted into electric energy without fuel combustion, the energy conversion efficiency is not limited by Carnot cycle, and theoretically, the total energy utilization efficiency is more than 60%.
The fuel cells mainly include alkaline fuel cells, proton Exchange Membrane Fuel Cells (PEMFC), phosphoric acid fuel cells, molten carbonate fuel cells, and solid oxide fuel cells, among which proton exchange membrane fuel cells are most successfully commercialized. The PEMFC automobile uses hydrogen as input fuel, only pure water is output in reaction, and the PEMFC automobile has the advantages of no noise and pollution in work, short refueling time (3-5 minutes), long driving range (500-700 kilometers) and the like.
One of the most central components of the PEMFC is a stack (power source), which mainly includes fastening bolts, end plates, current collecting plates, bipolar plates (BPP), gaskets, gas Diffusion Layers (GDL), micro-porous layers (MPL), catalyst Layers (CL), and Proton Exchange Membranes (PEM), where PEM, CL, MPL, and GDL are collectively called a Membrane Electrode Assembly (MEA).
Among them, a Gas Diffusion Layer (GDL) is one of important electrode elements of a Membrane Electrode (MEA) in a PEMFC. The gas diffusion layer mainly functions to efficiently transport gaseous reactants to the catalyst layer (transport hydrogen and oxygen), transfer heat, conduct electric charges (have high electrical conductivity), remove water generated from the catalyst layer during operation of the fuel cell, and withstand external pressure to maintain the structural frame. The gas diffusion layer is mainly composed of randomly oriented carbon fibers, and has a thickness of generally 150 to 300 μm and a fiber diameter of 7 to 10 μm. The operating condition and output performance of the PEMFC are affected by the thickness of the gas diffusion layer, the porosity (defined as the ratio of the sum of all pore volumes inside the gas diffusion layer to the product of the length, width and height of the gas diffusion layer), the fiber diameter, the content of Polytetrafluoroethylene (PTFE), and the like, so that it is very important to correctly identify the distribution of each component of the gas diffusion layer to evaluate and optimize the performance of the cell.
The components of the gas diffusion layer mainly include carbon fibers, a binder (phenolic resin), a hydrophobic agent (polytetrafluoroethylene), and some other additives. Currently, there is no method for accurately identifying various components of a gas diffusion layer, and the mainstream method is to use SEM or XCT to explore the microstructure of the gas diffusion layer, and both methods rely on human eyes to distinguish different components, but it is difficult for the naked eye to define the distinct boundaries between the components (such as adhesive and carbon fibers).
Accordingly, there is a need to provide an intelligent and efficient method for identifying various types of components in a gas diffusion layer.
Disclosure of Invention
The embodiment of the application provides a gas diffusion layer component identification method, a device, equipment and a storage medium, and the identification of each component in a gas diffusion layer is completed through an artificial intelligence image identification technology, so that the distribution condition of each component in the gas diffusion layer can be effectively distinguished, the intelligence is high, and the probability of artificial errors can be reduced.
In one aspect, an embodiment of the present application provides a method for identifying a component of a gas diffusion layer, including:
acquiring a scanned image of the gas diffusion layer;
performing component identification on the scanned image according to the component identification model to obtain component area information corresponding to the scanned image; the component region information includes at least one of carbon fiber region information, binder region information, and hydrophobizing agent region information.
Optionally, the training method of the component recognition model includes:
acquiring a training sample image; the training sample image is a scanned image of the gas diffusion layer with component area identification information;
constructing a preset machine learning model;
based on a preset machine learning model, carrying out component identification on a training sample image to obtain predicted component region information;
determining a loss value based on the predicted component region information and the component region identification information;
and training the preset machine learning model according to the loss value until a preset training end condition is met, and obtaining a component recognition model.
Optionally, determining the loss value based on the predicted component region information and the component region identification information includes:
obtaining an initial cross entropy loss function;
adding a regularization term to the initial cross entropy loss function to obtain a target loss function;
and determining a loss value based on the target loss function, the predicted component region information and the component region identification information.
