CN116825227B - Perovskite component proportion analysis method and device based on depth generation model - Google Patents
Perovskite component proportion analysis method and device based on depth generation model Download PDFInfo
- Publication number
- CN116825227B CN116825227B CN202311107647.XA CN202311107647A CN116825227B CN 116825227 B CN116825227 B CN 116825227B CN 202311107647 A CN202311107647 A CN 202311107647A CN 116825227 B CN116825227 B CN 116825227B
- Authority
- CN
- China
- Prior art keywords
- perovskite
- conversion
- analysis model
- efficiency
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 215
- 238000006243 chemical reaction Methods 0.000 claims abstract description 280
- 238000000034 method Methods 0.000 claims abstract description 29
- 230000006870 function Effects 0.000 claims description 144
- 238000012549 training Methods 0.000 claims description 106
- 230000004913 activation Effects 0.000 claims description 56
- 238000012360 testing method Methods 0.000 claims description 38
- 230000009466 transformation Effects 0.000 claims description 22
- 238000004140 cleaning Methods 0.000 claims description 18
- 238000004422 calculation algorithm Methods 0.000 claims description 17
- 229910052740 iodine Inorganic materials 0.000 claims description 16
- 150000004820 halides Chemical class 0.000 claims description 14
- 238000005315 distribution function Methods 0.000 claims description 12
- 238000010276 construction Methods 0.000 claims description 11
- 238000005457 optimization Methods 0.000 claims description 11
- 238000010606 normalization Methods 0.000 claims description 8
- 238000012795 verification Methods 0.000 claims description 7
- 229910052794 bromium Inorganic materials 0.000 claims description 4
- 229910052792 caesium Inorganic materials 0.000 claims description 4
- 238000002474 experimental method Methods 0.000 claims description 4
- 238000000926 separation method Methods 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 3
- 230000008485 antagonism Effects 0.000 claims 1
- 239000000463 material Substances 0.000 abstract description 22
- 238000011161 development Methods 0.000 abstract description 6
- 238000013461 design Methods 0.000 abstract description 5
- 239000000306 component Substances 0.000 description 148
- 239000000203 mixture Substances 0.000 description 14
- 238000003860 storage Methods 0.000 description 12
- KFZMGEQAYNKOFK-UHFFFAOYSA-N Isopropanol Chemical compound CC(C)O KFZMGEQAYNKOFK-UHFFFAOYSA-N 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 239000000460 chlorine Substances 0.000 description 5
- 238000001228 spectrum Methods 0.000 description 5
- 238000004590 computer program Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000009826 distribution Methods 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 2
- BAVYZALUXZFZLV-UHFFFAOYSA-N Methylamine Chemical compound NC BAVYZALUXZFZLV-UHFFFAOYSA-N 0.000 description 2
- 238000010521 absorption reaction Methods 0.000 description 2
- 238000000137 annealing Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 239000000470 constituent Substances 0.000 description 2
- 239000008358 core component Substances 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 239000010931 gold Substances 0.000 description 2
- RQQRAHKHDFPBMC-UHFFFAOYSA-L lead(ii) iodide Chemical compound I[Pb]I RQQRAHKHDFPBMC-UHFFFAOYSA-L 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- -1 property analysis Substances 0.000 description 2
- 230000005855 radiation Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 229910052701 rubidium Inorganic materials 0.000 description 2
- IGLNJRXAVVLDKE-UHFFFAOYSA-N rubidium atom Chemical compound [Rb] IGLNJRXAVVLDKE-UHFFFAOYSA-N 0.000 description 2
- XOLBLPGZBRYERU-UHFFFAOYSA-N tin dioxide Chemical compound O=[Sn]=O XOLBLPGZBRYERU-UHFFFAOYSA-N 0.000 description 2
- ZCYVEMRRCGMTRW-UHFFFAOYSA-N 7553-56-2 Chemical compound [I] ZCYVEMRRCGMTRW-UHFFFAOYSA-N 0.000 description 1
- WKBOTKDWSSQWDR-UHFFFAOYSA-N Bromine atom Chemical compound [Br] WKBOTKDWSSQWDR-UHFFFAOYSA-N 0.000 description 1
- ZAMOUSCENKQFHK-UHFFFAOYSA-N Chlorine atom Chemical compound [Cl] ZAMOUSCENKQFHK-UHFFFAOYSA-N 0.000 description 1
- IAZDPXIOMUYVGZ-UHFFFAOYSA-N Dimethylsulphoxide Chemical compound CS(C)=O IAZDPXIOMUYVGZ-UHFFFAOYSA-N 0.000 description 1
- PNKUSGQVOMIXLU-UHFFFAOYSA-N Formamidine Chemical compound NC=N PNKUSGQVOMIXLU-UHFFFAOYSA-N 0.000 description 1
- ZMXDDKWLCZADIW-UHFFFAOYSA-N N,N-Dimethylformamide Chemical compound CN(C)C=O ZMXDDKWLCZADIW-UHFFFAOYSA-N 0.000 description 1
- 229910006404 SnO 2 Inorganic materials 0.000 description 1
- ATJFFYVFTNAWJD-UHFFFAOYSA-N Tin Chemical compound [Sn] ATJFFYVFTNAWJD-UHFFFAOYSA-N 0.000 description 1
- 239000006096 absorbing agent Substances 0.000 description 1
- 239000012296 anti-solvent Substances 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- GDTBXPJZTBHREO-UHFFFAOYSA-N bromine Substances BrBr GDTBXPJZTBHREO-UHFFFAOYSA-N 0.000 description 1
- TVFDJXOCXUVLDH-UHFFFAOYSA-N caesium atom Chemical compound [Cs] TVFDJXOCXUVLDH-UHFFFAOYSA-N 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 229910052801 chlorine Inorganic materials 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 229910021419 crystalline silicon Inorganic materials 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 1
- 229910052737 gold Inorganic materials 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- AMGQUBHHOARCQH-UHFFFAOYSA-N indium;oxotin Chemical compound [In].[Sn]=O AMGQUBHHOARCQH-UHFFFAOYSA-N 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000011630 iodine Substances 0.000 description 1
- HWSZZLVAJGOAAY-UHFFFAOYSA-L lead(II) chloride Chemical compound Cl[Pb]Cl HWSZZLVAJGOAAY-UHFFFAOYSA-L 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 239000003960 organic solvent Substances 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 239000002904 solvent Substances 0.000 description 1
- 239000007858 starting material Substances 0.000 description 1
- 238000010408 sweeping Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 238000012418 validation experiment Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/20—Identification of molecular entities, parts thereof or of chemical compositions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0475—Generative networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/0985—Hyperparameter optimisation; Meta-learning; Learning-to-learn
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/70—Machine learning, data mining or chemometrics
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Crystallography & Structural Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Photovoltaic Devices (AREA)
Abstract
The invention relates to the technical field of new material performance analysis, and discloses a perovskite component proportion analysis method and device based on a depth generation model, wherein the method comprises the following steps: the perovskite conversion analysis model to be trained is built, the perovskite conversion analysis model is trained by an encoder, a generator, a tag discriminator and a feature discriminator by utilizing a perovskite standard data set, a trained perovskite conversion analysis model is obtained, a target efficiency conversion grade is reversely input into the trained perovskite conversion analysis model, a predicted target perovskite component proportion receiving target efficiency conversion grade is obtained, and the target efficiency conversion grade is reversely input into the trained perovskite conversion analysis model, so that the predicted target perovskite component proportion is obtained. The invention mainly aims to solve the problems that the existing perovskite component design only depends on the existing experience, and the development of perovskite with high photoelectric conversion efficiency needs to consume higher experimental trial-and-error cost.
Description
Technical Field
The invention relates to a perovskite component proportion analysis method and device based on a depth generation model, and belongs to the technical field of new material performance analysis.
Background
Under the development environment of the current science and technology, the continuous improvement of energy efficiency and energy conservation and emission reduction become a core way for realizing sustainable development. Solar energy has been attracting attention in recent years as the most abundant renewable clean energy source. To explore new photovoltaic materials with high efficiency and low cost, researchers have long been working on finding perovskite materials with excellent photovoltaic properties. Perovskite materials are of great interest because of their high absorption coefficient, high carrier mobility, and relatively high defect tolerance. The absorption wavelength range of the perovskite material to sunlight can be changed while the energy gap of the perovskite material is changed by adjusting the proportion of the perovskite material. According to theoretical calculation, the Photoelectric Conversion Efficiency (PCE) limit of the single junction perovskite solar cell can reach 33% which is higher than 29% of that of a crystalline silicon cell. The PCE of perovskite solar cells with 3D organic-inorganic hybrid halide perovskite as the absorber layer has been rapidly increased from the original 3.8% to the current 25.7% over the last decade by multiple optimizations of the perovskite chemistry. At present, precise regulation of the components of perovskite materials has become one of the effective ways to improve PCE, however, due to the diversity of the combining methods of perovskite raw materials, it is still difficult to form highly effective and convincing conclusions through the "trial and error method. Therefore, how to perform targeted component regulation and control on the perovskite material through a scientific and effective method becomes a difficult problem to be solved in the current research.
With the rapid development of computer science, artificial Intelligence (AI) has made a significant breakthrough in the fields of image processing, pattern recognition, natural language processing, and the like. Deep learning technology, an important method of machine learning, has been gradually applied to the discovery of new materials, property analysis, material interaction, and the like. The new material can be rapidly analyzed and screened according to the material properties by utilizing the deep learning and other technologies under the fourth scientific range, so that the searching range of candidate materials is greatly reduced, guidance is better provided for experimental verification, and the development process of the new perovskite material is accelerated. Therefore, the perovskite material composition is analyzed by combining a deep learning model, so that the photoelectric conversion efficiency of the perovskite to the solar cell is improved, and the method has a broad prospect.
