A perovskite battery power generation efficiency evaluation system and method based on artificial intelligence

Abstract

The invention discloses a system and method for evaluating power generation efficiency of a perovskite battery based on artificial intelligence, and relates to the field of power generation efficiency data processing. The invention collects various types of perovskite battery samples, thereby improving the richness of subsequently acquired preparation process parameter data and power generation efficiency reflection quantity data; iteratively classifies the reflection quantity data set matrix, reduces multiple reflection quantity data set matrices to category data, and improves the efficiency of subsequent classification evaluation of power generation efficiency levels corresponding to the preparation process parameter data; constructs an SVM model according to a first power generation efficiency level data matrix and a preparation process parameter set matrix, and provides better model support for subsequent evaluation of the power generation efficiency level of the perovskite battery to be evaluated; and finally adopts a second final SVM model to evaluate the power generation efficiency level of the perovskite battery to be evaluated, thereby achieving the purpose of evaluating the power generation efficiency of the perovskite battery.

Description

Technical Field

[0001] The present invention belongs to the field of power generation efficiency data processing, and in particular, relates to a perovskite battery power generation efficiency evaluation system and method based on artificial intelligence. Background Art

[0002] Currently, the power generation efficiency of perovskite batteries is usually evaluated using indicators such as short-circuit current, open-circuit voltage and maximum power. However, different evaluation methods use different test conditions, such as light intensity, temperature, humidity, etc. This will lead to differences in the measured data of indicators such as short-circuit current, open-circuit voltage and maximum power, which in turn leads to differences in the evaluation results and it is impossible to obtain a unified evaluation result. Summary of the invention

[0003] In response to the problems in the related technology, the present invention proposes a perovskite battery power generation efficiency evaluation system and method based on artificial intelligence to overcome the above-mentioned technical problems existing in the existing related technology.

[0004] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0005] The present invention is a method for evaluating the power generation efficiency of a perovskite battery based on artificial intelligence, comprising the following steps:

[0006] S1, perovskite battery sample collection module collects various types of perovskite battery samples;

[0007] The first preparation process parameter collection module collects the preparation process parameters of the perovskite battery sample to obtain a preparation process parameter set matrix;

[0008] The power generation efficiency reflection quantity data acquisition module calculates and acquires the power generation efficiency reflection quantity data of the perovskite battery sample to obtain a reflection quantity data set matrix;

[0009] S2, the reflection quantity data classification module iteratively classifies the reflection quantity data set matrix to obtain a final reflection quantity data set classification matrix;

[0010] S3, a power generation efficiency level data construction module constructs a corresponding first power generation efficiency level data matrix according to the final reflection amount data set classification matrix;

[0011] S4, a classification model building module builds an SVM model according to the first power generation efficiency level data matrix and the preparation process parameter set matrix to obtain a second final SVM model;

[0012] S5. The evaluation module uses the second final SVM model to evaluate the power generation efficiency level of the perovskite battery to be evaluated, and obtains power generation efficiency level data corresponding to the perovskite battery to be evaluated;

[0013] By collecting perovskite battery samples, a sample basis is provided for the subsequent acquisition of battery preparation process parameter data and power generation efficiency reflection data; collecting various types of perovskite battery samples improves the richness of samples, thereby improving the richness of the preparation process parameter data and power generation efficiency reflection data obtained subsequently, making the present technical solution more universal; by collecting perovskite battery sample preparation process parameters, data support is provided for the subsequent analysis of the power generation efficiency of battery samples from the level of battery sample preparation process parameters; then obtaining the power generation efficiency reflection data of the perovskite battery samples provides an evaluation standard for evaluating the power generation efficiency of the battery samples, and provides a data basis for the subsequent determination of the power generation efficiency level; the reflection data Iterative classification is performed according to the set matrix, and the reflection quantity data set matrix is ​​divided into multiple categories, each category corresponds to a power generation efficiency level, and multiple reflection quantity data set matrices are reduced to category data, which reduces the complexity of the data and improves the efficiency of subsequent classification and evaluation of the power generation efficiency level corresponding to the preparation process parameter data; an SVM model is constructed according to the first power generation efficiency level data matrix and the preparation process parameter set matrix, so that the obtained second final SVM model has a good classification ability for the preparation process parameter set matrix, which provides a good model support for the subsequent evaluation of the power generation efficiency level of the perovskite battery to be evaluated; finally, the second final SVM model is used to evaluate the power generation efficiency level of the perovskite battery to be evaluated, thereby achieving the purpose of evaluating the power generation efficiency of the perovskite battery.

[0014] Preferably, the S1 comprises the following steps:

[0015] S11. Set the perovskite battery type set , Indicates perovskite cell types, Indicates the total number of set perovskite battery types; collects the perovskite battery type set Multiple perovskite cell samples of each type are obtained to obtain a perovskite cell sample matrix ;as follows,

[0016] ;

[0017] in, Indicates The perovskite cell type corresponds to samples, Indicates the total number of samples collected for each perovskite cell type;

[0018] S12. Set the preparation process parameter type set , Indicates Preparation process parameter types, Indicates the total number of set preparation process parameter types; according to the preparation process parameter type set Collecting perovskite battery sample matrix The preparation process parameters of each perovskite battery sample are obtained to obtain the preparation process parameter set matrix ;as follows,

[0019] ;

[0020] in, Indicates The perovskite cell type corresponds to The preparation process parameter set of samples, , express The Parameter data of a preparation process parameter type;

[0021] S13, the perovskite battery sample matrix Each perovskite battery sample is in working state; set the initial working time period and the battery power generation efficiency reflection quantity set , and represent the maximum power point efficiency and the fill factor respectively; the initial working time period is divided into a test time period to obtain a test time period set , Indicates the divided Test time period, Indicates the total number of divided test time periods;

[0022] Calculate the perovskite battery sample matrix The maximum power point efficiency and fill factor of each perovskite battery sample in each test time period are obtained to obtain the reflected quantity data set matrix ;as follows,

[0023] ;

[0024] in, and Respectively represent The perovskite cell type corresponds to Maximum power point efficiency data and fill factor data of each sample during the initial working period;

[0025] The types of perovskite cells include planar perovskite cells, mesoporous perovskite cells, perovskite / Si stacked cells, quantum dot-sensitized perovskite cells and all-perovskite stacked cells. The sample richness is improved by collecting various types of perovskite cell samples. By dividing the initial working time period, the power generation efficiency reflection data of the perovskite cell samples in multiple time periods are calculated and obtained, and then the average value is calculated, so that the final power generation efficiency reflection data can more accurately reflect the power generation efficiency of the perovskite cell samples, thereby improving the accuracy of the final evaluation.