Optionally, the preset machine learning model includes an input layer, a convolution layer, a linear rectification layer, a discard layer, and an output layer;
the input of the input layer is a training sample image, the output end of the input layer is connected with the input end of the convolution layer, and the output end of the convolution layer is connected with the input end of the linear rectifying layer;
the output end of the linear rectification layer is connected with the input end of the discarding layer, the output end of the discarding layer is connected with the output layer, and the output of the output layer is the prediction component area information.
Optionally, obtaining a scanned image of the gas diffusion layer comprises:
acquiring an original scanning image of the gas diffusion layer;
and performing principal component analysis processing on the original scanning image to obtain a processed scanning image.
Optionally, the method further comprises:
and reconstructing the internal structure of the gas diffusion layer based on the component area information to obtain a three-dimensional model of the gas diffusion layer.
In another aspect, an embodiment of the present application provides a gas diffusion layer composition identification apparatus, including:
an acquisition module configured to perform acquiring a scan image of the gas diffusion layer;
the identification module is configured to perform component identification on the scanned image according to the component identification model to obtain component region information corresponding to the scanned image; the component region information includes at least one of carbon fiber region information, binder region information, and hydrophobizing agent region information.
Optionally, the apparatus further comprises a training module configured to perform:
acquiring a training sample image; the training sample image is a scanned image of the gas diffusion layer with component area identification information; constructing a preset machine learning model; based on a preset machine learning model, performing component identification on a training sample image to obtain predicted component region information; determining a loss value based on the predicted component region information and the component region identification information; and training the preset machine learning model according to the loss value until a preset training end condition is met, and obtaining a component recognition model.
In another aspect, an embodiment of the present application provides an apparatus, which includes a processor and a memory, where the memory stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded by the processor and executes the method for identifying a composition of a gas diffusion layer.
In another aspect, the present disclosure provides a computer storage medium, in which at least one instruction or at least one program is stored, and the at least one instruction or the at least one program is loaded and executed by a processor to implement the gas diffusion layer composition identification method.
The gas diffusion layer component identification method, device, equipment and storage medium provided by the embodiment of the application have the following beneficial effects:
by acquiring a scanned image of the gas diffusion layer; performing component identification on the scanned image according to the component identification model to obtain component area information corresponding to the scanned image; the component region information includes at least one of carbon fiber region information, binder region information, and hydrophobizing agent region information. The application finishes the identification of each component in the gas diffusion layer through the artificial intelligence image recognition technology, so that the distribution condition of each component in the gas diffusion layer can be effectively distinguished, the intelligence is high, and the probability of artificial errors can be reduced.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a schematic view of a scanned image of a gas diffusion layer provided in an embodiment of the present application;
FIG. 2 is a schematic flow chart of a method for identifying the composition of a gas diffusion layer according to an embodiment of the present disclosure;
FIG. 3 is a schematic flowchart of a method for training a component recognition model according to an embodiment of the present disclosure;
fig. 4 is a schematic flowchart of determining a loss value based on predicted component region information and component region identification information according to an embodiment of the present disclosure;
FIG. 5 is a schematic structural diagram of a gas diffusion layer composition identification device provided in an embodiment of the present application;
fig. 6 is a block diagram of a hardware structure of a server of a gas diffusion layer composition identification method according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and claims of this application and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or server that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
In many studies of fuel cells, the gas diffusion layer, as a key component in the fuel cell, accounts for 20-25% of the cost of the fuel cell, and the performance of the gas diffusion layer directly affects whether the fuel cell can work normally.