Because of the variety of methods of combining perovskite starting materials, it is still difficult to develop highly efficient and convincing conclusions by "trial and error" methods. Therefore, the current perovskite component design relies on only the prior experience, and has the problem that the development of perovskite with high photoelectric conversion efficiency needs to consume higher experimental trial-and-error cost.
Disclosure of Invention
The invention provides a perovskite component proportion analysis method and device based on a depth generation model and a computer readable storage medium, and mainly aims to solve the problem that the existing perovskite component design only depends on the existing experience, and the perovskite with high photoelectric conversion efficiency needs to be developed with higher experimental trial-and-error cost.
In order to achieve the above object, the perovskite component proportion analysis method based on the depth generation model provided by the invention comprises the following steps:
acquiring a perovskite original data set, wherein the perovskite original data set is derived from a public database and a discussion set;
performing data cleaning on the perovskite original data set to obtain a perovskite standard data set, wherein each perovskite standard data in the perovskite standard data set consists of proportion values and conversion efficiency grades of eight components, and the eight components are MA, FA, cs, rb, pb, sn, br and I;
constructing a perovskite conversion analysis model to be trained, wherein the perovskite conversion analysis model consists of an encoder, a generator, a tag discriminator and a feature discriminator;
the construction of the perovskite conversion analysis model to be trained comprises the following steps:
determining a first activation function and a second activation function, wherein the first activation function is a LeakyReLU, and the second activation function is a ReLU;
constructing and obtaining an encoder according to the sequence of the one-dimensional convolution layer, the batch normalization function and the first activation function;
constructing a generator based on three one-dimensional transpose convolutions and a second activation function;
constructing and obtaining the perovskite conversion analysis model according to the front of the encoder and the rear of the generator and by combining a pre-constructed tag discriminator and a feature discriminator; the tag identifier and the feature identifier respectively comprise two fully-connected layers, and share three one-dimensional convolution layers, and the layers are linked by using a LeakyReLU activation function;
Training a perovskite conversion analysis model by using the perovskite standard data set until the training condition is met and the training is exited, so as to obtain a trained perovskite conversion analysis model;
training the perovskite conversion analysis model by using the perovskite standard data set until the training condition is met, and quitting the training to obtain a trained perovskite conversion analysis model, wherein the training comprises the following steps:
setting a loss function and an optimization algorithm of a perovskite conversion analysis model, wherein the loss function comprises three loss functions, namely a reconstruction loss function, a KL divergence function and an anti-loss function;
constructing an input feature, wherein the input feature is composed of a first channel feature and a second channel feature, and the first channel feature comprises a perovskite host component and the second channel feature comprises a perovskite doping component;
the input features are imported into a perovskite conversion analysis model, and the dimensionality of the input features is reduced based on an encoder, so that low-dimensional features are obtained;
reconstructing the low-dimensional features into reconstructed features using the generator, wherein the dimensions of the reconstructed features are the same as the dimensions of the input features;
calculating the characteristic difference value of the reconstruction characteristic and the input characteristic by utilizing a characteristic discriminator, and simultaneously analyzing the conversion efficiency grade of the reconstruction characteristic to obtain an analysis efficiency grade;
Calculating a difference value of the analysis efficiency grade and the conversion efficiency grade based on the label discriminator to obtain a label difference value;
on the premise of optimizing an algorithm, calculating a characteristic difference value and a loss value of a label difference value based on the three loss functions, and iterating model parameters of a perovskite conversion analysis model through the loss values;
until the iteration times are greater than a preset iteration threshold, exiting training to obtain a trained perovskite conversion analysis model;
receiving a target efficiency conversion grade, and reversely inputting the target efficiency conversion grade into a trained perovskite conversion analysis model to obtain an analyzed target perovskite component proportion;
and preparing a target perovskite sample according to the target perovskite component proportion, and performing photoelectric conversion rate test on the target perovskite sample to obtain target photoelectric conversion rate, so as to complete perovskite component proportion analysis based on a depth generation model.
Optionally, the performing data cleaning on the perovskite raw data set to obtain a perovskite standard data set includes:
sequentially extracting each perovskite raw data from the perovskite raw data set, and performing the following operation on each perovskite raw data:
Judging the integrity of the perovskite original data, and if the perovskite original data lacks component names and proportions, directly eliminating the perovskite original data;
if the perovskite raw data are complete data, judging whether the solar cell constructed based on the perovskite raw data is a 3D organic-inorganic hybrid halide perovskite solar cell or not;
if the solar cell constructed based on the perovskite raw data is not a 3D organic-inorganic hybrid halide perovskite solar cell, directly rejecting the perovskite raw data;
if the solar cell constructed based on the perovskite raw data is a 3D organic-inorganic hybrid halide perovskite solar cell, acquiring photoelectric conversion efficiency corresponding to the 3D organic-inorganic hybrid halide perovskite solar cell constructed based on the perovskite raw data, and acquiring a test environment of the photoelectric conversion efficiency;
after eliminating perovskite original data with the photoelectric conversion efficiency of 0, judging whether the test environment is a standard test environment with the atmospheric quality AM1.5, the calibration irradiance of 1000W/m < 2 > and the calibration temperature of 25 ℃;
if the test environment is not the AM1.5, 1000W/m < 2 > and 25 ℃ standard test environment, directly eliminating the perovskite original data; if the test environment is AM1.5, 1000W/m 2 and standard test environment at 25 ℃, performing structural transformation on the perovskite original data to obtain perovskite standard data;
And converging each group of perovskite standard data and corresponding photoelectric conversion efficiency to obtain a perovskite standard data set.
Optionally, the performing structural transformation on the perovskite raw data to obtain perovskite standard data includes:
performing component separation operation on perovskite raw data to obtain eight components and ratio values, wherein the eight components are MA, FA, cs, rb, pb, sn, br and I respectively;
taking MA, FA, pb and I as perovskite main components, and taking Cs, rb, sn and Br as perovskite doping components; the method comprises the steps of,
according to a preset level interval, mapping the photoelectric conversion efficiency into a conversion efficiency level;
summarizing the ratio values and conversion efficiency grades of the perovskite main body component and the perovskite doping component to obtain the perovskite standard data.
Optionally, the conversion efficiency level comprises a high efficiency conversion efficiency, a medium efficiency conversion efficiency and a low efficiency conversion efficiency, and the high efficiency conversion efficiency is greater than or equal to 20%, the medium efficiency conversion efficiency is greater than or equal to 16% and less than 20%, and the low efficiency conversion efficiency is less than 16%.
Optionally, the KL divergence function is:
;
wherein,representing a gaussian distribution function +.>Representation generator generates the +.>Probability distribution function of individual dimensions- >Representing the +.>KL-divergence value of probability distribution function of individual dimensions relative to gaussian distribution function, +.>And->Represents the +.o of the reconstruction feature generated by the generator>Mean and standard deviation of the individual dimensions relative to the two-dimensional input feature, +.>Representing the dimensions of the reconstructed feature.
Optionally, the reconstruction loss function and the contrast loss function both employ cross entropy loss functions.
In order to solve the above problems, the present invention also provides a perovskite component proportion analysis apparatus based on a depth generation model, the apparatus comprising:
the perovskite data set acquisition module is used for acquiring a perovskite original data set, wherein the perovskite original data set is derived from a public database and a discussion set;
the data set cleaning module is used for performing data cleaning on the perovskite original data set to obtain a perovskite standard data set, wherein each perovskite standard data in the perovskite standard data set consists of a proportion value and a conversion efficiency grade of eight components, and the eight components are MA, FA, cs, rb, pb, sn, br and I;
the perovskite transformation analysis model building module is used for building a perovskite transformation analysis model to be trained, wherein the perovskite transformation analysis model consists of an encoder, a generator, a tag discriminator and a feature discriminator; the construction of the perovskite conversion analysis model to be trained comprises the following steps: determining a first activation function and a second activation function, wherein the first activation function is a LeakyReLU, and the second activation function is a ReLU; constructing and obtaining an encoder according to the sequence of the one-dimensional convolution layer, the batch normalization function and the first activation function; constructing a generator based on three one-dimensional transpose convolutions and a second activation function; constructing and obtaining the perovskite conversion analysis model according to the front of the encoder and the rear of the generator and by combining a pre-constructed tag discriminator and a feature discriminator; the tag identifier and the feature identifier respectively comprise two fully-connected layers, and share three one-dimensional convolution layers, and the layers are linked by using a LeakyReLU activation function;
The model training and application module is used for training the perovskite conversion analysis model by utilizing the perovskite standard data set until the training condition is met, and quitting the training to obtain a trained perovskite conversion analysis model; training the perovskite conversion analysis model by using the perovskite standard data set until the training condition is met, and quitting the training to obtain a trained perovskite conversion analysis model, wherein the training comprises the following steps: setting a loss function and an optimization algorithm of a perovskite conversion analysis model, wherein the loss function comprises three loss functions, namely a reconstruction loss function, a KL divergence function and an anti-loss function; constructing an input feature, wherein the input feature is composed of a first channel feature and a second channel feature, and the first channel feature comprises a perovskite host component and the second channel feature comprises a perovskite doping component; the input features are imported into a perovskite conversion analysis model, and the dimensionality of the input features is reduced based on an encoder, so that low-dimensional features are obtained; reconstructing the low-dimensional features into reconstructed features using the generator, wherein the dimensions of the reconstructed features are the same as the dimensions of the input features; calculating the characteristic difference value of the reconstruction characteristic and the input characteristic by utilizing a characteristic discriminator, and simultaneously analyzing the conversion efficiency grade of the reconstruction characteristic to obtain an analysis efficiency grade; calculating a difference value of the analysis efficiency grade and the conversion efficiency grade based on the label discriminator to obtain a label difference value; on the premise of optimizing an algorithm, calculating a characteristic difference value and a loss value of a label difference value based on the three loss functions, and iterating model parameters of a perovskite conversion analysis model through the loss values; until the iteration times are greater than a preset iteration threshold, exiting training to obtain a trained perovskite conversion analysis model; receiving a target efficiency conversion grade, and reversely inputting the target efficiency conversion grade into a trained perovskite conversion analysis model to obtain an analyzed target perovskite component proportion;
And the experiment verification module is used for preparing a target perovskite sample according to the proportion of the target perovskite components, and carrying out photoelectric conversion rate test on the target perovskite sample to obtain the target photoelectric conversion rate.