[0026] Preferably, in S13, the perovskite battery sample matrix is ​​calculated The maximum power point efficiency and fill factor of each perovskite cell sample in each test time period include the following steps:

[0027] S131, setting standard test conditions and test equipment; testing the perovskite battery sample matrix The external environment was set to standard test conditions when testing each perovskite cell sample;

[0028] The test equipment is used to test the perovskite battery sample matrix The current-voltage characteristic curve of each perovskite battery sample is plotted to obtain the current-voltage characteristic curve matrix set , Indicates The perovskite cell type corresponds to The samples in The current-voltage characteristic curve matrix within the test time period; as follows,

[0029] ;

[0030] in, Indicates The perovskite cell type corresponds to The samples in Current-voltage characteristic curve within a test period;

[0031] S132, calculating the current-voltage characteristic curve matrix set The maximum power point of each curve in the voltage-current-maximum power set matrix is ​​obtained. , Indicates The perovskite cell type corresponds to The samples in The voltage-current-maximum power set matrix within a test period; as follows,

[0032] ;

[0033] in, , and Respectively represent The perovskite cell type corresponds to The samples in The current data, voltage data and maximum power data corresponding to the maximum power point in the current-voltage characteristic curve within a test time period; ;

[0034] For the voltage-current-maximum power set matrix The voltage, current and maximum power in the voltage-current-maximum power set matrix of each time period are averaged to obtain the voltage-current-maximum power average value set matrix ;as follows,

[0035] ;

[0036] in, , and Respectively represent the voltage-current-maximum power set matrix set The average value of each voltage, current and maximum power in the voltage-current-maximum power set matrix of each time period in ; the calculation formula is as follows,

[0037] ; ; ;

[0038] S133, according to the voltage-current-maximum power average value set matrix Calculate the maximum power point efficiency data matrix ;as follows,

[0039] ;

[0040] in, Indicates The perovskite cell type corresponds to Maximum power point efficiency data of samples; ; Indicates the light intensity, Indicates The perovskite cell type corresponds to The battery area of ​​each sample;

[0041] S134, operating the test equipment to place the perovskite battery sample matrix in each test time period The voltage across each perovskite battery sample is adjusted to the voltage-current-maximum power set matrix The voltage value in each voltage-current-maximum power set corresponds to the value in the voltage-current-maximum power set, and the adjusted perovskite battery sample matrix is ​​obtained;

[0042] S135, setting a current threshold; adjusting the voltage across each perovskite battery sample in the adjusted perovskite battery sample matrix by the test device, keeping the maximum power unchanged, until the current of each perovskite battery sample in the adjusted perovskite battery sample matrix is ​​equal to the current threshold, taking the current of each perovskite battery sample at this time as the open circuit voltage, and obtaining the open circuit voltage data matrix ;as follows,

[0043] ;

[0044] in, Indicates The perovskite cell type corresponds to Open circuit voltage data of samples;

[0045] S136, according to the open circuit voltage data matrix And the voltage-current-maximum power average value set matrix Calculate the filling factor of each perovskite cell sample and obtain the filling factor data matrix ;

[0046] ;

[0047] in, Indicates The perovskite cell type corresponds to The filling factor data of the samples is as follows:

[0048] ;

[0049] The fill factor data matrix and maximum power point efficiency data matrix Merge to obtain the reflected amount data set matrix ;

[0050] By using testing equipment to draw the current-voltage characteristic curve of the perovskite battery sample, and then calculating the maximum power based on the current-voltage characteristic curve, and then calculating the maximum power point efficiency data; then calculating the filling factor data of the perovskite battery sample based on the maximum power and open circuit voltage data; thereby providing an evaluation standard for the subsequent evaluation of the power generation efficiency of the perovskite battery samples.

[0051] Preferably, S2 comprises the following steps:

[0052] S21, constructing a first initial SVM model; setting the initial penalty factor of the first initial SVM model to The kernel function is a radial basis kernel function, and the initial width of the radial basis kernel function of the first initial SVM model is set to ;

[0053] S22, the reflection amount data set matrix Input into the first initial SVM model for classification operation to obtain the classification matrix of the initial reflection quantity data set ;as follows,

[0054] ;

[0055] in, and Represents the classification matrix of the initial reflection data set Middle The first in the classification of the reflection quantity dataset Maximum power point efficiency data and fill factor data in the reflection quantity data set; Represents the classification matrix of the initial reflection data set Middle The total number of reflection quantity data sets in the reflection quantity data set category; Represents the matrix of the reflection quantity data set The total number of categories obtained by classification;

[0056] S23, classifying the initial reflection amount data set into a matrix The maximum power point efficiency data and fill factor data in each row are averaged to obtain the classification set of the initial reflection quantity mean data set. , Represents the classification matrix of the initial reflection data set Middle The mean value data set of the reflection quantity classified by the reflection quantity data set; , and Respectively The average maximum power point efficiency data and average fill factor data in the calculation formula are as follows:

[0057] ; ;

[0058] S24. Setting the Euclidean distance threshold ; Calculate the classification matrix of the initial reflection amount data set The Euclidean distances of each reflection quantity data set in each row to the corresponding row of the reflection quantity mean data set are summed up to obtain the initial Euclidean distance and data The calculation formula is as follows:

