Raw material purity optimization method for perovskite photovoltaic process based on learning curve model

Abstract

The invention relates to a raw material purity optimization method for a perovskite photovoltaic process based on a learning curve model, and the method comprises the steps: collecting production data of each process link in a perovskite photovoltaic production process, and carrying out the integration, classification and storage, thereby constructing a database; according to the production data, performing segmented fitting to obtain learning curves between the raw material purity and the cumulative yield in each process link, determining a learning coefficient of each learning curve, and performing integration to obtain a perovskite photovoltaic multi-process link learning model; learning coefficients of any two process links are successively selected for sensitivity analysis, and error constraint analysis is performed on the whole perovskite photovoltaic multi-process link learning model, so that the learning coefficients are optimized and adjusted, an optimal perovskite photovoltaic multi-process link learning model is obtained, and an optimal raw material purity scheme is determined. Compared with the prior art, the method has the advantages that the purity of the raw materials can be optimized, the cost is reduced while the yield of the raw materials is increased, and the prediction process is high in accuracy.

Description

Technical Field

[0001] The present invention relates to the field of perovskite photovoltaic production technology, and in particular to a method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model. Background Art

[0002] Renewable energy is the current direction of technological development. Among them, traditional solar cells represented by crystalline silicon technology have very mature industrial production lines and now account for more than 90% of the market share.

[0003] However, the extremely competitive environment has led to the photovoltaic industry currently being in a state of serious overcapacity. Excessive output of photovoltaic modules has caused the spot price of photovoltaic modules in the current market to be lower than the cost, which has caused a serious impact on parts of the photovoltaic industry. Against this background, the photovoltaic industry needs the promotion of emerging photovoltaic technology routes to further reduce the cost of photovoltaic modules.

[0004] Considering that the purity of raw materials directly affects the output in the perovskite photovoltaic production process, although excessively high raw material purity can produce high-quality products, it also brings high costs, reduces processability, and limits output. Therefore, appropriate raw material purity can achieve an increase in output and reduce costs. Summary of the invention

[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a raw material purity optimization method for perovskite photovoltaic process based on a learning curve model, so as to optimize the raw material purity, increase the raw material output and reduce the cost.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] A method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model comprises the following steps:

[0008] Collect production data from each process step in the perovskite photovoltaic production process, integrate and classify them for storage, and thus build a database;

[0009] According to the production data in the database, the learning curve between the raw material purity and the cumulative output in each process link is obtained by segmented fitting, the learning coefficient of each learning curve is determined, and the learning model of multi-process links of perovskite photovoltaics is obtained by integrating them;

[0010] In the perovskite photovoltaic multi-process link learning model, the learning coefficients of any two process links are selected one by one for sensitivity analysis, and the error constraint analysis of the overall perovskite photovoltaic multi-process link learning model is carried out to optimize and adjust the learning coefficients to obtain the optimal perovskite photovoltaic multi-process link learning model, thereby determining the optimal raw material purity solution.

[0011] Furthermore, the acquisition process of the perovskite photovoltaic multi-process link learning model is specifically as follows:

[0012] First, the learning curve of the first process link in perovskite photovoltaic production is fitted. Then, based on the learning curve of the first process link, the learning curves of the subsequent process links are fitted. Finally, the learning curves of each process link are combined to obtain a multi-process link learning model of perovskite photovoltaics.

[0013] Furthermore, the learning curve of the first process step is expressed as:

[0014] lnC0=β0lnP0+V0

[0015] In the formula, C0 is the purity of the input raw materials of the perovskite photovoltaic production line, P0 is the cumulative output of the input raw materials, β0 is the learning coefficient of the input raw material production, and V0 is the adjustment item of the learning curve of the first process link;

[0016] The expression of the learning curve of each subsequent process link is:

[0017] ln i =β i oeLh i +V i

[0018] In the formula, i is the process link, k is i is the ratio of the increase in raw material purity after the i process link compared with the previous process link, C i is the raw material purity of process step i, C i-1 is the raw material purity of process step i-1, P i is the cumulative output of process step i, β i is the learning coefficient of process step i, V i is the adjustment item of the learning curve of process step i;

[0019] The expression of the perovskite photovoltaic multi-process link learning model is:

[0020] LqCy p =(β0+∑t i β i )lnP0+V p

[0021] In the formula, C p To ensure the purity of the perovskite photovoltaic modules produced, t i V is the logarithmic ratio of the cumulative output of raw materials input into the perovskite photovoltaic production to the cumulative output of raw materials input into the i process link, pIt is the overall adjustment item and the cumulative value of the adjustment items of the learning curve of each process link.