In order to solve the above-mentioned problems, the present invention also provides an electronic apparatus including:
at least one processor; the method comprises the steps of,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to implement the depth generation model-based perovskite component ratio analysis method described above.
In order to solve the above-mentioned problems, the present invention also provides a computer-readable storage medium having stored therein at least one instruction that is executed by a processor in an electronic device to implement the above-mentioned depth generation model-based perovskite component proportion analysis method.
Compared with the prior art, the method firstly acquires the perovskite original data set, wherein the perovskite original data set is derived from the public database and the discussion set, and is to be interpreted, because the perovskite original data set constructed by the method is derived from the perovskite composition components and the influence on the photoelectric conversion efficiency disclosed by the authoritative paper and the database, the accuracy of the data set is higher, secondly, the data cleaning is carried out on the perovskite original data set to obtain the perovskite standard data set, wherein each part of perovskite standard data in the perovskite standard data set is composed of eight components in proportion and conversion efficiency grade, and the eight components are MA, FA, cs, rb, pb, sn, br and I, and because the eight components are core components composing the perovskite, the eight components are extracted from the perovskite original data set, thereby effectively constituting the data set for training a model, secondly, the perovskite conversion analysis model to be trained is constructed, wherein the perovskite conversion analysis model consists of a coder, a generator, a tag discriminator and a characteristic analyzer, and the perovskite conversion analysis model, and the method overcomes the problem that the number of the perovskite composed of the eight components is not satisfied by the perovskite is designed by the training model, the depth of the perovskite conversion model, the perovskite conversion model can be trained based on the training data of the training data set, the actual conversion condition is realized, the time-consuming time is reached, the training condition is reached, the training is realized by the training of the training model analysis model, and the actual conversion model is realized by the training of the perovskite conversion model, and the perovskite conversion model is realized by the training of the training model based on the training of the training mode, and the training efficiency model has been realized, therefore, the perovskite component proportion analysis method, the perovskite component proportion analysis device, the electronic equipment and the computer readable storage medium based on the depth generation model mainly aim to solve the problem that the perovskite component design at present only depends on the existing experience, and the perovskite with high photoelectric conversion efficiency needs to consume higher experimental trial-and-error cost.
Drawings
FIG. 1 is a schematic flow chart of a perovskite component ratio analysis method based on a depth generation model according to an embodiment of the invention;
FIG. 2 is a block diagram of a perovskite component ratio analysis model according to a depth generation model according to one embodiment of the invention;
FIG. 3 is a JV curve of perovskite MA0.09FA0.9Pb (I0.9Br0.1) 3 with respect to high efficiency photoelectric conversion levels in a depth generation model based perovskite composition ratio analysis method as defined by one embodiment of the invention;
FIG. 4 is a graph of the JV of perovskite MA0.52FA0.46PbI3 for the intermediate photoelectric conversion grade in the depth generation model based perovskite composition ratio analysis method as defined by one embodiment of the invention;
FIG. 5 is a training chart of perovskite component ratio analysis model based on depth generation model according to one embodiment of the invention;
FIG. 6 is a functional block diagram of a perovskite component ratio analysis device based on a depth generation model according to one embodiment of the invention;
fig. 7 is a schematic structural diagram of an electronic device for implementing the perovskite component proportion analysis method based on the depth generation model according to an embodiment of the invention.
The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
The embodiment of the application provides a perovskite component proportion analysis method based on a depth generation model. The execution subject of the perovskite component proportion analysis method based on the depth generation model comprises, but is not limited to, at least one of a server, a terminal and the like which can be configured to execute the method provided by the embodiment of the application. In other words, the perovskite component proportion analysis method based on the depth generation model may be performed by software or hardware installed in a terminal device or a server device. The service end includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like.
Example 1: referring to fig. 1, a flow chart of a perovskite component proportion analysis method based on a depth generation model according to an embodiment of the application is shown. In this embodiment, the perovskite component proportion analysis method based on the depth generation model includes:
S1, acquiring a perovskite original data set, wherein the perovskite original data set is derived from a public database and a discussion set.
It is to be construed that the perovskite raw dataset sources include, but are not limited to, public databases and discourse. The public database comprises CS, elsevier, wiley, RSC and the like, and the discussion comprises sample data used by 880 perovskite solar cell related research articles from 2013 to 2020. For example, as described in a published paper in 2020, if the perovskite is composed of the following components in proportions: the perovskite solar cell has the advantages that { 'Cs' (cesium) } '0.05,' MA '(methylamine) }, 0.095,' FA '(formamidine) (' 855), 'Rb' (rubidium), 'Pb' (lead) ('1),' Sn '(tin) (' I '(iodine) (' 0.9), 'Br' (bromine) ('0.1),' Cl '(chlorine) (' 0.0) }, and the photoelectric conversion efficiency of the perovskite solar cell formed by using the ratio can reach 22.7%, so that the ratio data and the photoelectric conversion efficiency can be extracted from the paper published in 2020, and thus the perovskite raw data can be constructed and obtained.
It can be determined that the perovskite raw data consists of constituent components, the duty ratio of the constituent components and the photoelectric conversion efficiency, and 2991 groups of perovskite raw data are obtained through public databases and discussion in one embodiment of the invention.
S2, performing data cleaning on the perovskite original data set to obtain a perovskite standard data set, wherein each perovskite standard data in the perovskite standard data set is composed of eight component proportion values and conversion efficiency grades of [ ' MA ', ' FA ', ' Cs ', ' Rb ', ' Pb ', ' Sn ', ' Br and ' I ' ].
It should be understood that, because different papers focus on the composition of perovskite, there may be a phenomenon that the composition of the perovskite raw data is different from one perovskite raw data set to another. Illustratively, for example, the perovskite raw data a has 10 components and the perovskite raw data B has 11 components, and compared with the perovskite raw data a, the perovskite raw data B further includes a rubidium component, so in order to ensure that the component composition of each group of perovskite raw data in the perovskite raw data set is the same, the embodiment of the present invention further needs to perform a data cleaning operation on the perovskite raw data set.
In detail, the performing data cleaning on the perovskite raw data set to obtain a perovskite standard data set comprises:
sequentially extracting each perovskite raw data from the perovskite raw data set, and performing the following operation on each perovskite raw data:
Judging the integrity of the perovskite original data, and if the perovskite original data lacks component names and proportions, directly eliminating the perovskite original data;
if the perovskite raw data are complete data, judging whether the solar cell constructed based on the perovskite raw data is a 3D organic-inorganic hybrid halide perovskite solar cell or not;
if the solar cell constructed based on the perovskite raw data is not a 3D organic-inorganic hybrid halide perovskite solar cell, directly rejecting the perovskite raw data;
if the solar cell constructed based on the perovskite raw data is a 3D organic-inorganic hybrid halide perovskite solar cell, acquiring photoelectric conversion efficiency corresponding to the 3D organic-inorganic hybrid halide perovskite solar cell constructed based on the perovskite raw data, and acquiring a test environment of the photoelectric conversion efficiency;
after eliminating perovskite original data with the photoelectric conversion efficiency of 0, judging whether the test environment is a standard test environment with the atmospheric quality AM1.5, the calibration irradiance of 1000W/m < 2 > and the calibration temperature of 25 ℃;
if the test environment is not the AM1.5, 1000W/m < 2 > and 25 ℃ standard test environment, directly eliminating the perovskite original data; if the test environment is AM1.5, 1000W/m 2 and standard test environment at 25 ℃, performing structural transformation on the perovskite original data to obtain perovskite standard data;
And converging each group of perovskite standard data and corresponding photoelectric conversion efficiency to obtain a perovskite standard data set.
It can be understood that even though the components and proportions of the perovskite are the same, the corresponding photoelectric conversion efficiency of the solar cells of different structures or types are different, and the main purpose of the embodiments of the present invention is to explore the effect of the components of different proportions on the photoelectric conversion efficiency of the perovskite, so that the embodiments of the present invention eliminate the perovskite raw data of the non-3D organic-inorganic hybrid halide perovskite solar cell for the purpose of controlling the variables. Illustratively, perovskite raw data a, perovskite raw data B, and perovskite raw data C are present in the perovskite raw data set, but perovskite raw data C was found to be used to construct 2D perovskite solar cells, not 3D perovskite solar cells, so perovskite raw data C could be directly culled.