[0059] ;

[0060] when When the initial reflection quantity data set is classified into the matrix As the final reflection quantity data set classification matrix; otherwise, the first initial SVM model is optimized and adjusted until So far, the final classification matrix of the reflection data set is obtained ;

[0061] ;

[0062] in, and Respectively represent the classification matrix of the final reflection data set Middle The first in the classification of the reflection quantity dataset Maximum power point efficiency data and fill factor data in the reflection quantity data set; Represents the final reflection data set classification matrix Middle The total number of reflection quantity data sets in the reflection quantity data set category;

[0063] The reflection quantity data set matrix is ​​classified by constructing the first initial SVM model, and the classification effect is evaluated by calculating the sum of the Euclidean distances of the classification matrix of the initial reflection quantity data set obtained after classification. If the classification effect is good, it means that the classification is successful; otherwise, the parameters of the first initial SVM model are adjusted and classified again to meet the classification standards; by using the above operation for classification, the classification effect is good, thereby improving the accuracy of the subsequent construction of power generation efficiency grade data based on the classified data.

[0064] Preferably, in S24, the first initial SVM model is optimized and adjusted until So far, obtaining the final classification matrix of the reflection quantity data set specifically includes the following steps:

[0065] S241. Constructing the first locust population , Indicates the first locust population A locust. represents the size of the first locust population; sets the first maximum number of iterations of the first locust population to , the first current iteration number is And the search space dimension is two-dimensional;

[0066] S242: initial penalty factor according to the first initial SVM model And the initial width Randomly generate the initial position of each locust in the first locust population to obtain the first initial position set , and Respectively represent the first locust population The one-dimensional and two-dimensional components of the initial position of the locust are generated by the following formulas:

[0067] ; ;

[0068] In the formula, and Respectively represent The first random number and the second random number generated by the

[0069] S243: Setting the first locust population The fitness function ;as follows,

[0070] ;

[0071] S244, start iterating, and set the first current iteration number Set to 1;

[0072] In each iteration, the first locust population The fitness function Calculate the fitness value of each locust to obtain a first fitness value set;

[0073] Selecting the highest fitness value in the first fitness value set and the position of the locust individual corresponding to the highest fitness value as the first optimal fitness value and the first optimal position, respectively;

[0074] The first locust population is selected according to the first optimal fitness value and the first optimal position. The position of each locust in is updated; after the update is completed, the first current iteration number Perform plus one;

[0075] S245, when When , stop the iteration and get the first optimal position , and Represent the first optimal penalty factor and the first optimal width respectively; and Substituting into the first initial SVM model to obtain a first optimized SVM model;

[0076] S246, the reflection amount data set matrix in S22 Input into the first optimized SVM model for classification to obtain a first reflection quantity data set classification matrix; process the first reflection quantity data set classification matrix in the manner of S23 and S24 to obtain a first Euclidean distance and data;

[0077] When the first Euclidean distance and the data are less than When the first reflection quantity data set classification matrix is ​​used as the final reflection quantity data set classification matrix ;

[0078] Otherwise, continue the iterative operation of S244 until the first Euclidean distance and the data are less than When the final reflection quantity data set classification matrix is ​​obtained ;

[0079] When the classification effect is not good, the locust optimization algorithm is used to adjust the penalty factor of the first initial SVM model and the width of the kernel function, so that the adjustment efficiency is higher and it is not easy to fall into the local optimal situation. Finally, the first optimal penalty factor and the first optimal width are obtained, so that the SVM model can meet the classification requirements.

[0080] Preferably, S3 comprises the following steps:

[0081] S31, according to the classification matrix of the final reflection amount data set Set the power generation efficiency level set , Represents the classification matrix of the final reflection quantity data set Middle The power generation efficiency level corresponding to the classification of the reflection quantity data set;

[0082] S32, according to the final reflection amount data set classification matrix And the power generation efficiency level set Get the preparation process parameter set matrix The power generation efficiency level data corresponding to each preparation process parameter set in the first power generation efficiency level data matrix is ​​obtained. ;as follows,

[0083] ;

[0084] in, Indicates The perovskite cell type corresponds to The power generation efficiency level data of the preparation process parameter set of samples;

[0085] By constructing the first power generation efficiency level data matrix based on the final reflection quantity data set classification matrix, the complexity of the data is reduced, which facilitates the subsequent construction of a classification model that can establish a connection between the preparation process parameter set matrix and the first power generation efficiency level data matrix.

[0086] Preferably, S4 comprises the following steps:

[0087] S41, constructing a second initial SVM model; the kernel function of the second initial SVM model is a radial basis kernel function;

[0088] S42, using the first power generation efficiency level data matrix And the preparation process parameter set matrix Training and testing the second initial SVM model to obtain a second final SVM model;

[0089] By using the first power generation efficiency level data matrix and the preparation process parameter set matrix to train and test the second initial SVM model, the obtained second final SVM model can better establish the correspondence between the preparation process parameter set matrix and the first power generation efficiency level data matrix, thereby facilitating the subsequent classification of the preparation process parameter set of the perovskite battery to be evaluated.

[0090] Preferably, the S42 comprises the following steps:

[0091] S421, setting a training data ratio; and performing a training on the first power generation efficiency level data matrix according to the training data ratio. And the preparation process parameter set matrix Performing data division to obtain a first power generation efficiency level training data matrix, a first power generation efficiency level test data matrix, a preparation process parameter training set matrix, and a preparation process parameter test set matrix;

[0092] S422, setting a training error threshold; inputting the first power generation efficiency level training data matrix and the preparation process parameter training set matrix into the second initial SVM model for training operation; during the training process, when the training error is less than the training error threshold, stopping the training to obtain a second trained SVM model;

[0093] Set the initial penalty factor of the second trained SVM model to ; Set the initial width of the radial basis kernel function of the second trained SVM model to ;

[0094] S423, setting accuracy threshold ; Input the first power generation efficiency level test data matrix and the preparation process parameter test set matrix into the second trained SVM model for testing to obtain the first test accuracy ;

[0095] when When the second trained SVM model is used as the second final SVM model; otherwise, the second trained SVM model is adjusted and optimized to obtain the second final SVM model;

[0096] Through the training and testing process, when the test result does not meet the requirements, the parameters of the second trained SVM model are optimized so that the second trained SVM model has better robustness.

[0097] Preferably, in S423, adjusting and optimizing the second trained SVM model to obtain the second final SVM model comprises the following steps:

[0098] S4231. Constructing the second locust population , Indicates the second locust population A locust. represents the size of the second locust population; sets the second maximum number of iterations of the second locust population to , the second current iteration number is And the search space dimension is two-dimensional;

[0099] S4232: initial penalty factor of the second trained SVM model And the initial width is Randomly generate the initial position of each locust in the second locust population to obtain the second initial position set , and Respectively represent the second locust population The one-dimensional and two-dimensional components of the initial position of the locust are generated by the following formulas:

[0100] ; ;

[0101] in, and Respectively represent The third random number and the fourth random number generated again;

[0102] S4233: Setting the second locust population The fitness function ;as follows,

[0103] ;

[0104] in, represents the correction parameter;

[0105] S4234, start iterating, and set the second current iteration number Set to 1;

[0106] In each iteration, the second locust population The fitness function Calculate the fitness value of each locust to obtain a second fitness value set;

[0107] Selecting the highest fitness value in the second fitness value set and the position of the locust individual corresponding to the highest fitness value as the second optimal fitness value and the second optimal position, respectively;

[0108] The second locust population is selected according to the second optimal fitness value and the second optimal position. The position of each locust in is updated; after the update is completed, the second current iteration number Perform plus one;

[0109] S4235, when When , stop the iteration and get the second best position , and Represent the second optimal penalty factor and the second optimal width respectively; and Substituting into the second trained SVM model to obtain an optimized second trained SVM model;

[0110] S4236: Input the first power generation efficiency level test data matrix and the preparation process parameter test set matrix in S423 into the optimized second trained SVM model for testing to obtain the second test accuracy. ;

[0111] when When the optimized second trained SVM model is used as the second final SVM model; otherwise, the iterative operation in S4234 is continued until So far, the second final SVM model is obtained;

[0112] By using the locust optimization algorithm to adjust the penalty factor and the width of the kernel function of the second trained SVM model, the adjustment efficiency is higher and it is not easy to fall into the local optimal situation. Finally, the second optimal penalty factor and the second optimal width are obtained, so that the classification accuracy of the SVM model can meet the requirements.

[0113] Preferably, the S5 comprises the following steps:

[0114] S51, setting the perovskite battery to be evaluated; according to the preparation process parameter type set Collect the preparation process parameters of the perovskite battery to be evaluated and obtain the preparation process parameter set to be evaluated. , represents the first Parameter data of a preparation process parameter type;

[0115] S52, the preparation process parameter set to be evaluated Input into the second final SVM model for classification operation to obtain the power generation efficiency level data corresponding to the perovskite battery to be evaluated;

[0116] Finally, the second final SVM model that has been trained and tested is used to classify the preparation process parameter set of the perovskite battery to be evaluated, so as to obtain the power generation efficiency level data corresponding to the perovskite battery to be evaluated, thereby achieving the purpose of evaluating the power generation efficiency of the perovskite battery to be evaluated.

[0117] A perovskite battery power generation efficiency evaluation system based on artificial intelligence, comprising a perovskite battery sample collection module, a first preparation process parameter collection module, a power generation efficiency reflection quantity data acquisition module, a reflection quantity data classification module, a power generation efficiency grade data construction module, a classification model construction module and an evaluation module;

[0118] The perovskite battery sample collection module is used to collect various types of perovskite battery samples;

[0119] The first preparation process parameter acquisition module is used to collect the preparation process parameters of the perovskite battery sample to obtain a preparation process parameter set matrix;

[0120] The power generation efficiency reflection amount data acquisition module is used to calculate and acquire the power generation efficiency reflection amount data of the perovskite battery sample to obtain a reflection amount data set matrix;

[0121] The reflection quantity data classification module is used to iteratively classify the reflection quantity data set matrix to obtain a final reflection quantity data set classification matrix;

[0122] The power generation efficiency level data construction module is used to construct a corresponding first power generation efficiency level data matrix according to the final reflection amount data set classification matrix;

[0123] The classification model building module is used to build an SVM model according to the first power generation efficiency level data matrix and the preparation process parameter set matrix to obtain a second final SVM model;

[0124] The evaluation module is used to evaluate the power generation efficiency level of the perovskite battery to be evaluated by using the second final SVM model to obtain power generation efficiency level data corresponding to the perovskite battery to be evaluated.

[0125] The present invention has the following beneficial effects:

[0126] 1. In the present invention, the preparation process parameters of the perovskite battery samples are collected to analyze the power generation efficiency of the battery samples from the level of the preparation process parameters of the battery samples. The preparation process parameters are generally fixed, so the evaluation results obtained by evaluating the power generation efficiency through the preparation process parameters are more stable and less diverse; the reflection quantity data set matrix is ​​iteratively classified to reduce multiple reflection quantity data set matrices to category data, thereby improving the efficiency of subsequent classification and evaluation of the power generation efficiency level corresponding to the preparation process parameter data; an SVM model is constructed based on the first power generation efficiency level data matrix and the preparation process parameter set matrix, thereby providing a better model support for the subsequent evaluation of the power generation efficiency level of the perovskite battery to be evaluated; finally, the second final SVM model is used to evaluate the power generation efficiency level of the perovskite battery to be evaluated, thereby achieving the purpose of evaluating the power generation efficiency of the perovskite battery.

[0127] 2. In the present invention, the power generation efficiency reflection data of the perovskite battery samples in multiple time periods are calculated and obtained, and then the average value is calculated, so that the power generation efficiency reflection data finally obtained can more accurately reflect the power generation efficiency of the perovskite battery samples, thereby improving the accuracy of the final evaluation.

[0128] 3. In the present invention, the locust optimization algorithm is used to optimize and adjust the first initial SVM model, so that the adjustment efficiency is higher and it is not easy to fall into the local optimal situation. Finally, the first optimal penalty factor and the first optimal width are obtained, so that the SVM model can meet the classification requirements.

[0129] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0130] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0131] Figure 1 The present invention is a flowchart of a perovskite battery power generation efficiency evaluation system based on artificial intelligence for evaluating the power generation efficiency of a perovskite battery. DETAILED DESCRIPTION

[0132] The following will be combined with the drawings in the embodiments of the invention to clearly and completely describe the technical solutions in the embodiments of the invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. Based on the embodiments in the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.

[0133] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0134] Embodiment 1

[0135] This embodiment is a method for evaluating the power generation efficiency of a perovskite battery based on artificial intelligence, comprising the following steps:

[0136] S1, perovskite battery sample collection module collects various types of perovskite battery samples;

[0137] The first preparation process parameter collection module collects the preparation process parameters of the perovskite battery sample to obtain a preparation process parameter set matrix;

[0138] The power generation efficiency reflection quantity data acquisition module calculates and acquires the power generation efficiency reflection quantity data of the perovskite battery sample to obtain a reflection quantity data set matrix;

[0139] The S1 comprises the following steps:

[0140] S11. Set the perovskite battery type set , Indicates perovskite cell types, Indicates the total number of set perovskite battery types; collects the perovskite battery type set Multiple perovskite cell samples of each type are obtained to obtain a perovskite cell sample matrix ;as follows,

[0141] ;

[0142] in, Indicates The perovskite cell type corresponds to samples, Indicates the total number of samples collected for each perovskite cell type;

[0143] S12. Set the preparation process parameter type set , Indicates Preparation process parameter types, Indicates the total number of set preparation process parameter types; according to the preparation process parameter type set Collecting perovskite battery sample matrix The preparation process parameters of each perovskite battery sample are obtained to obtain the preparation process parameter set matrix ;as follows,

[0144] ;

[0145] in, Indicates The perovskite cell type corresponds to The preparation process parameter set of samples, , express The Parameter data of a preparation process parameter type;

[0146] The types of preparation process parameters include perovskite precursor concentration, spin coating speed, spin coating time, annealing temperature, annealing time, type and thickness of electrode material, etc.;

[0147] S13, the perovskite battery sample matrix Each perovskite battery sample is in working state; set the initial working time period and the battery power generation efficiency reflection quantity set , and represent the maximum power point efficiency and the fill factor respectively; the initial working time period is divided into a test time period to obtain a test time period set , Indicates the divided Test time period, Indicates the total number of divided test time periods;

[0148] Calculate the perovskite battery sample matrix The maximum power point efficiency and fill factor of each perovskite battery sample in each test time period are obtained to obtain the reflected quantity data set matrix ;as follows,

[0149] ;

[0150] in, and Respectively represent The perovskite cell type corresponds to Maximum power point efficiency data and fill factor data of each sample during the initial working period;

[0151] S13 calculates the perovskite battery sample matrix The maximum power point efficiency and fill factor of each perovskite cell sample in each test time period include the following steps:

[0152] S131, setting standard test conditions and test equipment; testing the perovskite battery sample matrix The external environment was set to standard test conditions when testing each perovskite cell sample;

[0153] The test equipment is used to test the perovskite battery sample matrix The current-voltage characteristic curve of each perovskite battery sample is plotted to obtain the current-voltage characteristic curve matrix set , Indicates The perovskite cell type corresponds to The samples in The current-voltage characteristic curve matrix within the test time period; as follows,

[0154] ;

[0155] in, Indicates The perovskite cell type corresponds to The samples in Current-voltage characteristic curve within a test period;

[0156] S132, calculating the current-voltage characteristic curve matrix set The maximum power point of each curve in the voltage-current-maximum power set matrix is ​​obtained. , Indicates The perovskite cell type corresponds to The samples in The voltage-current-maximum power set matrix within a test period; as follows,

[0157] ;

[0158] in, , and Respectively represent The perovskite cell type corresponds to The samples in The current data, voltage data and maximum power data corresponding to the maximum power point in the current-voltage characteristic curve within a test time period; ;

[0159] For the voltage-current-maximum power set matrix The voltage, current and maximum power in the voltage-current-maximum power set matrix of each time period are averaged to obtain the voltage-current-maximum power average value set matrix ;as follows,

[0160] ;

[0161] in, , and Respectively represent the voltage-current-maximum power set matrix set The average value of each voltage, current and maximum power in the voltage-current-maximum power set matrix of each time period in ; the calculation formula is as follows,

[0162] ; ; ;

[0163] S133, according to the voltage-current-maximum power average value set matrix Calculate the maximum power point efficiency data matrix ;as follows,

[0164] ;

[0165] in, Indicates The perovskite cell type corresponds to Maximum power point efficiency data of samples; ; Indicates the light intensity, Indicates The perovskite cell type corresponds to The battery area of ​​each sample;

[0166] S134, operating the test equipment to place the perovskite battery sample matrix in each test time period The voltage across each perovskite battery sample is adjusted to the voltage-current-maximum power set matrix The voltage value in each voltage-current-maximum power set corresponds to the value in the voltage-current-maximum power set, and the adjusted perovskite battery sample matrix is ​​obtained;

[0167] S135, setting a current threshold; adjusting the voltage across each perovskite battery sample in the adjusted perovskite battery sample matrix by the test device, keeping the maximum power unchanged, until the current of each perovskite battery sample in the adjusted perovskite battery sample matrix is ​​equal to the current threshold, taking the current of each perovskite battery sample at this time as the open circuit voltage, and obtaining the open circuit voltage data matrix ;as follows,

[0168] ;

[0169] in, Indicates The perovskite cell type corresponds to Open circuit voltage data of samples;

[0170] S136, according to the open circuit voltage data matrix And the voltage-current-maximum power average value set matrix Calculate the filling factor of each perovskite cell sample and obtain the filling factor data matrix ;

[0171] ;

[0172] in, Indicates The perovskite cell type corresponds to The filling factor data of the samples is as follows:

[0173] ;

[0174] The fill factor data matrix and maximum power point efficiency data matrix Merge to obtain the reflected amount data set matrix ;

[0175] The standard test conditions include standard light intensity and standard temperature; the test equipment includes an IV curve tester;

[0176] S2, the reflection quantity data classification module iteratively classifies the reflection quantity data set matrix to obtain a final reflection quantity data set classification matrix;

[0177] The S2 comprises the following steps:

[0178] S21, constructing a first initial SVM model; setting the initial penalty factor of the first initial SVM model to The kernel function is a radial basis kernel function, and the initial width of the radial basis kernel function of the first initial SVM model is set to ;

[0179] S22, the reflection amount data set matrix Input into the first initial SVM model for classification operation to obtain the classification matrix of the initial reflection quantity data set ;as follows,

[0180] ;

[0181] in, and Represents the classification matrix of the initial reflection data set Middle The first in the classification of the reflection quantity dataset Maximum power point efficiency data and fill factor data in the reflection quantity data set; Represents the classification matrix of the initial reflection data set Middle The total number of reflection quantity data sets in the reflection quantity data set category; Represents the matrix of the reflection quantity data set The total number of categories obtained by classification;

[0182] S23, classifying the initial reflection amount data set into a matrix The maximum power point efficiency data and fill factor data in each row are averaged to obtain the classification set of the initial reflection quantity mean data set. , Represents the classification matrix of the initial reflection data set Middle The mean value data set of the reflection quantity classified by the reflection quantity data set; , and Respectively The average maximum power point efficiency data and average fill factor data in the calculation formula are as follows:

[0183] ; ;

[0184] S24. Setting the Euclidean distance threshold ; Calculate the classification matrix of the initial reflection amount data set The Euclidean distances of each reflection quantity data set in each row to the corresponding row of the reflection quantity mean data set are summed up to obtain the initial Euclidean distance and data ; The calculation formula is as follows,

[0185] ;

[0186] when When the initial reflection quantity data set is classified into the matrix As the final reflection quantity data set classification matrix; otherwise, the first initial SVM model is optimized and adjusted until So far, the final classification matrix of the reflection data set is obtained ;

[0187] ;

[0188] in, and Respectively represent the classification matrix of the final reflection data set Middle The first in the classification of the reflection quantity dataset Maximum power point efficiency data and fill factor data in the reflection quantity data set; Represents the final reflection data set classification matrix Middle The total number of reflection quantity data sets in the reflection quantity data set category;

[0189] In S24, the first initial SVM model is optimized and adjusted until So far, obtaining the final classification matrix of the reflection quantity data set specifically includes the following steps:

[0190] S241. Constructing the first locust population , Indicates the first locust population A locust. represents the size of the first locust population; sets the first maximum number of iterations of the first locust population to , the first current iteration number is And the search space dimension is two-dimensional;

[0191] S242: initial penalty factor according to the first initial SVM model And the initial width Randomly generate the initial position of each locust in the first locust population to obtain the first initial position set , and Respectively represent the first locust population The one-dimensional and two-dimensional components of the initial position of the locust are generated by the following formulas:

[0192] ; ;

[0193] In the formula, and Respectively represent The first random number and the second random number generated by the

[0194] S243: Setting the first locust population The fitness function ;as follows,

[0195] ;

[0196] S244, start iterating, and set the first current iteration number Set to 1;

[0197] In each iteration, the first locust population The fitness function Calculate the fitness value of each locust and update the position of each locust in the first locust population;

[0198] S245, when When , the iteration is stopped, the first optimal position is obtained, and the first optimized SVM model is obtained;

[0199] S246, input the reflection quantity data set matrix in S22 into the first optimized SVM model for classification, and obtain a first reflection quantity data set classification matrix; process the first reflection quantity data set classification matrix in the manner of S23 and S24 to obtain a first Euclidean distance and data; when the first Euclidean distance and data are less than , taking the first reflection quantity data set classification matrix as the final reflection quantity data set classification matrix;

[0200] Otherwise, continue the iterative operation of S244 until the first Euclidean distance and the data are less than When , the final classification matrix of the reflection quantity data set is obtained;

[0201] S3, a power generation efficiency level data construction module constructs a corresponding first power generation efficiency level data matrix according to the final reflection amount data set classification matrix;

[0202] The S3 comprises the following steps:

[0203] S31, setting a power generation efficiency level set according to the classification matrix of the final reflected quantity data set;

[0204] S32, obtaining power generation efficiency level data corresponding to each preparation process parameter set in the preparation process parameter set matrix according to the final reflection amount data set classification matrix and the power generation efficiency level set, and obtaining a first power generation efficiency level data matrix;

[0205] S4, a classification model building module builds an SVM model according to the first power generation efficiency level data matrix and the preparation process parameter set matrix to obtain a second final SVM model;

[0206] The S4 comprises the following steps:

[0207] S41, constructing a second initial SVM model; the kernel function of the second initial SVM model is a radial basis kernel function;

[0208] S42, training and testing the second initial SVM model using the first power generation efficiency level data matrix and the preparation process parameter set matrix to obtain a second final SVM model;

[0209] The S42 comprises the following steps:

[0210] S421, setting a training data ratio; dividing the first power generation efficiency level data matrix and the preparation process parameter set matrix according to the training data ratio to obtain a first power generation efficiency level training data matrix, a first power generation efficiency level test data matrix, a preparation process parameter training set matrix, and a preparation process parameter test set matrix;

[0211] S422, setting a training error threshold; inputting the first power generation efficiency level training data matrix and the preparation process parameter training set matrix into the second initial SVM model for training operation; during the training process, when the training error is less than the training error threshold, stopping the training to obtain a second trained SVM model; setting the initial penalty factor of the second trained SVM model to ; Set the initial width of the radial basis kernel function of the second trained SVM model to ;

[0212] S423, setting accuracy threshold ; Input the first power generation efficiency level test data matrix and the preparation process parameter test set matrix into the second trained SVM model for testing to obtain the first test accuracy ;when When the second trained SVM model is used as the second final SVM model; otherwise, the second trained SVM model is adjusted and optimized to obtain the second final SVM model;

[0213] S5. The evaluation module uses the second final SVM model to evaluate the power generation efficiency level of the perovskite battery to be evaluated, and obtains power generation efficiency level data corresponding to the perovskite battery to be evaluated;

[0214] The S5 comprises the following steps:

[0215] S51, setting a perovskite battery to be evaluated; collecting the preparation process parameters of the perovskite battery to be evaluated according to the preparation process parameter type set to obtain a preparation process parameter set to be evaluated;

[0216] S52, inputting the preparation process parameter set to be evaluated into the second final SVM model for classification operation, and obtaining power generation efficiency level data corresponding to the perovskite battery to be evaluated.

[0217] Embodiment 2

[0218] This embodiment discloses a perovskite battery power generation efficiency evaluation system based on artificial intelligence, which can implement the method of the above embodiment, including a perovskite battery sample collection module, a first preparation process parameter collection module, a power generation efficiency reflection quantity data acquisition module, a reflection quantity data classification module, a power generation efficiency grade data construction module, a classification model construction module and an evaluation module;

[0219] The perovskite battery sample collection module is used to collect various types of perovskite battery samples;

[0220] The first preparation process parameter acquisition module is used to collect the preparation process parameters of the perovskite battery sample to obtain a preparation process parameter set matrix;

[0221] The power generation efficiency reflection amount data acquisition module is used to calculate and acquire the power generation efficiency reflection amount data of the perovskite battery sample to obtain a reflection amount data set matrix;

[0222] The reflection quantity data classification module is used to iteratively classify the reflection quantity data set matrix to obtain a final reflection quantity data set classification matrix;

[0223] The power generation efficiency level data construction module is used to construct a corresponding first power generation efficiency level data matrix according to the final reflection amount data set classification matrix;

[0224] The classification model building module is used to build an SVM model according to the first power generation efficiency level data matrix and the preparation process parameter set matrix to obtain a second final SVM model;

[0225] The evaluation module is used to evaluate the power generation efficiency level of the perovskite battery to be evaluated by using the second final SVM model to obtain power generation efficiency level data corresponding to the perovskite battery to be evaluated.

[0226] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0227] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A method for evaluating the power generation efficiency of a perovskite battery based on artificial intelligence, characterized in that: The following steps are involved: S1, perovskite battery sample collection module collects various types of perovskite battery samples; The first preparation process parameter collection module collects the preparation process parameters of the perovskite battery sample to obtain a preparation process parameter set matrix; The power generation efficiency reflection quantity data acquisition module calculates and acquires the power generation efficiency reflection quantity data of the perovskite battery sample to obtain a reflection quantity data set matrix; The S1 specifically includes the following steps: S11, setting a perovskite battery type set; collecting multiple perovskite battery samples of each type in the perovskite battery type set to obtain a perovskite battery sample matrix; S12, setting a preparation process parameter type set; collecting preparation process parameters of each perovskite battery sample in the perovskite battery sample matrix according to the preparation process parameter type set to obtain a preparation process parameter set matrix; S13, each perovskite battery sample in the perovskite battery sample matrix is ​​in a working state; an initial working time period is set; the initial working time period is divided into a test time period to obtain a test time period set; the maximum power point efficiency and fill factor of each perovskite battery sample in the perovskite battery sample matrix in each test time period within the test time period are calculated to obtain a reflection quantity data set matrix; S2, the reflection quantity data classification module iteratively classifies the reflection quantity data set matrix to obtain a final reflection quantity data set classification matrix; S3, a power generation efficiency level data construction module constructs a corresponding first power generation efficiency level data matrix according to the final reflection amount data set classification matrix; S4, a classification model building module builds an SVM model according to the first power generation efficiency level data matrix and the preparation process parameter set matrix to obtain a second final SVM model; S5. The evaluation module uses the second final SVM model to evaluate the power generation efficiency level of the perovskite battery to be evaluated, and obtains power generation efficiency level data corresponding to the perovskite battery to be evaluated.

2. According to claim 1, a method for evaluating power generation efficiency of a perovskite battery based on artificial intelligence is characterized in that: Calculating the maximum power point efficiency and fill factor of each perovskite battery sample in the perovskite battery sample matrix in each test time period in the test time period in S13 includes the following steps: S131, setting standard test conditions and test equipment; setting the external environment to the standard test conditions when testing each perovskite battery sample in the perovskite battery sample matrix; using the test equipment to draw the current-voltage characteristic curve of each perovskite battery sample in the perovskite battery sample matrix to obtain a current-voltage characteristic curve matrix set; S132, calculating the maximum power point of each curve in the current-voltage characteristic curve matrix set to obtain a voltage-current-maximum power set matrix set; averaging each voltage, current and maximum power in the voltage-current-maximum power set matrix in each time period of the voltage-current-maximum power set matrix to obtain a voltage-current-maximum power average value set matrix; S133, calculating a maximum power point efficiency data matrix according to the voltage-current-maximum power average value set matrix; S134, operating the test equipment in each test time period to adjust the voltage across each perovskite battery sample in the perovskite battery sample matrix to the voltage value corresponding to each voltage-current-maximum power set in the voltage-current-maximum power set matrix, to obtain an adjusted perovskite battery sample matrix; S135, setting a current threshold; adjusting the voltage across each perovskite battery sample in the adjusted perovskite battery sample matrix by the test device, keeping the maximum power unchanged, until the current of each perovskite battery sample in the adjusted perovskite battery sample matrix is ​​equal to the current threshold, taking the current of each perovskite battery sample at this time as the open circuit voltage, and obtaining the open circuit voltage data matrix; S136. Calculate the filling factor of each perovskite battery sample according to the open circuit voltage data matrix and the voltage-current-maximum power average value set matrix to obtain a filling factor data matrix; merge the filling factor data matrix and the maximum power point efficiency data matrix to obtain the reflection quantity data set matrix.

3. The method for evaluating power generation efficiency of a perovskite battery based on artificial intelligence according to claim 2, characterized in that: The S2 comprises the following steps: S21, constructing a first initial SVM model; S22, inputting the reflection quantity data set matrix into the first initial SVM model for classification operation to obtain an initial reflection quantity data set classification matrix; S23, respectively calculating the average of the maximum power point efficiency data and the fill factor data in each row of the classification matrix of the initial reflection quantity data set to obtain a classification set of the initial reflection quantity average data set; S24. Setting the Euclidean distance threshold Calculate the Euclidean distances from each reflection amount data set in each row of the classification matrix of the initial reflection amount data set to the corresponding row of the reflection amount mean data set and sum them up to obtain the initial Euclidean distance and data ;when When the initial reflection amount data set classification matrix is ​​used as the final reflection amount data set classification matrix; otherwise, the first initial SVM model is optimized and adjusted until So far, the final classification matrix of the reflection quantity data set is obtained.

4. The method for evaluating power generation efficiency of a perovskite battery based on artificial intelligence according to claim 3, characterized in that: In S24, the first initial SVM model is optimized and adjusted until So far, obtaining the final classification matrix of the reflection quantity data set specifically includes the following steps: S241, constructing a first locust population; setting a first maximum number of iterations of the first locust population to , the first current iteration number is ; S242, randomly generating an initial position of each locust in the first locust population according to an initial penalty factor and an initial width of the first initial SVM model to obtain a first initial position set; S243, setting the fitness function of the first locust population; S244, starting iteration, calculating the fitness value of each locust according to the fitness function of the first locust population in each round of iteration and updating the position of each locust in the first locust population; S245, when When , the iteration is stopped, and the first optimal position and the first optimized SVM model are obtained; S246, inputting the reflection quantity data set matrix in S22 into the first optimized SVM model for classification, and obtaining a first reflection quantity data set classification matrix; processing the first reflection quantity data set classification matrix in the manner in S23 and S24, and obtaining a first Euclidean distance and data; When the first Euclidean distance and the data are less than , the first reflection quantity data set classification matrix is ​​used as the final reflection quantity data set classification matrix; otherwise, the iterative operation of S244 is continued until the first Euclidean distance and the data are less than When , the final classification matrix of the reflection quantity data set is obtained.

5. The method for evaluating power generation efficiency of a perovskite battery based on artificial intelligence according to claim 4, characterized in that: The S3 comprises the following steps: S31, setting a power generation efficiency level set according to the classification matrix of the final reflected quantity data set; S32. Obtain the power generation efficiency level data corresponding to each preparation process parameter set in the preparation process parameter set matrix according to the final reflection quantity data set classification matrix and the power generation efficiency level set, and obtain a first power generation efficiency level data matrix.

6. The method for evaluating power generation efficiency of a perovskite battery based on artificial intelligence according to claim 5, characterized in that: The S4 comprises the following steps: S41, constructing a second initial SVM model; the kernel function of the second initial SVM model is a radial basis kernel function; S42: Use the first power generation efficiency level data matrix and the preparation process parameter set matrix to train and test the second initial SVM model to obtain a second final SVM model.

7. The method for evaluating power generation efficiency of a perovskite battery based on artificial intelligence according to claim 6, characterized in that: The S42 comprises the following steps: S421, setting a training data ratio; dividing the first power generation efficiency level data matrix and the preparation process parameter set matrix according to the training data ratio to obtain a first power generation efficiency level training data matrix, a first power generation efficiency level test data matrix, a preparation process parameter training set matrix, and a preparation process parameter test set matrix; S422, setting a training error threshold; inputting the first power generation efficiency level training data matrix and the preparation process parameter training set matrix into the second initial SVM model for training operation; during the training process, when the training error is less than the training error threshold, stopping the training to obtain a second trained SVM model; S423, setting accuracy threshold ; Input the first power generation efficiency level test data matrix and the preparation process parameter test set matrix into the second trained SVM model for testing to obtain the first test accuracy ;when When the second trained SVM model is used as the second final SVM model; otherwise, the second trained SVM model is adjusted and optimized to obtain the second final SVM model.

8. The method for evaluating power generation efficiency of a perovskite battery based on artificial intelligence according to claim 7, characterized in that: The S5 comprises the following steps: S51, setting a perovskite battery to be evaluated; collecting the preparation process parameters of the perovskite battery to be evaluated according to the preparation process parameter type set to obtain a preparation process parameter set to be evaluated; S52, inputting the preparation process parameter set to be evaluated into the second final SVM model for classification operation, and obtaining power generation efficiency level data corresponding to the perovskite battery to be evaluated.

9. A system for evaluating power generation efficiency of a perovskite battery based on artificial intelligence, which adopts a method for evaluating power generation efficiency of a perovskite battery based on artificial intelligence according to any one of claims 1 to 8, characterized in that: It includes a perovskite battery sample collection module, a first preparation process parameter collection module, a power generation efficiency reflection quantity data acquisition module, a reflection quantity data classification module, a power generation efficiency grade data construction module, a classification model construction module and an evaluation module.

Images