[0022] Furthermore, the error constraint analysis is specifically as follows:

[0023] The error between the learning coefficient of the overall perovskite photovoltaic multi-process link learning model and the learning coefficient obtained by fitting the traditional learning model is within a reasonable range, and the corresponding expression is:

[0024]

[0025] In the formula, β t The formula is lnC p =β t lnP0+V t The learning coefficient obtained by direct linear fitting, α is the set reasonable error range.

[0026] Furthermore, the calculation expression of the sensitivity analysis is:

[0027]

[0028] In the formula, β j is the learning coefficient of process step j, P j is the cumulative output of process step j, C ii is the purity of the product obtained when all the products produced in process i are used to manufacture perovskite photovoltaic modules, C jj is the product purity obtained when all the products produced in process step j are used to manufacture perovskite photovoltaic modules, and γ is the sensitivity error range.

[0029] Furthermore, the process steps of the perovskite photovoltaic production process include thin film preparation, laser etching and packaging.

[0030] Further, the database stores data of production data in the perovskite photovoltaic production process according to the reference data source module, the cell detail module, the hierarchical structure module, the synthesis process module and the key indicator module;

[0031] The indicators of the reference data source module include: DOI number, publication date, author and sample ID;

[0032] The indicators of the battery detail module include: architecture, area, flexibility, transparency and component properties;

[0033] The indicators of the hierarchical structure module include: substrate, electron transport layer, perovskite layer, hole transport layer, back contact layer and additional layer;

[0034] The indicators of the synthesis process module include: chemicals, solvents, deposition process, deposition and dopants / additives;

[0035] The indicators of the key indicator module include: IV indicator, quality assessment indicator, stability indicator and outdoor test indicator.

[0036] Furthermore, the indicators of the perovskite layer in the hierarchical structure module include: constituent components, size dimensions, band gap, thickness, and doping;

[0037] The classification process of the constituent components is specifically as follows:

[0038] The perovskite composition is written in the form of ABX3, and its raw material components are divided into A element, B element and X element;

[0039] The A elements are divided into three categories: organic cations, inorganic cations and organic-inorganic phase doping;

[0040] The B elements are divided into two categories: lead and non-lead;

[0041] The X elements are divided into two categories: single element and multi-element doping.

[0042] Furthermore, the hole transport layer in the hierarchical structure module specifically includes an organic hole transport system and an inorganic hole transport system; the organic hole transport system includes a triphenylamine transport material doped with Li-TFSI and tBP; a high molecular conjugated polymer doped with Li-TFSI and tBP; a branched hole transport material Fused-F without adding a dopant; a conjugated small molecule oligothiophene derivative DR3TBDTT with benzothiophene as the core, terthiophene as the arm, and rhodanine as the end cap;

[0043] The inorganic hole transport system includes nickel oxide, cuprous iodide, cuprous thiocyanate, copper oxide, and copper sulfide.

[0044] Furthermore, the additional layers of the hierarchical structure module include a cathode buffer layer and an anode buffer layer.

[0045] Compared with the prior art, the present invention has the following advantages:

[0046] (1) The present invention is different from the traditional single-factor learning model. It uses a multi-factor learning model that integrates multiple process links. Compared with the traditional model, it can more clearly feedback the impact of changes in conditions in each process link on perovskite photovoltaic production, allowing users to adjust the learning coefficient of individual process links within a reasonable range according to their actual situation, and quantitatively analyze its impact on the entire perovskite photovoltaic production line from raw materials to finished components. At the same time, as a prediction model, the multi-factor learning model used in the present invention can predict in advance the future impact of changes in a certain process link (such as a breakthrough in the upgrade of a process link) on the entire industry, and assist users in making reasonable adjustments in advance.

[0047] (2) The present invention pairs the parameters in the constructed learning model, performs relevant sensitivity calculations on them respectively, and traverses to obtain the optimal raw material purity solution that meets the actual user usage constraints.

[0048] (3) The present invention deeply collects relevant information about perovskite photovoltaics, constructs a database related to each level of perovskite photovoltaics, and classifies it according to the data index structure of reference data, battery details, hierarchical structure, synthesis process, and key indicators. It grasps the product characteristics and technical routes of various existing perovskite photovoltaics, and on this basis, refines the technical details of each process link, so that users can select relevant perovskite photovoltaic technology routes according to actual needs and accurately calculate the corresponding relationship between the raw material input flow and product output flow of each process link and the raw material purity and output of each link. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A schematic flow chart of a method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model provided in an embodiment of the present invention;

[0050] Figure 2 A topological schematic diagram of a perovskite photovoltaic database provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0052] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0054] Example 1

[0055] like Figure 1 As shown, this embodiment provides a method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model, comprising the following steps:

[0056] S1: Collect production data of each process link in the perovskite photovoltaic production process, integrate and classify them for storage, and thus build a database;

[0057] S2: Based on the production data in the database, obtain the learning curve between the raw material purity and the cumulative output in each process link by segmented fitting, determine the learning coefficient of each learning curve, and integrate them to obtain the multi-process link learning model of perovskite photovoltaics;

[0058] S3: In the perovskite photovoltaic multi-process link learning model, the learning coefficients of any two process links are selected one by one for sensitivity analysis, and the error constraint analysis of the overall perovskite photovoltaic multi-process link learning model is performed to optimize and adjust the learning coefficients to obtain the optimal perovskite photovoltaic multi-process link learning model, thereby determining the optimal raw material purity solution.

[0059] The following is a detailed description of each step:

[0060] In step S1, Figure 2 As shown, the database stores the production data in the perovskite photovoltaic production process according to the reference data source module, the battery detail module, the hierarchical structure module, the synthesis process module and the key indicator module;

[0061] The indicators of the reference data source module include: DOI number, publication date, author and sample ID;

[0062] Metrics for the battery detail module include: architecture, area, flexibility, transparency, and component properties;

[0063] The indicators of the hierarchical structure module include: substrate, electron transport layer (ETL), perovskite layer (Perovskite), hole transport layer (HTL), back contact layer (Back contact), and additional layer;

[0064] Indicators for the synthesis process module include: chemicals, solvents, deposition steps, deposition, and dopants / additives;

[0065] The indicators of the key indicator module include: IV indicator, quality evaluation (QE) indicator, stability indicator and outdoor test indicator.

[0066] The indicators of the submodule perovskite layer of the hierarchical structure specifically include composition, size, band gap, thickness, and doping;

[0067] The indicators of the deposition process of the sub-module of the synthesis process module specifically include quenching / anti-solvent, annealing, gas environment, humidity, and storage.

[0068] The classification process of constituent components is as follows:

[0069] The perovskite composition is written in ABX3 format, and the components can be divided into three sub-modules: A element regulation, B element regulation and X element regulation.

[0070] Among them, the regulation of A elements is divided into three categories: organic cations, inorganic cations and organic-inorganic phase doping: organic A-type cations specifically include methylamine cations Formamidinium cation Organic cations such as cesium ions (Cs + ) and other inorganic cations, and organic-inorganic doping specifically includes doping a small amount of organic-inorganic hybrid cations such as cesium ions in the A-type cations with methylamine cations as the main body;

[0071] The B element regulation is divided into two modules: lead-based and non-lead-based. In the lead-based module, the B element is mainly composed of lead or entirely composed of lead, and in the lead-free module, the B element specifically includes metal elements such as Sn and Cu.

[0072] The X element control module is divided into a single element submodule and a multi-element doping module. The single element module is specifically characterized in that the X element is completely composed of one of the halogens such as Cl, Br, I, etc., and the multi-element doping module is specifically characterized in that multiple halogens are doped with each other to form the X element.

[0073] In the synthesis process module, the perovskite film formation methods specifically include one-step spin coating, distributed immersion method, two-step spin coating, vapor deposition method and other processes. Among them, the one-step spin coating method, the two-step spin coating method and the vapor deposition method require annealing process, and the vapor deposition method takes a long time for annealing.

[0074] The electron transport layer in the hierarchical structure module specifically includes an inorganic metal oxide electron transport system and an organic electron transport system.

[0075] The inorganic oxide electron transport system mainly includes titanium dioxide, zinc oxide and other compounds, among which titanium dioxide can form a dense layer and also a mesoporous structure on the dense layer: the preparation process of dense layer titanium dioxide mainly includes: sol-gel method, aerosol spray pyrolysis method and spin coating, etc.; mesoporous layer titanium dioxide includes TiO2 nanoparticles, TiO2 nanorods and TiO2 nanofibers, among which the preparation process of TiO2 nanoparticles mainly includes screen printing, spin coating, sol-gel method, etc.; the preparation process of TiO2 nanorods includes organic metal chemical vapor deposition (MOVCD), electrochemical corrosion, hydrothermal method, etc.; the preparation process of TiO2 nanofibers includes electrospinning, etc. The application of zinc oxide in electron transport systems mainly includes dense planes and nanorods and other structural morphologies. The preparation process of ZnO thin films mainly includes chemical bath deposition, sol-gel method, sputtering, electrodeposition and other methods; at the same time, ZnO can be doped with aluminum, PCBM, C3-SAM and other materials to improve the surface morphology, and its doping process mainly includes electrostatic spraying, spin coating, sol-gel method and other methods. Other oxides specifically include insulating oxides such as aluminum oxide, zirconium oxide, and silicon dioxide, and their preparation processes mainly include spin coating, sol-gel method and other methods. Organic electron transport materials mainly include fullerene (C 60 ) and its derivatives, and its preparation process mainly includes distributed spin coating and other methods.

[0076] The hole transport layer in the hierarchical structure module specifically includes an organic hole transport system and an inorganic hole transport system. The organic hole transport system mainly includes triphenylamine transport materials doped with Li-TFSI and tBP (triphenylamine materials specifically include spiro-MeOTAD, PTAA and other materials), high molecular conjugated polymers doped with Li-TFSI and tBP (such as PDPPDBTE), branched hole transport materials Fused-F without adding dopants, and conjugated small molecule oligothiophene derivatives DR3TBDTT with benzothiophene as the core, trithiophene as the arm, and rhodanine end-capping. Its preparation process mainly includes two-step spin coating, hot substrate spin coating, electrochemical in-situ polymerization and other methods. The inorganic hole transport system mainly includes nickel oxide, cuprous iodide, cuprous thiocyanate, copper oxide, and copper sulfide. Among them, the nickel oxide film-making process mainly includes solution method, sol-gel method, combustion chemical method and other methods. Copper components can be doped into the nickel oxide layer to improve the grain morphology of the thin film; the cuprous iodide film-making process mainly includes spin coating and other methods; the cuprous thiocyanate film-making process mainly includes scraping method, electrochemical deposition method, one-step rapid crystallization method, two-step continuous deposition method and other methods; copper oxides mainly include cupric oxide and cuprous oxide, and their film-making processes mainly include solution spin coating and other methods; the copper sulfide film-making process mainly includes solution spin coating and other methods.

[0077] The additional layers of the hierarchical structure module include cathode buffer layer, anode buffer layer, etc. to improve the functional layers of perovskite photovoltaic modules. Among them, the cathode buffer layer materials mainly include metal oxides (such as ZnO nanocrystals), inorganic salts (such as CaMnO3, Cs2CO3), polymer materials (such as PN4N, polyethyleneimine (PEIE), poly [3- (6-trimethylammoniumhexyl) thiophene] (P3TMAHT)), etc.); the anode buffer layer materials mainly include [poly (3, 4-ethylenedioxythiophene): polystyrene sulfonic acid] (PEDOT: PSS), Ni x O etc.

[0078] This embodiment adopts the IEC61215 standard for the determination of key indicator module parameters, wherein the equipment conditions required by the standard include: ultraviolet accelerated aging test machine, temperature control equipment, humidity and freeze test equipment, solar simulator, pulse power supply, current-voltage characteristic tester and other equipment. At the same time, the photovoltaic module needs to be subjected to a double 85 test (high temperature and high humidity environment test), which is mainly completed in a double 85 test experimental box.

[0079] In step S2, the specific process steps of perovskite photovoltaics mainly include three major processes: thin film preparation, laser etching, and packaging.

[0080] The specific process of acquiring the multi-process learning model of perovskite photovoltaics is as follows:

[0081] First, the learning curve of the first process link in perovskite photovoltaic production is fitted. Then, based on the learning curve of the first process link, the learning curves of the subsequent process links are fitted. Finally, the learning curves of each process link are combined to obtain a multi-process link learning model of perovskite photovoltaics.

[0082] The expression of the learning curve of the first process link is:

[0083] lnC0=β0lnP0+V0

[0084] In the formula, C0 is the input raw material purity of the perovskite photovoltaic production line, P0 is the cumulative output of the input raw material, β0 is the learning coefficient of the input raw material production, and V0 is the adjustment item of the learning curve of the first process link. It has no clear specific meaning, and its value is also obtained by linear fitting of the logarithmic raw material purity and cumulative output;

[0085] Among them, the β0 obtained by fitting can be obtained by the learning rate formula Obtain the learning rate of raw material production. At the same time, the input raw material P0 here needs to undergo coefficient transformation. The effect after transformation is: the unit perovskite photovoltaic input raw material P0 can obtain the unit final output perovskite photovoltaic module. The coefficient here is mainly affected by factors such as material loss and product conversion rate. Users can set it according to their actual situation.