Furthermore, different test environments also affect the obtained photoelectric conversion efficiency, so in order to unify the test environments, the embodiment of the invention sets the photoelectric conversion efficiency obtained by testing each group of perovskite original data, and the test environments are AM1.5, 1000W/m 2 and 25 ℃. It should be explained that the standard solar spectrum: refers to the spectrum of solar radiation reaching the ground after passing through the atmosphere when emitted outside the earth's atmosphere. The AM1.5 standard is generally used to represent the spectrum at an atmospheric mass of 1.5, which is the benchmark for solar cell testing. Nominal irradiance of 1000W/m 2 means that at AM1.5 solar spectrum, the average irradiance of solar radiation on a surface of 1 square meter area perpendicular to the ground is 1000 watts per square meter. Calibration temperature 25℃: it means that the ambient temperature should be kept at 25 c in order to compare the test results under different ambient conditions when testing the solar cell. That is, when the photoelectric conversion efficiency test of the solar cell is performed, irradiance of 1000W/m 2 under the standard solar spectrum is required to be used, and the ambient temperature is kept at 25 ℃ so as to ensure comparability of test results.
In addition, the performing structural transformation on the perovskite raw data to obtain perovskite standard data comprises the following steps:
performing component separation operation on perovskite raw data to obtain eight components and ratio values, wherein the eight components are MA, FA, cs, rb, pb, sn, br and I respectively;
taking MA, FA, pb and I as perovskite main components, and taking Cs, rb, sn and Br as perovskite doping components; the method comprises the steps of,
according to a preset level interval, mapping the photoelectric conversion efficiency into a conversion efficiency level;
summarizing the ratio values and conversion efficiency grades of the perovskite main body component and the perovskite doping component to obtain the perovskite standard data.
Importantly, the component separation operation needs to be adaptively operated according to the structure of each part of perovskite raw data, for example, in perovskite raw data A, pbI2 of 735.3 mg, FAI of 224.4 mg, MABr of 16.2 mg and CsI of 19.8 mg are recorded to be configured into perovskite solutions, so that the molar percentages obtained by dividing the perovskite solutions by molar masses (PbI 2:461g/mol, FAI:171.97g/mol, MABr:111.96g/mol and CsI:259.8 g/mol) are PbI2: FAI: MABr: csI is 1.595 mmol:1.305 mmol:0.145 mmol:0.076 The composition and the proportion of the perovskite are Cs0.05, FA0.85, MA0.10, pb1.00, I0.97 and Br0.03.
It should be emphasized that if MACl or PbCl2 is contained in the perovskite raw data, it is known in the art that annealing operation is required to be performed in the construction of solar cells, and Cl is not generally contained in the perovskite composition because Cl is largely volatilized during the shipment process and Cl residues do not occur in the perovskite.
It should be explained that the preset class interval in the embodiment of the present invention is 3 classes, that is, the conversion efficiency class includes high efficiency, medium efficiency and low efficiency. Wherein the photoelectric conversion efficiency of the high-efficiency grade is more than or equal to 20%, the medium-efficiency conversion efficiency is more than or equal to 16% and less than 20%, and the low-efficiency photoelectric conversion efficiency is less than 16%.
The embodiment of the invention also comprises operations of removing blank spaces, special symbols, photoelectric conversion efficiency correction and the like in the perovskite original data when constructing the perovskite standard data set, and the operations are not repeated here.
S3, constructing a perovskite conversion analysis model to be trained, wherein the perovskite conversion analysis model consists of an encoder, a generator, a tag discriminator and a feature discriminator.
In detail, the construction of the perovskite transformation analysis model to be trained comprises the following steps:
determining a first activation function and a second activation function, wherein the first activation function is a LeakyReLU, and the second activation function is a ReLU;
Constructing and obtaining an encoder according to the sequence of the one-dimensional convolution layer, the batch normalization function and the first activation function;
constructing a generator based on three one-dimensional transpose convolutions and a second activation function;
the perovskite transformation analysis model is constructed by combining a pre-constructed tag discriminator and a feature discriminator according to the front of the encoder and the rear of the generator.
Referring to fig. 2, there is shown a perovskite conversion analysis model according to an embodiment of the present invention, which is composed of four modules, namely, an encoder, a generator, a tag discriminator, and a feature discriminator. And it can be understood that the perovskite conversion analysis model is constructed based on a deep learning model, so that before the perovskite conversion analysis model is used, the perovskite conversion analysis model needs to be trained, and the perovskite data of unknown conversion efficiency level can not be analyzed by using the trained perovskite conversion analysis model until the trained perovskite conversion analysis model is obtained after the training condition is met.
It should be explained that the tag discriminator and the feature discriminator each comprise two fully connected layers and share three one-dimensional convolution layers, and the layers are linked by using a LeakyReLU activation function. Therefore, after the perovskite conversion analysis model is built, the perovskite conversion analysis model needs to be trained in the next step.
And S4, training a perovskite conversion analysis model by using the perovskite standard data set until the training condition is met, and exiting the training to obtain a trained perovskite conversion analysis model.
In detail, training the perovskite conversion analysis model by using the perovskite standard data set until the training condition is met, and quitting the training to obtain a trained perovskite conversion analysis model, wherein the training comprises the following steps:
setting a loss function and an optimization algorithm of a perovskite conversion analysis model, wherein the loss function comprises three loss functions, namely a reconstruction loss function, a KL divergence function and an anti-loss function;
constructing an input feature, wherein the input feature is composed of a first channel feature and a second channel feature, and the first channel feature comprises a perovskite host component and the second channel feature comprises a perovskite doping component;
the input features are imported into a perovskite conversion analysis model, and the dimensionality of the input features is reduced based on an encoder, so that low-dimensional features are obtained;
reconstructing the low-dimensional features into reconstructed features using the generator, wherein the dimensions of the reconstructed features are the same as the dimensions of the input features;
calculating the characteristic difference value of the reconstruction characteristic and the input characteristic by utilizing a characteristic discriminator, and simultaneously analyzing the conversion efficiency grade of the reconstruction characteristic to obtain an analysis efficiency grade;
Calculating a difference value of the analysis efficiency grade and the conversion efficiency grade based on the label discriminator to obtain a label difference value;
on the premise of optimizing an algorithm, calculating a characteristic difference value and a loss value of a label difference value based on the three loss functions, and iterating model parameters of a perovskite conversion analysis model through the loss values;
and (3) until the iteration times are greater than a preset iteration threshold, exiting training to obtain a trained perovskite conversion analysis model.
It is explained in turn that the embodiment of the present invention uses a cross entropy function as a reconstruction loss function, which is mainly aimed at measuring the input features and the error between the reconstructed features obtained by the encoder generator according to the input features.
The main purpose of the KL divergence function is to restrict the error of the input feature converted into the reconstructed feature, so that the error distribution accords with the normal distribution. In the perovskite conversion analysis model, the embodiment of the invention hopes that the reconstruction feature can better have a corresponding relation with the input feature, namely in order to achieve the aim, the KL divergence function is used as a regular term (regularization term) to restrict the variance and standard deviation of the reconstruction feature and the input feature to conform to Gaussian distribution, and the purpose of adding the KL divergence function can enable the reconstruction feature generated by the generator to have controllable generation change. In the embodiment of the present invention, the KL divergence function may be:
;
Wherein,representing a gaussian distribution function +.>Representation generator generates the +.>Probability distribution function of individual dimensions->Representing the +.>KL-divergence value of probability distribution function of individual dimensions relative to gaussian distribution function, +.>And->Represents the +.o of the reconstruction feature generated by the generator>Mean and standard deviation of the individual dimensions relative to the two-dimensional input feature, +.>Representing the dimensions of the reconstructed feature.
Furthermore, the challenge loss function is a loss function constructed for training the feature discriminator. The basic idea is to convert the judgment result of the feature discriminator into a 0-1 label, set the label of the input feature as 1, and set the label of the reconstructed feature as 0. Then, the cross entropy loss (cross-entropy loss) is used to measure the analysis error of the feature discriminator. The contrast loss function can cause the feature discriminator to better resolve the input features and the reconstructed features, thereby helping the generator produce more realistic reconstructed features.
In other words, the main purpose of the training is to improve the analysis capability of the perovskite transformation analysis model, in each training iteration, the tag discriminator and the feature discriminator are first trained so that the feature discriminator can accurately distinguish between the input data and the reconstructed data, the target discriminator classifies the data according to a given transformation efficiency level, then the reconstructed data is generated according to the generator, and the given transformation efficiency level is trained so that the encoder and the generator can generate perovskite data (i.e. the reconstructed data) meeting the conditions, and it is seen that the analysis model capable of generating the perovskite sample meeting the given transformation efficiency level is finally obtained through continuous iterative optimization.
In addition, it should be explained that, in the embodiment of the present invention, the input features are obtained by constructing the main component and the doping component, and the input features of the main component and the doping component are used to implement two-channel input, so that the model training effect can be improved while the data itself is not changed, where the improved aspects include: the variance of single-channel input is reduced, namely when the traditional method directly inputs the features in a single-channel mode, the features can only change along one dimension, and the problems of over fitting or under fitting and the like of the model are easy to cause. After splitting the data, different types of features can be respectively input into the two channels, so that the variance of the features on a single channel can be reduced, and the robustness of the model can be improved; in addition, the tolerance to noise can be increased, because some features have slight noise, and after the features are split into a main component and a doped component, the model can learn richer feature representations through a plurality of channels, so that the tolerance of the model to noise can be improved. Finally, the influence of special features is reduced: the method has the advantages that the data are split into the main body and the doped components, the dimension of the feature space is increased, the feature learning capacity of the model is improved, meanwhile, the influence of certain special features in the data set on the performance of the model is reduced, the probability of misjudgment of the model is further reduced, and the model is more stable.