There are currently two main approaches to experimental study of gas diffusion layer microstructure: scanning Electron Microscopy (SEM) and X-ray tomography (XCT):
the electron beam emitted by the SEM from the sub-gun passes through the condenser lens along the optical axis of the lens body in the vacuum channel, and is converged into a beam of sharp, bright and uniform light spot by the condenser lens to irradiate on a sample in the sample chamber; the electron beam after passing through the sample carries the structural information of the sample. Fig. 1 (a) is a SEM image of a microstructure of a gas diffusion layer, and when observing the morphology, a composition analysis of a micro region is manually performed, and the carbon fibers are regarded as uniform in diameter, and are judged as carbon fibers all having a fixed radius along the axis direction of the carbon fibers, and the rest are binder phenol resin and hydrophobing agent polytetrafluoroethylene.
However, SEM has the following disadvantages:
1) When the scanning electron microscope method is applied to the research of the gas diffusion layer, although the characterization of the gas diffusion layer is clear, different components can be distinguished only by human eyes.
2) The method can not accurately distinguish the interfaces of the carbon fiber and other components by manually distinguishing the components, is difficult to distinguish the hydrophobic agent and the adhesive, and can not obtain an accurate structure because the SEM image belongs to the observation of three-dimensional shapes of the surface.
3) The SEM image does not reflect its internal three-dimensional structure, so numerical reconstruction can be performed only from the SEM image when exploring the three-dimensional structure.
The XCT technique is a non-invasive imaging method that generates a multi-slice two-dimensional image of an object in a specific region (which may be in multiple directions, as shown in fig. 1 (b), i.e., an XCT slice in a planar direction and a thickness direction) by tomography, and can characterize the microstructure inside the object without cutting the object. The basic principle of XCT is to rotate and scan an object along a single axis to obtain a series of two-dimensional images with different rotation angles, and further synthesize a three-dimensional structural image of a sample. Threshold segmentation is realized through custom codes, and two substances are distinguished: solid microstructures (fibers, binder, PTFE), and air.
XCT, however, has the following disadvantages:
1) XCT is costly and complex to operate.
2) The 3D structure directly obtained by the XCT threshold segmentation cannot separate the components of the gas diffusion layer, and the carbon fiber, the binder, and the water repellent agent are regarded as an integral structure.
It can be seen that the existing two common methods cannot accurately distinguish carbon fibers, adhesives and hydrophobing agents from pores in a gas diffusion layer of a proton exchange membrane fuel cell.
In view of this, the embodiment of the present application provides a method for identifying gas diffusion layer components, which completes extraction and identification of gas diffusion layer component characteristics through an artificial intelligence image identification technology, and the method can effectively distinguish distribution conditions of each component in a gas diffusion layer, has high intelligence, and can reduce the probability of artificial errors.
While specific embodiments of a method for identifying a composition of a gas diffusion layer according to the present application are described below, fig. 2 is a schematic flow chart of a method for identifying a composition of a gas diffusion layer according to embodiments of the present application, and the present specification provides the method steps as in the embodiments or the flow chart, but may include more or fewer steps based on conventional or non-inventive labor. The order of steps recited in the embodiments is merely one manner of performing the steps in a multitude of orders and does not represent the only order of execution. In practice, the system or server product may be implemented in a sequential or parallel manner (e.g., parallel processor or multi-threaded environment) according to the embodiments or methods shown in the figures. Specifically, as shown in fig. 2, the method may include:
in step S201, a scanned image of the gas diffusion layer is acquired.
In the embodiment of the present application, the SEM image of the gas diffusion layer is selected as the data source of the component identification method because of the high availability of the SEM image, the cost thereof is much lower than that of the XCT image, and the SEM image is a clear structural feature of the reaction gas diffusion layer. Therefore, the scanned image may be specifically an SEM image of the gas diffusion layer.
In an alternative embodiment, obtaining a scanned image of a gas diffusion layer comprises:
acquiring an original scanning image of the gas diffusion layer;
and performing principal component analysis processing on the original scanning image to obtain a processed scanning image.
Principal Component Analysis (PCA) is a mathematical dimension reduction method, in which a series of variables that may be linearly related are converted into a set of new linearly unrelated variables, also called principal components, by orthogonal transformation (orthogonal transformation), so that the new variables are used to display the features of the data in a smaller dimension.