[0086] The expression of the learning curve of each subsequent process link is:

[0087] ln i =β i oeLh i +V i

[0088] In the formula, i is the process link, k is i is the ratio of the increase in raw material purity after the i process link compared with the previous process link, C i is the raw material purity of process step i, C i-1 is the raw material purity of process step i-1, P i is the cumulative output of process step i, β i is the learning coefficient of process step i, and its specific value is obtained by linear fitting of the logarithmic increase rate and the cumulative output of process step i, V i is the adjustment item of the learning curve of process step i, and its value is also obtained by linear fitting of the logarithmic increase rate and the cumulative output of process step i;

[0089] in, i-1 represents the process before process i, i.e. C i =k i Ci-1 In this embodiment, when analyzing the purity of the product in the process i, the purity of the product in the process i-1 will be determined first. Therefore, in the learning model of this embodiment, C i Proportional to k i At the same time, considering that the specific process i may be applicable to the production of multiple similar technology products (such as OLED, quantum dots, etc.), the raw material P of the input process i is i , will be greater than or equal to the input perovskite photovoltaic production raw material P0.

[0090] The learning curve expression obtained by combining various process links, that is, the expression of the multi-process link learning model of perovskite photovoltaics is:

[0091] LqCy p =(β0+∑t i β i )lnP0+V p

[0092] In the formula, C p To ensure the purity of the perovskite photovoltaic modules produced, t i It is the ratio of the cumulative output of the raw materials input into the perovskite photovoltaic system after logarithmization to the cumulative output of the raw materials input into the i process link. The specific expression is: V p It is the overall adjustment item, which has no clear specific meaning. It is the cumulative value of the adjustment items of the learning curve of each process link. The specific expression is V p =∑V i .

[0093] This embodiment will select the learning coefficients of any two processes one by one for sensitivity analysis, and meet the optimal raw material purity solution under the actual constraints. After traversing all the selected situations, the optimal solution in the first sensitivity test is selected from all the results obtained. The same operation is then performed on the remaining process learning coefficients until all parameters have determined specific values ​​or only one parameter remains undetermined, and the optimal raw material purity solution that meets the actual needs of the user is obtained.

[0094] The specific analysis and constraints are as follows:

[0095] 1. The learning coefficient error of the multi-process learning model must be within a reasonable range compared with the traditional learning model. The specific expression is:

[0096]

[0097] In the formula, β t The formula is lnC p =β t lnP0+Vt The learning coefficient obtained by direct linear fitting, α is the set reasonable error range, which is determined by the user according to actual needs.

[0098] 2. For the two learning coefficients β in sensitivity analysis i With β j The following specific expressions must be satisfied:

[0099]

[0100] In the formula, β j is the learning coefficient of process step j, P j is the cumulative output of process step j, C ii is the purity of the product obtained when all the products produced in process i are used to manufacture perovskite photovoltaic modules, C jj It is the purity of the products obtained when all the products produced in the j process link are used to manufacture perovskite photovoltaic modules. The specific data is input by the user according to the specific situation. γ is the sensitivity error range, which is set by the user according to the specific situation.

[0101] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model, characterized in that: The following steps are involved: Collect production data from each process step in the perovskite photovoltaic production process, integrate and classify them for storage, and thus build a database; According to the production data in the database, the learning curve between the raw material purity and the cumulative output in each process link is obtained by segmented fitting, the learning coefficient of each learning curve is determined, and the learning model of multi-process links of perovskite photovoltaics is obtained by integrating them; In the perovskite photovoltaic multi-process link learning model, the learning coefficients of any two process links are selected one by one for sensitivity analysis, and the error constraint analysis of the overall perovskite photovoltaic multi-process link learning model is carried out to optimize and adjust the learning coefficients to obtain the optimal perovskite photovoltaic multi-process link learning model, thereby determining the optimal raw material purity solution.

2. The method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model according to claim 1, characterized in that: The acquisition process of the perovskite photovoltaic multi-process link learning model is specifically as follows: First, the learning curve of the first process link in perovskite photovoltaic production is fitted. Then, based on the learning curve of the first process link, the learning curves of the subsequent process links are fitted. Finally, the learning curves of each process link are combined to obtain a multi-process link learning model of perovskite photovoltaics.

3. The method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model according to claim 2, characterized in that: The expression of the learning curve of the first process link is: lnC0=β0lnP0+V0 In the formula, C0 is the purity of the input raw materials of the perovskite photovoltaic production line, P0 is the cumulative output of the input raw materials, β0 is the learning coefficient of the input raw material production, and V0 is the adjustment item of the learning curve of the first process link; The expression of the learning curve of each subsequent process link is: k i =β i lnP i +V i In the formula, i is the process link, k is i is the ratio of the raw material purity increase after the i process step compared with the previous process step, C i is the raw material purity of process step i, C i-1 is the raw material purity of process step i-1, P i is the cumulative output of process step i, β i is the learning coefficient of process step i, V i is the adjustment item of the learning curve of process step i; The expression of the perovskite photovoltaic multi-process link learning model is: lnC p =(β0+∑t i b i )lnP0+V p In the formula, C p To ensure the purity of the perovskite photovoltaic modules produced, t i V is the logarithmic ratio of the cumulative output of raw materials input into the perovskite photovoltaic production to the cumulative output of raw materials input into the i process link, p It is the overall adjustment item and the cumulative value of the adjustment items of the learning curve of each process link.

4. The method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model according to claim 3, characterized in that: The error constraint analysis is specifically as follows: The error between the learning coefficient of the overall perovskite photovoltaic multi-process link learning model and the learning coefficient obtained by fitting the traditional learning model is within a reasonable range, and the corresponding expression is: In the formula, β t The formula is lnC p =β t lnP0+V t The learning coefficient obtained by direct linear fitting, α is the set reasonable error range.

5. The method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model according to claim 3, characterized in that: The calculation expression of the sensitivity analysis is: In the formula, β j is the learning coefficient of process step j, P j is the cumulative output of process step j, C ii is the purity of the product obtained when all the products produced in process i are used to manufacture perovskite photovoltaic modules, C jj is the product purity obtained when all the products produced in process step j are used to manufacture perovskite photovoltaic modules, and γ is the sensitivity error range.

6. The method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model according to claim 1, characterized in that: The process steps of the perovskite photovoltaic production process include thin film preparation, laser etching and packaging.

7. The method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model according to claim 1, characterized in that: The database stores production data in the perovskite photovoltaic production process according to the reference data source module, the battery detail module, the hierarchical structure module, the synthesis process module and the key indicator module; The indicators of the reference data source module include: DOI number, publication date, author and sample ID; The indicators of the battery detail module include: architecture, area, flexibility, transparency and component properties; The indicators of the hierarchical structure module include: substrate, electron transport layer, perovskite layer, hole transport layer, back contact layer and additional layer; The indicators of the synthesis process module include: chemicals, solvents, deposition process, deposition and dopants / additives; The indicators of the key indicator module include: IV indicator, quality assessment indicator, stability indicator and outdoor test indicator.

8. The method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model according to claim 7, characterized in that: The indicators of the perovskite layer in the hierarchical structure module include: constituent composition, size dimension, band gap, thickness, and doping; The classification process of the constituent components is specifically as follows: The perovskite composition is written in the form of ABX3, and its raw material components are divided into A element, B element and X element; The A elements are divided into three categories: organic cations, inorganic cations and organic-inorganic phase doping; The B elements are divided into two categories: lead and non-lead; The X elements are divided into two categories: single element and multi-element doping.

9. The method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model according to claim 7, characterized in that: The hole transport layer in the hierarchical structure module specifically includes an organic hole transport system and an inorganic hole transport system; the organic hole transport system includes a triphenylamine transport material doped with Li-TFSI and tBP; a high molecular conjugated polymer doped with Li-TFSI and tBP; a branched hole transport material Fused-F without adding a dopant; a conjugated small molecule oligothiophene derivative DR3TBDTT with benzothiophene as the core, terthiophene as the arm, and rhodanine as the end cap; The inorganic hole transport system includes nickel oxide, cuprous iodide, cuprous thiocyanate, copper oxide, and copper sulfide.

10. The method for optimizing the raw material purity of a perovskite photovoltaic process based on a learning curve model according to claim 7, characterized in that: The additional layers of the hierarchical structure module include a cathode buffer layer and an anode buffer layer.

Images