In the embodiment of the invention, adam is selected as an optimization algorithm, the learning rate of an Adam optimizer is set to be 0.001, and in addition, the iteration number is set to be 160 to perform training, and the training diagram of the perovskite conversion analysis model is shown in fig. 5, wherein the horizontal axis represents the iteration number and the vertical axis represents the loss value.
In summary, the training process of the perovskite conversion analysis model may be shown by an example, it is assumed that a set of perovskite standard data M is included in the perovskite standard data set, and according to the foregoing, the perovskite standard data M is composed of the perovskite main component, the ratio value of the perovskite doping component, and the conversion efficiency level, so that the ratio value of the perovskite main component and the perovskite doping component is input to the encoder, and the encoder is used to reduce the dimensions of the ratio values of the perovskite main component and the perovskite doping component, and then the generator is used to increase the dimensions again, so as to obtain the reconstruction feature. In order to prevent the difference between the reconstruction feature and the input feature from being too high, the encoder and the generator are constrained by the feature discriminator, and the conversion efficiency grade corresponding to the reconstruction feature is analyzed at the same time, and the real efficiency grade and the analysis efficiency grade are not provided with errors if the conversion efficiency grade of the perovskite standard data M is high-efficiency and the analyzed conversion efficiency grade is also high-efficiency, but along with continuous input of the perovskite standard data, the perovskite conversion analysis model always has errors in analysis of the conversion efficiency grade, so that the embodiment of the invention constructs a loss function comprising three loss functions, and the error values of the reconstruction feature, the input feature, the real efficiency grade and the analysis efficiency grade are respectively constrained, thereby ensuring that the perovskite conversion analysis model has stronger application value.
S5, receiving a target efficiency conversion grade, and reversely inputting the target efficiency conversion grade into a trained perovskite conversion analysis model to obtain the analyzed target perovskite component proportion.
It is understood that the target efficiency conversion level refers to a specific photoelectric conversion efficiency level that needs to be obtained, for example: high efficiency. According to the description of S4, after the training condition is finished and the training is exited, a perovskite conversion analysis model meeting the subsequent requirements can be obtained, so that after receiving the required target efficiency conversion grade, the perovskite conversion analysis model can be directly used for reversely analyzing the component proportions in the perovskite, namely, by utilizing the reverse analysis technology of the perovskite conversion analysis model, the first channel characteristic and the second channel characteristic are obtained by inputting the target efficiency conversion grade into a tag discriminator of the perovskite conversion analysis model, wherein the first channel characteristic comprises the perovskite main body component composition proportion, and the second channel characteristic comprises the perovskite doping component composition proportion.
Further, when the target efficiency conversion grade is high-efficiency, the generation quantity is set, a perovskite conversion analysis model is adopted to generate potential materials, 8 characteristics of 'MA', 'FA', 'Cs', 'Rb', 'Pb', 'Sn', 'Br', 'I' data generated by generation are checked, the purpose of checking component proportions at different positions is achieved, when the component proportions in the perovskite are obtained, whether the sum of the component proportions is 1 needs to be judged, and if the sum is not 1, the proportions need to be recalculated according to the generation result. Finally, the obtained characteristics are put into a forward model ML for analysis of photoelectric conversion efficiency, so that the analysis photoelectric conversion efficiency of the generated perovskite material is obtained, and the proportion of components in the perovskite material is obtained through further screening. After the proportion of components in the required perovskite material is generated, the perovskite is combined with the ETL layer, the HTL layer, the solvent, the antisolvent and other devices in different modes and the technological treatment method, and the perovskite battery device with high photoelectric conversion efficiency is more efficiently and effectively screened by combining a forward model through a screening method.
Further, referring to fig. 3 and 4, to demonstrate the effectiveness of the perovskite conversion analysis model, two sets of data points representing medium and high efficiency were selected in the resulting material, wherein the perovskite compositions were ma0.52fa0.46pbi3 and ma0.09fa0.9pb (i 0.9br0.1) 3, respectively, and each set of validation experiments was repeated twice. A perovskite layer was prepared using a two-step deposition method during the validation process, using DMSO (dimethyl sulfoxide) +dmf (dimethylformamide) and IPA (isopropyl alcohol) as organic solvents for dissolving PbI2 and organic solutions, respectively, and annealing at 150 degrees celsius for 30min. For measuring PCE (photoelectric conversion efficiency), ITO (indium tin oxide semiconductor transparent conductive film)/SnO 2 (tin dioxide)/perovskite layer/spira-ome tad (2,,7,/>-tetrakis [ N, N-bis (4-methoxyphenyl) amino group]-9,/>-spirobifluorene)/Au (gold) battery device structure. The highest reverse sweeping efficiency measured under the standard sunlight is 19.13 and 21.8, which respectively meet the medium efficiency grade 16-20 and the high efficiency grade>20, the model proved to be reliable. Thus setting the PCE generation conditions when screening high PCE materialsFor high efficiency, then the PCE may be found >20, and further the perovskite composition of the potential high PCE was discovered through experimental verification.
S6, preparing a target perovskite sample according to the target perovskite component proportion, and performing photoelectric conversion rate test on the target perovskite sample to obtain target photoelectric conversion rate, so as to complete perovskite component proportion analysis based on a depth generation model.
Further, after the target perovskite component proportion is obtained, a target perovskite sample with a corresponding target efficiency conversion grade can be prepared according to the target perovskite component proportion, the actual photoelectric conversion rate of the target perovskite sample is tested, the actual photoelectric conversion rate is compared with the target efficiency conversion grade, and when the actual photoelectric conversion rate belongs to the target efficiency conversion grade, the target perovskite sample with the target perovskite component proportion can reach the target efficiency conversion grade. And the effect verification of perovskite component proportion analysis based on the depth generation model is realized.
Compared with the prior art, the method firstly acquires the perovskite original data set, wherein the perovskite original data set is derived from the public database and the discussion set, and is to be interpreted, because the perovskite original data set constructed by the method is derived from the perovskite composition components and the influence on the photoelectric conversion efficiency disclosed by the authoritative paper and the database, the accuracy of the data set is higher, secondly, the data cleaning is carried out on the perovskite original data set to obtain the perovskite standard data set, wherein each part of perovskite standard data in the perovskite standard data set is composed of eight components in proportion and conversion efficiency grade, and the eight components are MA, FA, cs, rb, pb, sn, br and I, and because the eight components are core components composing the perovskite, the eight components are extracted from the perovskite original data set, thereby effectively constituting the data set for training a model, secondly, the perovskite conversion analysis model to be trained is constructed, wherein the perovskite conversion analysis model consists of a coder, a generator, a tag discriminator and a characteristic analyzer, and the perovskite conversion analysis model, and the method overcomes the problem that the number of the perovskite composed of the eight components is not satisfied by the perovskite is designed by the training model, the depth of the perovskite conversion model, the perovskite conversion model can be trained based on the training data of the training data set, the actual conversion condition is realized, the time-consuming time is reached, the training condition is reached, the training is realized by the training of the training model analysis model, and the actual conversion model is realized by the training of the perovskite conversion model, and the perovskite conversion model is realized by the training of the training model based on the training of the training mode, and the training efficiency model has been realized, therefore, the perovskite component proportion analysis method, the perovskite component proportion analysis device, the electronic equipment and the computer readable storage medium based on the depth generation model mainly aim to solve the problem that the perovskite component design at present only depends on the existing experience, and the perovskite with high photoelectric conversion efficiency needs to consume higher experimental trial-and-error cost.
Example 2: fig. 6 is a functional block diagram of a perovskite component ratio analysis device based on a depth generation model according to an embodiment of the invention.
The perovskite component proportion analysis apparatus 100 based on the depth generation model according to the present invention may be mounted in an electronic device. Depending on the functions implemented, the depth generation model-based perovskite component ratio analysis apparatus 100 may include a perovskite data set acquisition module 101, a data set cleaning module 102, a transformation analysis model construction module 103, a model training and application module 104, and an experiment verification module 105. The module of the invention, which may also be referred to as a unit, refers to a series of computer program segments, which are stored in the memory of the electronic device, capable of being executed by the processor of the electronic device and of performing a fixed function.