The main purpose of the principal component analysis processing is to perform data classification and feature extraction on SEM images of other diffusion layers, reduce the complexity of data and facilitate the reading-in of a subsequent component recognition model. The amount of information in the SEM images is relatively large, and using principal component analysis processing, the amount of model training can be reduced, so that the final trained model can handle simpler input data, e.g., lower magnification images from a laser scanning confocal microscope.
In step S203, performing component identification on the scanned image according to the component identification model to obtain component region information corresponding to the scanned image; the component region information includes at least one of carbon fiber region information, binder region information, and hydrophobizing agent region information.
In the embodiment of the application, a component identification model is obtained through training, a scanned image is used as the input of the component identification model, the component identification model carries out component identification on the scanned image, and corresponding component region information is output; the component region information includes at least one of carbon fiber region information, binder region information, and hydrophobizing agent region information.
In an alternative embodiment, as shown in fig. 3, the training method of the component recognition model may include the following steps:
in step S301, a training sample image is acquired; the training sample image is a scanned image of the gas diffusion layer with identification information of the component areas.
In the step, the SEM image of the collected gas diffusion layer is used as a training sample, the training sample image is processed in a manual mode, areas of carbon fibers, adhesives and hydrophobing agents in the training sample image are divided, and different components are marked by different area identification information. The characterization of different components on SEM images is different, the carbon fibers are straight cylinders with regular shapes, the adhesive is irregular shapes at the junctions of the carbon fibers, and the hydrophobic agent is usually attached to the adhesive.
It should be noted that, for gas diffusion layers of different categories, the component distribution conditions are greatly different, and a component identification model corresponding to each category can be trained separately by using training sample images of corresponding categories; for example, for a hydrophobic gas diffusion layer and a hydrophilic gas diffusion layer, training sample images can be collected and trained to obtain two component recognition models. Thus, the accuracy of model identification can be improved.
In step S303, a preset machine learning model is constructed.
The preset machine learning model can adopt a Convolutional Neural Network (CNN), which is a feed-forward Neural Network and has excellent performance on large-scale image processing.
In step S305, based on a preset machine learning model, component recognition is performed on a training sample image, so as to obtain predicted component region information.
The convolutional neural network is composed of an input layer and an output layer and a plurality of hidden layers. Therefore, in an alternative embodiment, the pre-set machine learning model may include an input layer, a convolutional layer, a linear rectifying layer, a discard layer, and an output layer;
the input of the input layer is a training sample image, the output end of the input layer is connected with the input end of the convolution layer, and the output end of the convolution layer is connected with the input end of the linear rectifying layer;
the output end of the linear rectification layer is connected with the input end of the discarding layer, the output end of the discarding layer is connected with the output layer, and the output of the output layer is the prediction component region information.
The convolution layer is used for extracting bottom layer characteristics of the image. A convolutional layer is a set of parallel feature maps (feature maps) that are composed by sliding different convolutional kernels over the input image and performing certain operations. In addition, at each sliding position, the convolution kernel and the input image perform an element-by-element multiplication and summation operation to project the information in the receptive field to an element in the feature map. This sliding process can be called stride, which is a factor in controlling the size of the output feature map and is a hyper-parameter that we need to set. The convolution kernel is much smaller in size than the input image and acts in the input image in an overlapping or parallel manner, and all elements in a feature map are calculated by one convolution kernel, that is, a feature map shares the same weight and bias term.
The Linear rectification layer (ReLU layer) uses Linear rectification (ReLU) as an excitation function (Activation function) of the nerves of this layer. It can enhance the non-linear characteristics of the decision function and the entire neural network without itself altering the convolutional layer.
Often, overfitting results when there are not enough training samples, or overfertiming. To prevent overfitting, many methods can be used, such as early stopping, data augmentation (Data augmentation), regularization (Regularization) including L1, L2 (L2 Regularization is also called weight reduction), dropout.
In this optional embodiment, by discarding the layer, when training the model is started, a part of hidden layer units is randomly "discarded" in the first iteration, the input and output are kept unchanged, and the weight of the neural network is updated according to the BP algorithm. And continuously deleting part of hidden layer units in the second iteration by adopting the same method until the training is finished.
In step S307, a loss value is determined based on the predicted component region information and the component region identification information.
Specifically, as shown in fig. 4, the step S307 may include the following steps:
in step S401, an initial cross entropy loss function is obtained;
the loss function is used for evaluating the degree of the model with different predicted values and actual values, and the better the loss function is, the better the performance of the model is generally. This step takes the form of a cross entropy loss function as follows:
wherein x represents a sample; y represents the actual label; a represents the predicted output; n represents the total number of samples. The problem that the updating of the weight of the perfect outcome square loss function is too slow can be solved by using the matching of the cross entropy loss function and the sigmoid activation function.
In step S403, adding a regularization term to the initial cross entropy loss function to obtain a target loss function;
in the step, an L2 regularization method is adopted to prevent the model from being over-fitted, a regularization item is called after the cross entropy loss function, and the obtained target loss function is as follows:
wherein, C 0 Is the original loss function;the L2 regularization term is obtained by dividing the sum of the squares of all parameters by the gas diffusion layer image data training set sample size.
In step S405, a loss value is determined based on the target loss function, the predicted component region information, and the component region identification information.
In the step, according to the determined target loss function, the predicted component area information and the component area identification information, a loss value is obtained through solving.
In step S309, the preset machine learning model is trained according to the loss value until a preset training end condition is satisfied, so as to obtain a component recognition model.
The preset training end condition may include any one of the following: the iteration times exceed the preset times; alternatively, the current loss value is less than the threshold.
Specifically, mass fractions of components in the SEM image predicted by the model are compared with the true values, and when errors are less than 20%, a component identification model is obtained.
In summary, the component identification method of the gas diffusion layer provided by the application can solve the problem of component identification of the gas diffusion layer of the proton exchange membrane fuel cell by the image identification technology based on the artificial intelligence algorithm, and accurately distinguish the component distribution of carbon fiber, adhesive (phenolic resin), hydrophobic agent (polytetrafluoroethylene), holes, breakpoints and the like.
Furthermore, the identified components of the finished product of the gas diffusion layer can be fed back to a gas diffusion layer manufacturer in time, so that the manufacturer can adjust the product production material ratio according to the information, and can indicate that the gas diffusion layer and other porous media and other related material manufacturing companies better plan the process flow and the manufacturing scheme.
In an optional implementation manner, the gas diffusion layer group separation method in the examples of the present application further includes:
and reconstructing the internal structure of the gas diffusion layer based on the component area information to obtain a three-dimensional model of the gas diffusion layer.
In the embodiment of the application, the distribution conditions of different materials can be accurately distinguished based on the component region information; further, based on the correlation between the surface characteristics and the three-dimensional characteristics of the gas diffusion layer in the hot pressing process, the reduction of the whole three-dimensional structure of the gas diffusion layer can be established by accurately identifying and summarizing the surface components and the component growth rules of the gas diffusion layer of the fuel cell.
Meanwhile, after scanned images are effectively distinguished, accurate microstructure parameters of the gas diffusion layer, such as the diameter of carbon fibers, the content of the adhesive and the distribution rule of the adhesive, can be provided for scientific researchers, the precision of a reconstructed model can be improved, and simulation research is closer to the real situation.
In addition, the gas diffusion layer component identification method provided by the embodiment of the application can be popularized from the fuel cell gas diffusion layer industry to other carbon paper and porous medium industries, can solve the research and manufacturing problems of porous media in the fields of petrifaction, air purification, spinning, papermaking, aerospace, atomic energy and the like, and has great scientific significance and industrial value.
An embodiment of the present application further provides a gas diffusion layer composition identification device, and fig. 5 is a schematic structural diagram of the gas diffusion layer composition identification device provided in the embodiment of the present application, and as shown in fig. 5, the device includes:
an acquisition module 501 configured to perform acquisition of a scanned image of the gas diffusion layer;
the identification module 502 is configured to perform component identification on the scanned image according to the component identification model, so as to obtain component region information corresponding to the scanned image; the component region information includes at least one of carbon fiber region information, binder region information, and hydrophobizing agent region information.
Optionally, the apparatus further comprises a training module configured to perform:
acquiring a training sample image; the training sample image is a scanned image of the gas diffusion layer with component area identification information; constructing a preset machine learning model; based on a preset machine learning model, carrying out component identification on a training sample image to obtain predicted component region information; determining a loss value based on the predicted component region information and the component region identification information; and training the preset machine learning model according to the loss value until a preset training end condition is met, and obtaining a component recognition model.
Optionally, the training module is configured to perform:
obtaining an initial cross entropy loss function;
adding a regularization term to the initial cross entropy loss function to obtain a target loss function;
and determining a loss value based on the target loss function, the predicted component region information and the component region identification information.
Optionally, the preset machine learning model includes an input layer, a convolution layer, a linear rectification layer, a discard layer, and an output layer;
the input of the input layer is a training sample image, the output end of the input layer is connected with the input end of the convolution layer, and the output end of the convolution layer is connected with the input end of the linear rectifying layer;
the output end of the linear rectification layer is connected with the input end of the discarding layer, the output end of the discarding layer is connected with the output layer, and the output of the output layer is the prediction component area information.
Optionally, the obtaining module 501 is configured to perform:
acquiring an original scanning image of the gas diffusion layer;
and performing principal component analysis processing on the original scanning image to obtain a processed scanning image.
Optionally, the apparatus further comprises a reconstruction module configured to perform:
and reconstructing the internal structure of the gas diffusion layer based on the component area information to obtain a three-dimensional model of the gas diffusion layer.
The device and method embodiments in the embodiments of the present application are based on the same application concept.
The method provided by the embodiment of the application can be executed in a computer terminal, a server or a similar operation device. Taking the example of the application running on a server, fig. 6 is a hardware structure block diagram of the server of the gas diffusion layer component identification method provided in the embodiment of the present application. As shown in fig. 6, the server 600 may have a relatively large difference due to different configurations or performances, and may include one or more Central Processing Units (CPUs) 610 (the processors 610 may include but are not limited to a Processing device such as a microprocessor NCU or a programmable logic device FPGA), a memory 630 for storing data, and one or more storage media 620 (e.g., one or more mass storage devices) for storing applications 623 or data 622. Memory 630 and storage medium 620 may be, among other things, transient storage or persistent storage. The program stored on the storage medium 620 may include one or more modules, each of which may include a series of instruction operations for the server. Still further, the central processor 610 may be configured to communicate with the storage medium 620 to execute a series of instruction operations in the storage medium 620 on the server 600. The server 600 may also include one or more power supplies 660, one or more wired or wireless network interfaces 650, one or more input-output interfaces 640, and/or one or more operating systems 621, such as Windows, mac OS, unix, linux, freeBSD, and the like.
The input/output interface 640 may be used to receive or transmit data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider of the server 600. In one example, i/o Interface 640 includes a Network adapter (NIC) that may be coupled to other Network devices via a base station to communicate with the internet. In one example, the input/output interface 640 may be a Radio Frequency (RF) module, which is used for communicating with the internet in a wireless manner.
It will be understood by those skilled in the art that the structure shown in fig. 6 is only an illustration and is not intended to limit the structure of the electronic device. For example, server 600 may also include more or fewer components than shown in FIG. 6, or have a different configuration than shown in FIG. 6.
Embodiments of the present application also provide a storage medium that may be disposed in a server to store at least one instruction, at least one program, a set of codes, or a set of instructions related to implementing a gas diffusion layer composition identification method in method embodiments, the at least one instruction, the at least one program, the set of codes, or the set of instructions being loaded and executed by the processor to implement the gas diffusion layer composition identification method.
Alternatively, in this embodiment, the storage medium may be located in at least one network server of a plurality of network servers of a computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to: a U-disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a removable hard disk, a magnetic or optical disk, and other various media capable of storing program codes.
As can be seen from the above embodiments of a method, an apparatus, a device and a storage medium for identifying components of a gas diffusion layer provided by the present application, the present application obtains a scanned image of a gas diffusion layer; performing component identification on the scanned image according to the component identification model to obtain component area information corresponding to the scanned image; the component region information includes at least one of carbon fiber region information, binder region information, and hydrophobizing agent region information. The application finishes the identification of each component in the gas diffusion layer through the artificial intelligence image recognition technology, so that the distribution condition of each component in the gas diffusion layer can be effectively distinguished, the intelligence is high, and the probability of artificial errors can be reduced.
It should be noted that: the sequence of the embodiments of the present application is only for description, and does not represent the advantages or disadvantages of the embodiments. And specific embodiments thereof have been described above. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the apparatus embodiment, since it is substantially similar to the method embodiment, the description is relatively simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, where the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.
The above description is only exemplary of the present application and should not be taken as limiting the present application, as any modification, equivalent replacement, or improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.
Claims (10)
1. A method of identifying a composition of a gas diffusion layer, comprising:
acquiring a scanned image of the gas diffusion layer;
performing component identification on the scanned image according to a component identification model to obtain component area information corresponding to the scanned image; the component region information includes at least one of carbon fiber region information, binder region information, and hydrophobizing agent region information.
2. The method of claim 1, wherein the training method of the component recognition model comprises:
acquiring a training sample image; the training sample image is a scanned image of the gas diffusion layer with component area identification information;
constructing a preset machine learning model;
based on the preset machine learning model, carrying out component identification on the training sample image to obtain predicted component region information;
determining a loss value based on the predicted component region information and the component region identification information;
and training the preset machine learning model according to the loss value until a preset training end condition is met, and obtaining the component recognition model.
3. The method of claim 2, wherein determining a loss value based on the predicted component region information and the component region identification information comprises:
obtaining an initial cross entropy loss function;
adding a regularization term to the initial cross entropy loss function to obtain a target loss function;
determining the loss value based on the target loss function, the predicted component region information, and the component region identification information.
4. The method of claim 2 or 3, wherein the pre-set machine learning model comprises an input layer, a convolution layer, a linear rectification layer, a discard layer, and an output layer;
the input of the input layer is the training sample image, the output end of the input layer is connected with the input end of the convolutional layer, and the output end of the convolutional layer is connected with the input end of the linear rectifying layer;
the output end of the linear rectification layer is connected with the input end of the discarding layer, the output end of the discarding layer is connected with the output layer, and the output of the output layer is the prediction component region information.
5. The method of claim 1, wherein the obtaining a scanned image of the gas diffusion layer comprises:
acquiring an original scanning image of the gas diffusion layer;
and performing principal component analysis processing on the original scanning image to obtain a processed scanning image.
6. The method according to any one of claims 1-5, further comprising:
and reconstructing the internal structure of the gas diffusion layer based on the component area information to obtain a three-dimensional model of the gas diffusion layer.
7. A gas diffusion layer composition identification device, comprising:
an acquisition module configured to perform acquiring a scan image of the gas diffusion layer;
the identification module is configured to perform component identification on the scanned image according to a component identification model to obtain component region information corresponding to the scanned image; the component region information includes at least one of carbon fiber region information, binder region information, and hydrophobizing agent region information.
8. The apparatus of claim 5, further comprising a training module configured to perform:
acquiring a training sample image; the training sample image is a scanned image of the gas diffusion layer with component area identification information; constructing a preset machine learning model; based on the preset machine learning model, performing component identification on the training sample image to obtain predicted component region information; determining a loss value based on the predicted component region information and the component region identification information; and training the preset machine learning model according to the loss value until a preset training end condition is met, and obtaining the component recognition model.
9. An apparatus comprising a processor and a memory, wherein the memory has stored therein at least one instruction or at least one program, the at least one instruction or the at least one program being loaded by the processor and executing the gas diffusion layer composition identification method according to any one of claims 1 to 6.
10. A computer storage medium having stored therein at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by a processor to perform a gas diffusion layer composition identification method according to any one of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111083066.8A CN115830591A (en) | 2021-09-15 | 2021-09-15 | Gas diffusion layer component identification method, device, equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111083066.8A CN115830591A (en) | 2021-09-15 | 2021-09-15 | Gas diffusion layer component identification method, device, equipment and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115830591A true CN115830591A (en) | 2023-03-21 |
Family
ID=85515587
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111083066.8A Pending CN115830591A (en) | 2021-09-15 | 2021-09-15 | Gas diffusion layer component identification method, device, equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115830591A (en) |
-
2021
- 2021-09-15 CN CN202111083066.8A patent/CN115830591A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ming et al. | A systematic review of machine learning methods applied to fuel cells in performance evaluation, durability prediction, and application monitoring | |
Choi et al. | An SAO-based text mining approach to building a technology tree for technology planning | |
US20230306761A1 (en) | Methods for identifying cross-modal features from spatially resolved data sets | |
CN114677677B (en) | Method for predicting material proportion of gas diffusion layer of proton exchange membrane fuel cell | |
Xu et al. | Microstructure reconstruction of battery polymer separators by fusing 2D and 3D image data for transport property analysis | |
Hwang et al. | Integrated application of semantic segmentation-assisted deep learning to quantitative multi-phased microstructural analysis in composite materials: Case study of cathode composite materials of solid oxide fuel cells | |
Zhang et al. | Machine learning-facilitated multiscale imaging for energy materials | |
Rui et al. | Design of proton exchange membranes with high durability for fuel cells: From the perspective of machine learning | |
Lan et al. | Image recognition of cracks and the effect in the microporous layer of proton exchange membrane fuel cells on performance | |
CN114565716A (en) | Image three-dimensional reconstruction method, device, equipment and medium | |
Rubio et al. | A Novel System-Level Model for a Fuel Cell in a Strategic Context | |
CN115830591A (en) | Gas diffusion layer component identification method, device, equipment and storage medium | |
CN115017803B (en) | Method for designing local and global directional regulation and control of thermal mass resistance in porous structure of fuel cell stack | |
Arcot et al. | Morphological characteristics of catalyst layer defects in catalyst-coated membranes in PEM fuel cells | |
CN115620104A (en) | Proton exchange membrane fuel cell gas diffusion layer performance prediction method | |
CN114863028A (en) | Three-dimensional reconstruction method for gas diffusion layer of proton exchange membrane fuel cell | |
Xu et al. | The recycling of waste per-fluorinated sulfonic acid for reformulation and membrane application in iron-chromium redox flow batteries | |
Han et al. | Intelligent design and synthesis of energy catalytic materials | |
CN112465838A (en) | Ceramic crystal grain image segmentation method, system, storage medium and computer equipment | |
Chen et al. | Using multi-threshold non-local means joint distribution method to analysis the spatial distribution patterns of binder and fibers in gas diffusion layers of fuel cells | |
CN116070513A (en) | Method for predicting assembly compression performance of gas diffusion layer | |
Thiele et al. | Three dimensional Reconstruction of PEFC Catalyst Layers | |
CN113889646B (en) | Performance analysis method, device, equipment and storage medium for fuel cell catalytic layer | |
CN116341362A (en) | Method for predicting compression deformation of three-dimensional structure of gas diffusion layer | |
Zuo | Texture Appreciation of Central Plain Folk Handicrafts Based on Electron Microscope. |
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 |