The perovskite data set acquisition module 101 is configured to acquire a perovskite original data set, where the perovskite original data set is derived from a public database and a discussion set;
the data set cleaning module 102 is configured to perform data cleaning on a perovskite original data set to obtain a perovskite standard data set, where each perovskite standard data set includes a proportion value and a conversion efficiency grade of eight components, and the eight components are MA, FA, cs, rb, pb, sn, br and I;
The transformation analysis model construction module 103 is configured to construct a perovskite transformation analysis model to be trained, where the perovskite transformation analysis model is composed of an encoder, a generator, a tag discriminator, and a feature discriminator; the construction of the perovskite conversion analysis model to be trained comprises the following steps: determining a first activation function and a second activation function, wherein the first activation function is a LeakyReLU, and the second activation function is a ReLU; constructing and obtaining an encoder according to the sequence of the one-dimensional convolution layer, the batch normalization function and the first activation function; constructing a generator based on three one-dimensional transpose convolutions and a second activation function; constructing and obtaining the perovskite conversion analysis model according to the front of the encoder and the rear of the generator and by combining a pre-constructed tag discriminator and a feature discriminator; the tag identifier and the feature identifier respectively comprise two fully-connected layers, and share three one-dimensional convolution layers, and the layers are linked by using a LeakyReLU activation function;
the model training and applying module 104 is configured to train the perovskite conversion analysis model by using the perovskite standard data set until the training condition is met, and quit training, so as to obtain a trained perovskite conversion analysis model; training the perovskite conversion analysis model by using the perovskite standard data set until the training condition is met, and quitting the training to obtain a trained perovskite conversion analysis model, wherein the training comprises the following steps: setting a loss function and an optimization algorithm of a perovskite conversion analysis model, wherein the loss function comprises three loss functions, namely a reconstruction loss function, a KL divergence function and an anti-loss function; constructing an input feature, wherein the input feature is composed of a first channel feature and a second channel feature, and the first channel feature comprises a perovskite host component and the second channel feature comprises a perovskite doping component; the input features are imported into a perovskite conversion analysis model, and the dimensionality of the input features is reduced based on an encoder, so that low-dimensional features are obtained; reconstructing the low-dimensional features into reconstructed features using the generator, wherein the dimensions of the reconstructed features are the same as the dimensions of the input features; calculating the characteristic difference value of the reconstruction characteristic and the input characteristic by utilizing a characteristic discriminator, and simultaneously analyzing the conversion efficiency grade of the reconstruction characteristic to obtain an analysis efficiency grade; calculating a difference value of the analysis efficiency grade and the conversion efficiency grade based on the label discriminator to obtain a label difference value; on the premise of optimizing an algorithm, calculating a characteristic difference value and a loss value of a label difference value based on the three loss functions, and iterating model parameters of a perovskite conversion analysis model through the loss values; until the iteration times are greater than a preset iteration threshold, exiting training to obtain a trained perovskite conversion analysis model; receiving a target efficiency conversion grade, and reversely inputting the target efficiency conversion grade into a trained perovskite conversion analysis model to obtain an analyzed target perovskite component proportion;
The experiment verification module 105 is configured to prepare a target perovskite sample according to the target perovskite component ratio, and perform a photoelectric conversion rate test on the target perovskite sample to obtain a target photoelectric conversion rate.
In detail, the modules in the perovskite component proportion analysis device 100 based on the depth generation model in the embodiment of the present invention use the same technical means as the perovskite component proportion analysis method based on the depth generation model described in fig. 1, and can produce the same technical effects, which are not repeated here.
Example 3: fig. 7 is a schematic structural diagram of an electronic device for implementing a perovskite component proportion analysis method based on a depth generation model according to an embodiment of the invention.
The electronic device 1 may comprise a processor 10, a memory 11, a bus 12 and a communication interface 13, and may further comprise a computer program stored in the memory 11 and executable on the processor 10, such as a perovskite component proportion analysis program based on a depth generation model.
The memory 11 includes at least one type of readable storage medium, including flash memory, a mobile hard disk, a multimedia card, a card memory (e.g., SD or DX memory, etc.), a magnetic memory, a magnetic disk, an optical disk, etc. The memory 11 may in some embodiments be an internal storage unit of the electronic device 1, such as a removable hard disk of the electronic device 1. The memory 11 may in other embodiments also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart memory card (SmartMediaCard, SMC), a secure digital (SecureDigital, SD) card, a flash card (FlashCard) or the like, provided on the electronic device 1. Further, the memory 11 may also include both an internal storage unit and an external storage device of the electronic device 1. The memory 11 may be used not only for storing application software installed in the electronic device 1 and various types of data, such as codes of perovskite component proportion analysis programs based on depth generation models, but also for temporarily storing data that has been output or is to be output.
The processor 10 may be comprised of integrated circuits in some embodiments, for example, a single packaged integrated circuit, or may be comprised of multiple integrated circuits packaged with the same or different functions, including one or more central processing units (CentralProcessingunit, CPU), microprocessors, digital processing chips, graphics processors, a combination of various control chips, and the like. The processor 10 is a control unit (control unit) of the electronic device, connects the respective components of the entire electronic device using various interfaces and lines, executes or executes programs or modules (e.g., perovskite component ratio analysis programs based on a depth generation model, etc.) stored in the memory 11, and invokes data stored in the memory 11 to perform various functions of the electronic device 1 and process the data.
The bus may be an Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The bus may be classified as an address bus, a data bus, a control bus, etc. The bus is arranged to enable a connection communication between the memory 11 and at least one processor 10 etc.
Fig. 7 shows only an electronic device with components, it being understood by a person skilled in the art that the structure shown in fig. 7 does not constitute a limitation of the electronic device 1, and may comprise fewer or more components than shown, or may combine certain components, or may be arranged in different components.
For example, although not shown, the electronic device 1 may further include a power source (such as a battery) for supplying power to each component, and preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that functions of charge management, discharge management, power consumption management, and the like are implemented through the power management device. The power supply may also include one or more of any of a direct current or alternating current power supply, recharging device, power failure detection circuit, power converter or inverter, power status indicator, etc. The electronic device 1 may further include various sensors, bluetooth modules, wi-Fi modules, etc., which will not be described herein.
Further, the electronic device 1 may also comprise a network interface, optionally the network interface may comprise a wired interface and/or a wireless interface (e.g. WI-FI interface, bluetooth interface, etc.), typically used for establishing a communication connection between the electronic device 1 and other electronic devices.
The electronic device 1 may optionally further comprise a user interface, which may be a Display, an input unit, such as a Keyboard (Keyboard), or a standard wired interface, a wireless interface. Alternatively, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (organic light-emitting diode) touch, or the like. The display may also be referred to as a display screen or display unit, as appropriate, for displaying information processed in the electronic device 1 and for displaying a visual user interface.
It should be understood that the embodiments described are for illustrative purposes only and are not limited to this configuration in the scope of the patent application.
The perovskite component proportion analysis program based on the depth generation model stored in the memory 11 in the electronic device 1 is a combination of a plurality of instructions, which when executed in the processor 10, can realize:
acquiring a perovskite original data set, wherein the perovskite original data set is derived from a public database and a discussion set;
performing data cleaning on the perovskite original data set to obtain a perovskite standard data set, wherein each perovskite standard data in the perovskite standard data set consists of proportion values and conversion efficiency grades of eight components, and the eight components are MA, FA, cs, rb, pb, sn, br and I;
Constructing a perovskite conversion analysis model to be trained, wherein the perovskite conversion analysis model consists of an encoder, a generator, a tag discriminator and a feature discriminator;
the construction of the perovskite conversion analysis model to be trained comprises the following steps:
determining a first activation function and a second activation function, wherein the first activation function is a LeakyReLU, and the second activation function is a ReLU;
constructing and obtaining an encoder according to the sequence of the one-dimensional convolution layer, the batch normalization function and the first activation function;
constructing a generator based on three one-dimensional transpose convolutions and a second activation function;
constructing and obtaining the perovskite conversion analysis model according to the front of the encoder and the rear of the generator and by combining a pre-constructed tag discriminator and a feature discriminator; the tag identifier and the feature identifier respectively comprise two fully-connected layers, and share three one-dimensional convolution layers, and the layers are linked by using a LeakyReLU activation function;
training a perovskite conversion analysis model by using the perovskite standard data set until the training condition is met and the training is exited, so as to obtain a trained perovskite conversion analysis model;
training the perovskite conversion analysis model by using the perovskite standard data set until the training condition is met, and quitting the training to obtain a trained perovskite conversion analysis model, wherein the training comprises the following steps:
Setting a loss function and an optimization algorithm of a perovskite conversion analysis model, wherein the loss function comprises three loss functions, namely a reconstruction loss function, a KL divergence function and an anti-loss function;
constructing an input feature, wherein the input feature is composed of a first channel feature and a second channel feature, and the first channel feature comprises a perovskite host component and the second channel feature comprises a perovskite doping component;
the input features are imported into a perovskite conversion analysis model, and the dimensionality of the input features is reduced based on an encoder, so that low-dimensional features are obtained;
reconstructing the low-dimensional features into reconstructed features using the generator, wherein the dimensions of the reconstructed features are the same as the dimensions of the input features;
calculating the characteristic difference value of the reconstruction characteristic and the input characteristic by utilizing a characteristic discriminator, and simultaneously analyzing the conversion efficiency grade of the reconstruction characteristic to obtain an analysis efficiency grade;
calculating a difference value of the analysis efficiency grade and the conversion efficiency grade based on the label discriminator to obtain a label difference value;
on the premise of optimizing an algorithm, calculating a characteristic difference value and a loss value of a label difference value based on the three loss functions, and iterating model parameters of a perovskite conversion analysis model through the loss values;
Until the iteration times are greater than a preset iteration threshold, exiting training to obtain a trained perovskite conversion analysis model;
receiving a target efficiency conversion grade, and reversely inputting the target efficiency conversion grade into a trained perovskite conversion analysis model to obtain an analyzed target perovskite component proportion;
and preparing a target perovskite sample according to the target perovskite component proportion, and performing photoelectric conversion rate test on the target perovskite sample to obtain target photoelectric conversion rate, so as to complete perovskite component proportion analysis based on a depth generation model.
Specifically, the specific implementation method of the above instruction by the processor 10 may refer to descriptions of related steps in the corresponding embodiments of fig. 1 to 6, which are not repeated herein.
Further, the modules/units integrated in the electronic device 1 may be stored in a computer readable storage medium if implemented in the form of software functional units and sold or used as separate products. The computer readable storage medium may be volatile or nonvolatile. For example, the computer readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer memory, a Read-only memory (ROM).
The present invention also provides a computer readable storage medium storing a computer program which, when executed by a processor of an electronic device, can implement:
acquiring a perovskite original data set, wherein the perovskite original data set is derived from a public database and a discussion set;
performing data cleaning on the perovskite original data set to obtain a perovskite standard data set, wherein each perovskite standard data in the perovskite standard data set consists of proportion values and conversion efficiency grades of eight components, and the eight components are MA, FA, cs, rb, pb, sn, br and I;
constructing a perovskite conversion analysis model to be trained, wherein the perovskite conversion analysis model consists of an encoder, a generator, a tag discriminator and a feature discriminator;
the construction of the perovskite conversion analysis model to be trained comprises the following steps:
determining a first activation function and a second activation function, wherein the first activation function is a LeakyReLU, and the second activation function is a ReLU;
constructing and obtaining an encoder according to the sequence of the one-dimensional convolution layer, the batch normalization function and the first activation function;
constructing a generator based on three one-dimensional transpose convolutions and a second activation function;
Constructing and obtaining the perovskite conversion analysis model according to the front of the encoder and the rear of the generator and by combining a pre-constructed tag discriminator and a feature discriminator; the tag identifier and the feature identifier respectively comprise two fully-connected layers, and share three one-dimensional convolution layers, and the layers are linked by using a LeakyReLU activation function;
training a perovskite conversion analysis model by using the perovskite standard data set until the training condition is met and the training is exited, so as to obtain a trained perovskite conversion analysis model;
training the perovskite conversion analysis model by using the perovskite standard data set until the training condition is met, and quitting the training to obtain a trained perovskite conversion analysis model, wherein the training comprises the following steps:
setting a loss function and an optimization algorithm of a perovskite conversion analysis model, wherein the loss function comprises three loss functions, namely a reconstruction loss function, a KL divergence function and an anti-loss function;
constructing an input feature, wherein the input feature is composed of a first channel feature and a second channel feature, and the first channel feature comprises a perovskite host component and the second channel feature comprises a perovskite doping component;
the input features are imported into a perovskite conversion analysis model, and the dimensionality of the input features is reduced based on an encoder, so that low-dimensional features are obtained;
Reconstructing the low-dimensional features into reconstructed features using the generator, wherein the dimensions of the reconstructed features are the same as the dimensions of the input features;
calculating the characteristic difference value of the reconstruction characteristic and the input characteristic by utilizing a characteristic discriminator, and simultaneously analyzing the conversion efficiency grade of the reconstruction characteristic to obtain an analysis efficiency grade;
calculating a difference value of the analysis efficiency grade and the conversion efficiency grade based on the label discriminator to obtain a label difference value;
on the premise of optimizing an algorithm, calculating a characteristic difference value and a loss value of a label difference value based on the three loss functions, and iterating model parameters of a perovskite conversion analysis model through the loss values;
until the iteration times are greater than a preset iteration threshold, exiting training to obtain a trained perovskite conversion analysis model;
receiving a target efficiency conversion grade, and reversely inputting the target efficiency conversion grade into a trained perovskite conversion analysis model to obtain an analyzed target perovskite component proportion;
and preparing a target perovskite sample according to the target perovskite component proportion, and performing photoelectric conversion rate test on the target perovskite sample to obtain target photoelectric conversion rate, so as to complete perovskite component proportion analysis based on a depth generation model.
The modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical units, may be located in one place, or may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional module in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units can be realized in a form of hardware or a form of hardware and a form of software functional modules.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof.
Finally, it should be noted that the above-mentioned embodiments are merely for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made to the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention.
Claims (7)
1. A perovskite component proportion analysis method based on a depth generation model, the method comprising:
acquiring a perovskite original data set, wherein the perovskite original data set is derived from a public database and a discussion set;
performing data cleaning on the perovskite original data set to obtain a perovskite standard data set, wherein each perovskite standard data in the perovskite standard data set consists of proportion values and conversion efficiency grades of eight components, and the eight components are MA, FA, cs, rb, pb, sn, br and I;
constructing a perovskite conversion analysis model to be trained, wherein the perovskite conversion analysis model consists of an encoder, a generator, a tag discriminator and a feature discriminator;
the construction of the perovskite conversion analysis model to be trained comprises the following steps:
determining a first activation function and a second activation function, wherein the first activation function is a LeakyReLU, and the second activation function is a ReLU;
constructing and obtaining an encoder according to the sequence of the one-dimensional convolution layer, the batch normalization function and the first activation function;
constructing a generator based on three one-dimensional transpose convolutions and a second activation function;
constructing and obtaining the perovskite conversion analysis model according to the front of the encoder and the rear of the generator and by combining a pre-constructed tag discriminator and a feature discriminator; the tag identifier and the feature identifier respectively comprise two fully-connected layers, and share three one-dimensional convolution layers, and the layers are linked by using a LeakyReLU activation function;
Training a perovskite conversion analysis model by using the perovskite standard data set until the training condition is met and the training is exited, so as to obtain a trained perovskite conversion analysis model;
training the perovskite conversion analysis model by using the perovskite standard data set until the training condition is met, and quitting the training to obtain a trained perovskite conversion analysis model, wherein the training comprises the following steps:
setting a loss function and an optimization algorithm of a perovskite conversion analysis model, wherein the loss function comprises three loss functions, namely a reconstruction loss function, a KL divergence function and an anti-loss function;
constructing an input feature, wherein the input feature is composed of a first channel feature and a second channel feature, and the first channel feature comprises a perovskite host component and the second channel feature comprises a perovskite doping component;
the input features are imported into a perovskite conversion analysis model, and the dimensionality of the input features is reduced based on an encoder, so that low-dimensional features are obtained;
reconstructing the low-dimensional features into reconstructed features using the generator, wherein the dimensions of the reconstructed features are the same as the dimensions of the input features;
calculating the characteristic difference value of the reconstruction characteristic and the input characteristic by utilizing a characteristic discriminator, and simultaneously analyzing the conversion efficiency grade of the reconstruction characteristic to obtain an analysis efficiency grade;
Calculating a difference value of the analysis efficiency grade and the conversion efficiency grade based on the label discriminator to obtain a label difference value;
on the premise of optimizing an algorithm, calculating a characteristic difference value and a loss value of a label difference value based on the three loss functions, and iterating model parameters of a perovskite conversion analysis model through the loss values;
until the iteration times are greater than a preset iteration threshold, exiting training to obtain a trained perovskite conversion analysis model;
receiving a target efficiency conversion grade, and reversely inputting the target efficiency conversion grade into a trained perovskite conversion analysis model to obtain an analyzed target perovskite component proportion;
and preparing a target perovskite sample according to the target perovskite component proportion, and performing photoelectric conversion rate test on the target perovskite sample to obtain target photoelectric conversion rate, so as to complete perovskite component proportion analysis based on a depth generation model.
2. The depth generation model based perovskite component ratio analysis method as claimed in claim 1, wherein performing data cleaning on the perovskite raw dataset to obtain a perovskite standard dataset comprises:
sequentially extracting each perovskite raw data from the perovskite raw data set, and performing the following operation on each perovskite raw data:
Judging the integrity of the perovskite original data, and if the perovskite original data lacks component names and proportions, directly eliminating the perovskite original data;
if the perovskite raw data are complete data, judging whether the solar cell constructed based on the perovskite raw data is a 3D organic-inorganic hybrid halide perovskite solar cell or not;
if the solar cell constructed based on the perovskite raw data is not a 3D organic-inorganic hybrid halide perovskite solar cell, directly rejecting the perovskite raw data;
if the solar cell constructed based on the perovskite raw data is a 3D organic-inorganic hybrid halide perovskite solar cell, acquiring photoelectric conversion efficiency corresponding to the 3D organic-inorganic hybrid halide perovskite solar cell constructed based on the perovskite raw data, and acquiring a test environment of the photoelectric conversion efficiency;
after eliminating perovskite original data with the photoelectric conversion efficiency of 0, judging whether the test environment is a standard test environment with the atmospheric quality AM1.5, the calibration irradiance of 1000W/m < 2 > and the calibration temperature of 25 ℃;
if the test environment is not the AM1.5, 1000W/m < 2 > and 25 ℃ standard test environment, directly eliminating the perovskite original data; if the test environment is AM1.5, 1000W/m 2 and standard test environment at 25 ℃, performing structural transformation on the perovskite original data to obtain perovskite standard data;
And converging each group of perovskite standard data and corresponding photoelectric conversion efficiency to obtain a perovskite standard data set.
3. The depth generation model-based perovskite component ratio analysis method as claimed in claim 2, wherein the performing structural transformation on the perovskite raw data to obtain perovskite standard data includes:
performing component separation operation on perovskite raw data to obtain eight components and ratio values, wherein the eight components are MA, FA, cs, rb, pb, sn, br and I respectively;
taking MA, FA, pb and I as perovskite main components, and taking Cs, rb, sn and Br as perovskite doping components; the method comprises the steps of,
according to a preset level interval, mapping the photoelectric conversion efficiency into a conversion efficiency level;
summarizing the ratio values and conversion efficiency grades of the perovskite main body component and the perovskite doping component to obtain the perovskite standard data.
4. The depth generation model-based perovskite component ratio analysis method as claimed in claim 3, wherein the conversion efficiency level includes high efficiency conversion efficiency, medium efficiency conversion efficiency and low efficiency conversion efficiency, and the high efficiency conversion efficiency is 20% or more, the medium efficiency conversion efficiency is 16% or more and 20% or less, and the low efficiency conversion efficiency is 16% or less.
5. The depth generation model based perovskite component ratio analysis method as claimed in claim 1, wherein the KL divergence function is:
;
wherein,representing a gaussian distribution function +.>Representation generator generates the +.>Probability distribution function of individual dimensions->Representing the +.>KL-divergence value of probability distribution function of individual dimensions relative to gaussian distribution function, +.>And->Represents the +.o of the reconstruction feature generated by the generator>The mean and standard deviation of each dimension relative to the two-dimensional input feature,representing the dimensions of the reconstructed feature.
6. The depth generation model based perovskite component ratio analysis method as claimed in claim 1, wherein the reconstruction loss function and the antagonism loss function each employ a cross entropy loss function.
7. A perovskite component proportion analysis device based on a depth generation model, the device comprising:
the perovskite data set acquisition module is used for acquiring a perovskite original data set, wherein the perovskite original data set is derived from a public database and a discussion set;
the data set cleaning module is used for performing data cleaning on the perovskite original data set to obtain a perovskite standard data set, wherein each perovskite standard data in the perovskite standard data set consists of a proportion value and a conversion efficiency grade of eight components, and the eight components are MA, FA, cs, rb, pb, sn, br and I;
The perovskite transformation analysis model building module is used for building a perovskite transformation analysis model to be trained, wherein the perovskite transformation analysis model consists of an encoder, a generator, a tag discriminator and a feature discriminator; the construction of the perovskite conversion analysis model to be trained comprises the following steps: determining a first activation function and a second activation function, wherein the first activation function is a LeakyReLU, and the second activation function is a ReLU; constructing and obtaining an encoder according to the sequence of the one-dimensional convolution layer, the batch normalization function and the first activation function; constructing a generator based on three one-dimensional transpose convolutions and a second activation function; constructing and obtaining the perovskite conversion analysis model according to the front of the encoder and the rear of the generator and by combining a pre-constructed tag discriminator and a feature discriminator; the tag identifier and the feature identifier respectively comprise two fully-connected layers, and share three one-dimensional convolution layers, and the layers are linked by using a LeakyReLU activation function;
the model training and application module is used for training the perovskite conversion analysis model by utilizing the perovskite standard data set until the training condition is met, and quitting the training to obtain a trained perovskite conversion analysis model; training the perovskite conversion analysis model by using the perovskite standard data set until the training condition is met, and quitting the training to obtain a trained perovskite conversion analysis model, wherein the training comprises the following steps: setting a loss function and an optimization algorithm of a perovskite conversion analysis model, wherein the loss function comprises three loss functions, namely a reconstruction loss function, a KL divergence function and an anti-loss function; constructing an input feature, wherein the input feature is composed of a first channel feature and a second channel feature, and the first channel feature comprises a perovskite host component and the second channel feature comprises a perovskite doping component; the input features are imported into a perovskite conversion analysis model, and the dimensionality of the input features is reduced based on an encoder, so that low-dimensional features are obtained; reconstructing the low-dimensional features into reconstructed features using the generator, wherein the dimensions of the reconstructed features are the same as the dimensions of the input features; calculating the characteristic difference value of the reconstruction characteristic and the input characteristic by utilizing a characteristic discriminator, and simultaneously analyzing the conversion efficiency grade of the reconstruction characteristic to obtain an analysis efficiency grade; calculating a difference value of the analysis efficiency grade and the conversion efficiency grade based on the label discriminator to obtain a label difference value; on the premise of optimizing an algorithm, calculating a characteristic difference value and a loss value of a label difference value based on the three loss functions, and iterating model parameters of a perovskite conversion analysis model through the loss values; until the iteration times are greater than a preset iteration threshold, exiting training to obtain a trained perovskite conversion analysis model; receiving a target efficiency conversion grade, and reversely inputting the target efficiency conversion grade into a trained perovskite conversion analysis model to obtain an analyzed target perovskite component proportion;
And the experiment verification module is used for preparing a target perovskite sample according to the proportion of the target perovskite components, and carrying out photoelectric conversion rate test on the target perovskite sample to obtain the target photoelectric conversion rate.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311107647.XA CN116825227B (en) | 2023-08-31 | 2023-08-31 | Perovskite component proportion analysis method and device based on depth generation model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311107647.XA CN116825227B (en) | 2023-08-31 | 2023-08-31 | Perovskite component proportion analysis method and device based on depth generation model |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116825227A CN116825227A (en) | 2023-09-29 |
CN116825227B true CN116825227B (en) | 2023-11-14 |
Family
ID=88122520
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311107647.XA Active CN116825227B (en) | 2023-08-31 | 2023-08-31 | Perovskite component proportion analysis method and device based on depth generation model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116825227B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117114143B (en) * | 2023-10-23 | 2024-02-02 | 桑若(厦门)光伏产业有限公司 | Perovskite passivating agent passivation strategy evaluation method, device, equipment and medium |
CN117275634B (en) * | 2023-11-20 | 2024-05-28 | 桑若(厦门)光伏产业有限公司 | Perovskite solar cell design method and device based on machine learning |
CN117709132B (en) * | 2024-02-05 | 2024-04-19 | 安徽大学 | Diagnostic method for internal loss mechanism of solar cell |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113223639A (en) * | 2021-03-15 | 2021-08-06 | 电子科技大学 | Method for exploring structure, composition and property of perovskite oxide |
CN115034959A (en) * | 2022-06-20 | 2022-09-09 | 重庆大学 | High-definition image translation method based on cross-channel fusion space attention mechanism |
CN115547425A (en) * | 2022-11-25 | 2022-12-30 | 安徽大学 | Method for predicting perovskite lattice constant based on atomic scale feature descriptor |
CN115579089A (en) * | 2022-11-09 | 2023-01-06 | 昆明理工大学 | Method for screening ideal band gap perovskite material based on machine learning |
CN115984117A (en) * | 2023-02-13 | 2023-04-18 | 西安邮电大学 | Variational self-coding image super-resolution method and system based on channel attention |
CN116153446A (en) * | 2023-02-23 | 2023-05-23 | 南京信息工程大学 | Perovskite solar cell HTL material prediction method based on NLP (non-linear liquid phase) processing |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
AU2020293389A1 (en) * | 2019-06-12 | 2022-02-03 | Ivy Mawusi ASUO | Doped mixed cation perovskite materials and devices exploiting same |
-
2023
- 2023-08-31 CN CN202311107647.XA patent/CN116825227B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113223639A (en) * | 2021-03-15 | 2021-08-06 | 电子科技大学 | Method for exploring structure, composition and property of perovskite oxide |
CN115034959A (en) * | 2022-06-20 | 2022-09-09 | 重庆大学 | High-definition image translation method based on cross-channel fusion space attention mechanism |
CN115579089A (en) * | 2022-11-09 | 2023-01-06 | 昆明理工大学 | Method for screening ideal band gap perovskite material based on machine learning |
CN115547425A (en) * | 2022-11-25 | 2022-12-30 | 安徽大学 | Method for predicting perovskite lattice constant based on atomic scale feature descriptor |
CN115984117A (en) * | 2023-02-13 | 2023-04-18 | 西安邮电大学 | Variational self-coding image super-resolution method and system based on channel attention |
CN116153446A (en) * | 2023-02-23 | 2023-05-23 | 南京信息工程大学 | Perovskite solar cell HTL material prediction method based on NLP (non-linear liquid phase) processing |
Non-Patent Citations (1)
Title |
---|
基于深度学习与迁移学习的钙钛矿功能材料筛选研究;李想;中国优秀硕士学位论文全文数据库 工程科技Ⅰ辑(第04期);1-79 * |
Also Published As
Publication number | Publication date |
---|---|
CN116825227A (en) | 2023-09-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN116825227B (en) | Perovskite component proportion analysis method and device based on depth generation model | |
Liu et al. | Fault diagnosis approach for photovoltaic array based on the stacked auto-encoder and clustering with IV curves | |
CN109934222A (en) | A kind of insulator chain self-destruction recognition methods based on transfer learning | |
CN114048896B (en) | Method, system, equipment and medium for predicting photovoltaic power generation data | |
CN113781466A (en) | Solar panel defect identification method based on multi-scale attention-intensive network | |
CN115936177A (en) | Photovoltaic output power prediction method and system based on neural network | |
CN116345555A (en) | CNN-ISCA-LSTM model-based short-term photovoltaic power generation power prediction method | |
CN116054740A (en) | Fault detection and fault detection model processing method and device for photovoltaic power generation equipment | |
Bouguera et al. | A novel solar energy predictor for communicating sensors | |
CN117613883A (en) | Method and device for predicting generated power, computer equipment and storage medium | |
Choulli et al. | Hybrid optimization based on the analytical approach and the particle swarm optimization algorithm (Ana-PSO) for the extraction of single and double diode models parameters | |
CN115759365A (en) | Photovoltaic power generation power prediction method and related equipment | |
Liu et al. | Experiment‐based supervised learning approach toward condition monitoring of PV array mismatch | |
WO2024061141A1 (en) | Method for remote-sensing sample transfer under common knowledge constraints | |
El Marghichi | A Solar PV Model Parameter Estimation Based on the Enhanced Self-Organization Maps | |
CN116316629A (en) | Data-driven optimal power flow calculation method considering topology feature learning | |
CN113449466B (en) | Solar radiation prediction method and system for optimizing RELM based on PCA and chaos GWO | |
CN115968488A (en) | Model generation method, object detection method, controller, and electronic device | |
CN114861993A (en) | Regional photovoltaic power generation prediction method based on federal learning and deep neural network | |
CN117114143B (en) | Perovskite passivating agent passivation strategy evaluation method, device, equipment and medium | |
CN114092838A (en) | Photovoltaic defect detection method based on deep learning target detection | |
CN117275634B (en) | Perovskite solar cell design method and device based on machine learning | |
CN112153347A (en) | Coal mine underground intelligent visual perception terminal, perception method, storage medium and electronic equipment | |
CN117748500B (en) | Photovoltaic power prediction method, device, equipment and medium | |
Tan et al. | Parameters identification of solar cell models using LSHADE hybrid with TSO |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |