Monocrystalline silicon assembly performance determination method and device, equipment and storage medium
Through multi-dimensional data evaluation and clustering algorithm, the problems of inflexible boundaries and lack of adaptability of traditional loss rate division methods are solved, and a more accurate evaluation of the performance of single crystal silicon components is achieved.
Patent Information
- Application Number
- CN202510197451.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-21
- Publication Date
- 2025-05-30
AI Technical Summary
The traditional method of loss rate division has problems such as inflexible boundaries, lack of adaptability and ignoring the relationship between data, and it is difficult to fully reflect the actual performance of single-crystal silicon battery modules.
Through multi-dimensional data evaluation, a clustering algorithm is used to cluster single-crystalline silicon component data sets to determine component levels, and a model is determined based on these data sets and grade training performance, thereby accurately evaluating component performance.
This method can more accurately reflect the performance of single crystal silicon components, consider multiple influencing factors, overcome the limitations of traditional methods, and provide a more detailed and accurate component performance evaluation.
Smart Images
Figure CN120067719A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of battery packs, and in particular, to a method, device, equipment, and storage medium for determining the performance of monocrystalline silicon components. Background Art
[0002] In the context of the global energy structure transformation and response to climate change, solar energy, as a clean and renewable energy source, has received extensive attention. Monocrystalline silicon photovoltaic cells occupy an important position in the photovoltaic market due to their high energy conversion efficiency and long lifespan. However, the performance of monocrystalline silicon cell components is not only limited by the improvement of the photoelectric conversion efficiency during the manufacturing process but also affected by packaging processes, electrical performance, optical performance, thermal performance, and long-term stability. Traditional methods for dividing the encapsulation loss rate rely on fixed intervals set manually. For example, 0 - 1%, 1% - 3%, >3%, etc. This method has the following problems: (1) Inflexible boundaries: If the encapsulation loss rate data is not evenly distributed, or the data volume of certain categories is small, the divided intervals may not accurately reflect the actual situation of the data.
[0003] (2) Lack of adaptability: Traditional division methods cannot adjust the category boundaries according to the actual distribution of the data. If there are some outliers or uneven data distribution in the dataset, the fixed intervals may cause some categories to be too large or too small, affecting the clustering effect.
[0004] (3) Ignoring the relationships between data: Traditional methods only focus on whether the data falls into a predefined interval and ignore the similarities and relationships between the data, which may result in a weak ability of the model to interpret the data.
[0005] Currently, using a single indicator of encapsulation loss rate (ELR) to evaluate the component power loss of monocrystalline silicon cells cannot comprehensively reflect the actual performance of the components and is difficult to meet the actual application requirements. Summary of the Invention
[0006] Embodiments of the present invention provide a method, device, equipment, and storage medium for determining the performance of monocrystalline silicon components, which evaluate the performance of monocrystalline silicon components through multi-dimensional data to overcome the problem of evaluating component performance using only the encapsulation loss rate traditionally.
[0007] In a first aspect, embodiments of the present invention provide a method for determining the performance of monocrystalline silicon components, including: Obtaining multiple monocrystalline silicon component datasets from the historical database; wherein, each monocrystalline silicon component dataset includes: open-circuit voltage, short-circuit current, maximum output power, maximum output power point voltage, maximum output power point current, and fill factor.
[0008] Cluster multiple datasets of monocrystalline silicon components through a clustering algorithm to obtain multiple clusters, and determine the component level corresponding to each dataset of monocrystalline silicon components according to the cluster to which each dataset of monocrystalline silicon components belongs; wherein, each cluster corresponds to a component level.
[0009] Based on each dataset of monocrystalline silicon components and the component level corresponding to each dataset of monocrystalline silicon components, train and obtain a monocrystalline silicon component performance determination model.
[0010] Input the dataset of the target monocrystalline silicon component whose component level is to be determined into the monocrystalline silicon component performance determination model to obtain the component level of the target monocrystalline silicon component.
[0011] In a possible implementation manner, training and obtaining a monocrystalline silicon component performance determination model based on each dataset of monocrystalline silicon components and the component level corresponding to each dataset of monocrystalline silicon components includes: Input each dataset of monocrystalline silicon components and the component level corresponding to each dataset of monocrystalline silicon components into multiple preset models for training respectively to obtain multiple pre-output models; wherein, the multiple preset models include a decision tree model, a random forest model, a support vector machine, a K-nearest neighbor model, and a naive Bayes classifier model.
[0012] Based on the receiver operating characteristic curve, obtain the accuracy of the multiple pre-output models.
[0013] Optimize the model with the maximum accuracy among the multiple pre-output models based on K-fold cross-validation to obtain the monocrystalline silicon component performance determination model.
[0014] In a possible implementation manner, the method further includes: Through the monocrystalline silicon component performance determination model, obtain the component levels of multiple target monocrystalline silicon components on the target production line, and obtain the sealing loss rate of all target monocrystalline silicon components in each component level.
[0015] For each component level, based on the sealing loss rate of all target monocrystalline silicon components in this component level, obtain the first sealing loss rate range.
[0016] For each cluster, based on the sealing loss rate corresponding to the dataset of monocrystalline silicon components in this cluster, obtain the second sealing loss rate range.
[0017] Successively determine whether the fluctuation value of the first sealing loss rate range under each component level relative to the second sealing loss rate range is less than a preset threshold.
[0018] If the fluctuation value of the first seal loss rate range relative to the second seal loss rate range under each component level is less than or equal to the preset threshold, the detection result of the target production line state is normal; if the fluctuation value of the first seal loss rate range relative to the second seal loss rate range under at least one component level is greater than the preset threshold, the detection result of the target production line state is abnormal.
[0019] In a possible implementation, the method further includes: Based on the decision tree algorithm, the dataset of monocrystalline silicon components of the target monocrystalline silicon component, and the component level, determine the importance coefficient of each data in the dataset of monocrystalline silicon components of the target monocrystalline silicon component.
[0020] In a possible implementation, obtain multiple datasets of monocrystalline silicon components from the historical database, including: Obtain multiple first datasets from the historical database.
[0021] Perform standardization processing and data normalization on the multiple first datasets to obtain multiple datasets of monocrystalline silicon components.
[0022] In a possible implementation, performing standardization processing and data normalization on the multiple first datasets to obtain multiple datasets of monocrystalline silicon components includes: Perform standardization processing on the multiple first datasets to obtain multiple second datasets.
[0023] For each second dataset, perform data normalization on the second dataset in combination with the first formula, the second formula, the third formula, the fourth formula, the fifth formula, and the sixth formula to obtain multiple datasets of monocrystalline silicon components.
[0024] The first formula is: VOCnorm = VOC × (1 + β × (Tref - T)) Where VOCnorm represents the open-circuit voltage after normalization, VOC represents the open-circuit voltage before normalization, β is the voltage temperature coefficient, Tref is the reference temperature, and T is the actual test temperature.
[0025] The second formula is: ISCnorm = ISC × Gref / G Where ISCnorm represents the short-circuit current after normalization, ISC represents the short-circuit current before normalization, Gref is the reference light intensity, and G is the actual test light intensity.
[0026] The third formula is: Pmaxnorm = Pmax × Gref / G × (1 + γ × (Tref - T)) Among them, Pmaxnorm represents the maximum output power after normalization, Pmax represents the maximum output power before normalization, and γ represents the power temperature coefficient.
[0027] The fourth formula is: VPMnorm = VPM × (1 + β × (Tref - T)) Among them, VPMnorm represents the voltage at the maximum power point after normalization, and VPM represents the voltage at the maximum power point before normalization; The fifth formula is: IPMnorm = IPM × Gref / G Among them, IPMnorm represents the current at the maximum power point after normalization, and IPM represents the current at the maximum power point before normalization.
[0028] The sixth formula is: FFnorm = Pmaxnorm / (VOCnorm × ISCnorm) Among them, FFnorm represents the fill factor after normalization.
[0029] In a second aspect, an embodiment of the present invention provides a device for determining the performance of a monocrystalline silicon component, including: A data acquisition module, configured to acquire multiple monocrystalline silicon component data sets from a historical database; among them, each monocrystalline silicon component data set includes: open circuit voltage, short circuit current, maximum output power, voltage at the maximum output power point, current at the maximum output power point, and fill factor.
[0030] A level discrimination module, configured to cluster the multiple monocrystalline silicon component data sets through a clustering algorithm to obtain multiple clustering clusters, and determine the component level corresponding to each monocrystalline silicon component data set according to the clustering cluster to which each monocrystalline silicon component data set belongs; among them, each clustering cluster corresponds to a component level.
[0031] A model training module, configured to train and obtain a monocrystalline silicon component performance determination model based on each monocrystalline silicon component data set and the component level corresponding to each monocrystalline silicon component data set.
[0032] A result output module, configured to input the monocrystalline silicon component data set of a target monocrystalline silicon component whose component level is to be determined into the monocrystalline silicon component performance determination model to obtain the component level of the target monocrystalline silicon component.
[0033] In a third aspect, an embodiment of the present invention provides an electronic device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, it implements the method in the first aspect or any possible implementation manner of the first aspect above.
[0034] Fourthly, an embodiment of the present invention provides a computer-readable storage medium storing a computer program, which when executed by a processor implements the method in the first aspect above or any possible implementation manner of the first aspect.
[0035] Fifthly, an embodiment of the present invention provides a computer program product including a computer program, which when executed by a processor implements the method in the first aspect above or any possible implementation manner of the first aspect.
[0036] In the embodiment of the present invention, firstly, the component grades of monocrystalline silicon components are obtained by clustering, and then training is performed according to the monocrystalline silicon component dataset and the component grades of monocrystalline silicon components to obtain a monocrystalline silicon component performance determination model. When determining the component performance of a target monocrystalline silicon component, the monocrystalline silicon component dataset of the target monocrystalline silicon component is directly input into the monocrystalline silicon component performance determination model, and the component grade of the target monocrystalline silicon component can be accurately obtained. Compared with the traditional method of solely using the encapsulation loss rate to evaluate the component grade, more dimensions (influencing factors) are considered, and the obtained results are more accurate. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] Figure 1 is a flowchart of the implementation of the monocrystalline silicon component performance determination method provided by the embodiment of the present invention; Figure 2 is a schematic structural diagram of the monocrystalline silicon component performance determination device provided by the embodiment of the present invention; Figure 3 is a schematic diagram of an electronic device provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0038] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
[0039] Refer to Figure 1 , which shows a flowchart of the implementation of the monocrystalline silicon component performance determination method provided by the embodiment of the present invention, and is described in detail as follows: Step 101, obtain multiple monocrystalline silicon component datasets from the historical database; wherein, each monocrystalline silicon component dataset includes: open circuit voltage, short circuit current, maximum output power, maximum output power point voltage, maximum output power point current, and fill factor.
[0040] In some specific embodiments, step 101 may include: Obtain multiple first datasets from the historical database.
[0041] Perform normalization processing and data normalization on the multiple first datasets to obtain multiple monocrystalline silicon component datasets.
[0042] In some specific embodiments, normalizing and data normalization are performed on multiple first data sets to obtain multiple single-crystalline silicon module data sets, which may include: Normalize multiple first data sets to obtain multiple second data sets.
[0043] For each second data set, combine the first formula, the second formula, the third formula, the fourth formula, the fifth formula, and the sixth formula to perform data normalization on the second data set to obtain multiple single-crystalline silicon module data sets.
[0044] The first formula may be: VOCnorm = VOC × (1 + β × (Tref - T)) Where VOCnorm represents the open-circuit voltage after normalization, VOC represents the open-circuit voltage before normalization, β is the voltage temperature coefficient, Tref is the reference temperature, and T is the actual test temperature.
[0045] The second formula may be: ISCnorm = ISC × Gref / G Where ISCnorm represents the short-circuit current after normalization, ISC represents the short-circuit current before normalization, Gref is the reference light intensity, and G is the actual test light intensity.
[0046] The third formula may be: Pmaxnorm = Pmax × Gref / G × (1 + γ × (Tref - T)) Where Pmaxnorm represents the maximum output power after normalization, Pmax represents the maximum output power before normalization, and γ represents the power temperature coefficient.
[0047] The fourth formula may be: VPMnorm = VPM × (1 + β × (Tref - T)) Where VPMnorm represents the voltage at the maximum output power point after normalization, and VPM represents the voltage at the maximum output power point before normalization; The fifth formula may be: IPMnorm = IPM × Gref / G Where IPMnorm represents the current at the maximum output power point after normalization, and IPM represents the current at the maximum output power point before normalization.
[0048] The sixth formula may be: FFnorm = Pmaxnorm / (VOCnorm × ISCnorm) Where FFnorm represents the fill factor after normalization.
[0049] Specifically, the standardization process can be z-score standardization, which is transformed into standardized values with a mean of 0 and a variance of 1. Performing the standardization process can ensure that there is no unit difference between different performance metrics.
[0050] Step 102: Cluster multiple single-crystalline silicon component datasets through a clustering algorithm to obtain multiple clusters, and determine the component level corresponding to each single-crystalline silicon component dataset according to the cluster to which each single-crystalline silicon component dataset belongs; wherein, each cluster corresponds to a component level.
[0051] In some specific embodiments, the clustering algorithm can be the K-means clustering algorithm. Among them, the k value of the K-means clustering algorithm generally takes 3, corresponding to three component levels (good performance, average performance, and poor performance).
[0052] Step 103: Train and obtain a single-crystalline silicon component performance determination model based on each single-crystalline silicon component dataset and the component level corresponding to each single-crystalline silicon component dataset.
[0053] In some specific embodiments, Step 103 may include: Input each single-crystalline silicon component dataset and the component level corresponding to each single-crystalline silicon component dataset into multiple preset models for training respectively to obtain multiple pre-output models; wherein, the multiple preset models include a decision tree model, a random forest model, a support vector machine, a K-nearest neighbor model, and a naive Bayes classifier model.
[0054] Based on the receiver operating characteristic curve, obtain the accuracy of multiple pre-output models.
[0055] Optimize the model with the maximum accuracy among multiple pre-output models based on K-fold cross-validation to obtain a single-crystalline silicon component performance determination model.
[0056] In some specific embodiments, evaluate the accuracy of the model through the receiver operating characteristic curve (ROC curve) and the confusion matrix, and select the model with the best performance for further optimization. Optimize the selected model through K-fold cross-validation (such as k = 10) to improve the prediction accuracy and generalization ability of the model.
[0057] Step 104: Input the single-crystalline silicon component dataset of the target single-crystalline silicon component whose component level is to be determined into the single-crystalline silicon component performance determination model to obtain the component level of the target single-crystalline silicon component.
[0058] In some specific embodiments, the method further includes: Through the single-crystalline silicon component performance determination model, obtain the component levels of multiple target single-crystalline silicon components on the target production line, and obtain the sealing loss rate of all target single-crystalline silicon components in each component level.
[0059] For each component level, a first seal loss rate range is obtained based on the seal loss rates of all target monocrystalline silicon components in that component level.
[0060] For each cluster, a second seal loss rate range is obtained based on the seal loss rate corresponding to the dataset of monocrystalline silicon components in that cluster.
[0061] Successively determine whether the fluctuation value of the first seal loss rate range under each component level relative to the second seal loss rate range is less than a preset threshold.
[0062] If the fluctuation value of the first seal loss rate range under each component level relative to the second seal loss rate range is less than or equal to the preset threshold, the detection result of the target production line status is normal; if the fluctuation value of the first seal loss rate range under at least one component level relative to the second seal loss rate range is greater than the preset threshold, the detection result of the target production line status is abnormal.
[0063] In some specific embodiments, the method further includes: Through the monocrystalline silicon component performance determination model, obtain the component levels of multiple target monocrystalline silicon components on the target production line.
[0064] Based on the component levels of each target monocrystalline silicon component, obtain the first component level ratio; Based on the number of datasets of monocrystalline silicon components in each cluster, obtain the second component level ratio; Judge whether the fluctuation value of the first component level ratio relative to the second component level ratio is less than the preset threshold.
[0065] If the fluctuation value of the first component level ratio relative to the second component level ratio is less than or equal to the preset threshold, the detection result of the target production line status is normal; If the fluctuation value of the first component level ratio relative to the second component level ratio is greater than the preset threshold, the detection result of the target production line status is abnormal.
[0066] In some specific embodiments, the method further includes: Through the monocrystalline silicon component performance determination model, obtain the component levels of multiple target monocrystalline silicon components on the target production line, and obtain the seal loss rates of all target monocrystalline silicon components in each component level.
[0067] Divide the multiple target monocrystalline silicon components into a first target group and a second target group in chronological order; among them, the acquisition time of the time label of each target monocrystalline silicon component in the first target group is before the acquisition time of the time label of each target monocrystalline silicon component in the second target group.
[0068] In the first target group, for each component level, a third seal damage rate range is obtained based on the seal damage rates of all target monocrystalline silicon components at that component level.
[0069] In the second target group, for each component level, a fourth seal damage rate range is obtained based on the seal damage rates of all target monocrystalline silicon components at that component level.
[0070] Successively determine whether the fluctuation value of the third seal damage rate range relative to the fourth seal damage rate range for each component level is less than a preset threshold.
[0071] If the fluctuation value of the third seal damage rate range relative to the fourth seal damage rate range for each component level is less than or equal to the preset threshold, the detection result of the target production line status is normal; if the fluctuation value of the third seal damage rate range relative to the fourth seal damage rate range for at least one component level is greater than the preset threshold, the detection result of the target production line status is abnormal.
[0072] Specifically, the following uses a specific embodiment to illustrate that "the acquisition time of the time tag of each target monocrystalline silicon component in the first target group is before the acquisition time of the time tag of each target monocrystalline silicon component in the second target group": Suppose there are 100 target monocrystalline silicon components on the target production line. These 100 target monocrystalline silicon components are determined by the monocrystalline silicon component performance determination model according to their six characteristic indicators (open circuit voltage, short circuit current, maximum output power, maximum output power point voltage, maximum output power point current, and fill factor) to obtain the component level (at this time, the target monocrystalline silicon component will obtain a time tag for obtaining the component level). Since there is a sequence in this process, there will be a difference in the acquisition time of the time tag at this time. In short, the target monocrystalline silicon components in the second target group can be the latter 50 target monocrystalline silicon components among these 100 target monocrystalline silicon components that obtain the component level through the monocrystalline silicon component performance determination model.
[0073] In some specific embodiments, the method further includes: Obtain the component levels of multiple target monocrystalline silicon components on the target production line through the monocrystalline silicon component performance determination model.
[0074] Divide the multiple target monocrystalline silicon components into a first target group and a second target group in chronological order; among them, the acquisition time of the time tag of each target monocrystalline silicon component in the first target group is before the acquisition time of the time tag of each target monocrystalline silicon component in the second target group.
[0075] Based on the component levels corresponding to all target monocrystalline silicon components in the first target group, obtain the third component level ratio.
[0076] Based on the component grades corresponding to all the target monocrystalline silicon components in the second target group, obtain the fourth component grade ratio.
[0077] Determine whether the fluctuation value of the third component grade ratio relative to the fourth component grade ratio is less than a preset threshold.
[0078] If the fluctuation value of the third component grade ratio relative to the fourth component grade ratio is less than or equal to the preset threshold, the detection result of the target production line status is normal; if the fluctuation value of the first sealing loss rate range relative to the second sealing loss rate range under at least one component grade is greater than the preset threshold, the detection result of the target production line status is abnormal.
[0079] In some specific embodiments, the method further includes: Based on the decision tree algorithm, the monocrystalline silicon component dataset of the target monocrystalline silicon components, and the component grades, determine the importance coefficient of each data in the monocrystalline silicon component dataset of the target monocrystalline silicon components.
[0080] Specifically, when it is detected that the status of the target production line is abnormal, the target production line needs to be inspected in a timely manner. However, when inspecting the target production line, it is necessary to distinguish priorities. Distinguishing priorities can improve the inspection efficiency, ensure that faults are discovered and processed in a timely manner. Therefore, it is necessary to calculate through the decision tree algorithm the importance coefficient of each of the six indicators (open circuit voltage, short circuit current, maximum output power, maximum output power point voltage, maximum output power point current, and fill factor) of the target monocrystalline silicon components produced by the target production line at this time, and obtain an inspection order for the target production line according to the importance coefficient.
[0081] For example, if the open circuit voltage is the indicator with the largest importance coefficient, then when inspecting the target production line, the part related to affecting the open circuit voltage of the monocrystalline silicon components (such as the welding machine, laminator, etc.) can be inspected preferentially.
[0082] For the above monocrystalline silicon component performance determination method, first obtain the component grades of the monocrystalline silicon components through clustering, and then train according to the monocrystalline silicon component dataset and the component grades of the monocrystalline silicon components to obtain a monocrystalline silicon component performance determination model. When determining the component performance of the target monocrystalline silicon components, directly input the monocrystalline silicon component dataset of the target monocrystalline silicon components into the monocrystalline silicon component performance determination model, and the component grades of the target monocrystalline silicon components can be accurately obtained. Compared with the traditional method of solely using the encapsulation loss rate to evaluate the component grades, more dimensions (influence factors) are considered, and the obtained results are more accurate.
[0083] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.
[0084] The following is an apparatus embodiment of the present invention. For details not described in detail, reference may be made to the corresponding method embodiment above.
[0085] Figure 2 FIG. shows a schematic structural diagram of a single-crystalline silicon component performance determination apparatus provided by an embodiment of the present invention. For ease of description, only parts related to the embodiment of the present invention are shown and are described in detail as follows: As Figure 2 shown, the single-crystalline silicon component performance determination apparatus includes: A data acquisition module 201, configured to acquire a plurality of single-crystalline silicon component data sets from a historical database; wherein, each single-crystalline silicon component data set includes: open-circuit voltage, short-circuit current, maximum output power, maximum output power point voltage, maximum output power point current, and fill factor.
[0086] A grade discrimination module 202, configured to cluster the plurality of single-crystalline silicon component data sets through a clustering algorithm to obtain a plurality of clustering clusters, and determine the component grade corresponding to each single-crystalline silicon component data set according to the clustering cluster to which each single-crystalline silicon component data set belongs; wherein, each clustering cluster corresponds to a component grade.
[0087] A model training module 203, configured to train and obtain a single-crystalline silicon component performance determination model based on each single-crystalline silicon component data set and the component grade corresponding to each single-crystalline silicon component data set.
[0088] A result output module 204, configured to input the single-crystalline silicon component data set of a target single-crystalline silicon component whose component grade is to be determined into the single-crystalline silicon component performance determination model to obtain the component grade of the target single-crystalline silicon component.
[0089] In a possible implementation manner, the model training module 203 is configured to: Input each single-crystalline silicon component data set and the component grade corresponding to each single-crystalline silicon component data set into a plurality of preset models for training respectively to obtain a plurality of pre-output models; wherein, the plurality of preset models include a decision tree model, a random forest model, a support vector machine, a K-nearest neighbor model, and a naive Bayes classifier model.
[0090] Obtain the accuracy of the plurality of pre-output models based on the receiver operating characteristic curve.
[0091] Optimize the model with the maximum accuracy among the plurality of pre-output models based on K-fold cross-validation to obtain the single-crystalline silicon component performance determination model.
[0092] In a possible implementation manner, the result output module 204 may further be configured to: Through the performance determination model of monocrystalline silicon components, obtain the component grades of multiple target monocrystalline silicon components on the target production line, and obtain the encapsulation loss rate of all target monocrystalline silicon components in each component grade.
[0093] For each component grade, based on the encapsulation loss rate of all target monocrystalline silicon components in this component grade, obtain the first encapsulation loss rate range.
[0094] For each clustering cluster, based on the encapsulation loss rate corresponding to the monocrystalline silicon component data set in this clustering cluster, obtain the second encapsulation loss rate range.
[0095] Successively determine whether the fluctuation value of the first encapsulation loss rate range under each component grade relative to the second encapsulation loss rate range is less than the preset threshold.
[0096] If the fluctuation value of the first encapsulation loss rate range under each component grade relative to the second encapsulation loss rate range is less than or equal to the preset threshold, the detection result of the target production line state is normal; if the fluctuation value of the first encapsulation loss rate range under at least one component grade relative to the second encapsulation loss rate range is greater than the preset threshold, the detection result of the target production line state is abnormal.
[0097] In a possible implementation manner, the result output module 204 can also be used for: Based on the decision tree algorithm, the monocrystalline silicon component data set of the target monocrystalline silicon component, and the component grade, determine the importance coefficient of each data in the monocrystalline silicon component data set of the target monocrystalline silicon component.
[0098] In a possible implementation manner, the data acquisition module 201 can be used for: Obtain multiple first data sets from the historical database.
[0099] Perform standardization processing and data normalization on the multiple first data sets to obtain multiple monocrystalline silicon component data sets.
[0100] In a possible implementation manner, the data acquisition module 201 can be used for: Perform standardization processing on the multiple first data sets to obtain multiple second data sets.
[0101] For each second data set, perform data normalization on this second data set in combination with the first formula, the second formula, the third formula, the fourth formula, the fifth formula, and the sixth formula to obtain multiple monocrystalline silicon component data sets.
[0102] The first formula can be: VOCnorm = VOC × (1 + β × (Tref - T)) Among them, VOCnorm represents the open-circuit voltage after normalization, VOC represents the open-circuit voltage before normalization, β is the voltage temperature coefficient, Tref is the reference temperature, and T is the actual test temperature.
[0103] The second formula can be: ISCnorm = ISC × Gref / G Among them, ISCnorm represents the short-circuit current after normalization, ISC represents the short-circuit current before normalization, Gref is the reference light intensity, and G is the actual test light intensity.
[0104] The third formula can be: Pmaxnorm = Pmax × Gref / G × (1 + γ × (Tref - T)) Among them, Pmaxnorm represents the maximum output power after normalization, Pmax represents the maximum output power before normalization, and γ represents the power temperature coefficient.
[0105] The fourth formula can be: VPMnorm = VPM × (1 + β × (Tref - T)) Among them, VPMnorm represents the voltage at the maximum output power point after normalization, and VPM represents the voltage at the maximum output power point before normalization; The fifth formula can be: IPMnorm = IPM × Gref / G Among them, IPMnorm represents the current at the maximum output power point after normalization, and IPM represents the current at the maximum output power point before normalization.
[0106] The sixth formula can be: FFnorm = Pmaxnorm / (VOCnorm × ISCnorm) Among them, FFnorm represents the fill factor after normalization.
[0107] Figure 3 It is a schematic diagram of the electronic device provided by the embodiment of the present invention. As Figure 3 shown, the electronic device 5 of this embodiment includes: a processor 50 and a memory 51. The memory 51 stores a computer program 52. When the processor 50 executes the computer program 52, the steps in the above-mentioned method embodiments are implemented. Alternatively, when the processor 50 executes the computer program 52, the functions of each module / unit in the above-mentioned device embodiments are implemented.
[0108] Exemplarily, the computer program 52 can be divided into one or more modules / units, which are stored in the memory 51 and executed by the processor 50 to implement the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program 52 in the electronic device 5.
[0109] The electronic device 5 may include, but is not limited to, the processor 50 and the memory 51. Those skilled in the art can understand that Figure 3 merely being examples of the electronic device 5, they do not constitute a limitation to the electronic device 5. It may include more or fewer components than those shown in the figure, or combine certain components, or different components. For example, the electronic device 5 may further include input / output devices, network access devices, buses, etc.
[0110] The processor 50 can be a central processing unit (CPU), or can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.
[0111] The memory 51 can be an internal storage unit of the electronic device 5, such as the hard disk or memory of the electronic device 5. The memory 51 can also be an external storage device of the electronic device 5, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 5. Further, the memory 51 can also include both the internal storage unit and the external storage device of the electronic device 5. The memory 51 is used to store the computer program 52 and other programs and data required by the electronic device 5. The memory 51 can also be used to temporarily store the data that has been output or will be output.
[0112] For the convenience and brevity of description, only the above division of each functional module / unit is used as an example. In actual applications, the above functions can be allocated to different functional modules / units according to needs. The above modules / units can be implemented in the form of hardware, or in the form of software, or in the form of a combination of hardware and software.
[0113] An embodiment of the present invention also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the methods in the above method embodiments are implemented.
[0114] An embodiment of the present invention also provides a computer program product including a computer program. When the computer program is executed by a processor, the methods in the above method embodiments are implemented.
[0115] Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.
[0116] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For parts not detailed or recorded in a certain embodiment, reference may be made to the relevant descriptions of other embodiments. Without special instructions and logical conflicts, the terms and / or descriptions between different embodiments are consistent and can be mutually referred to, and the technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationships.
[0117] The above-described embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.
Claims
1. A method for determining the performance of a monocrystalline silicon module, characterized in that: include: Acquire multiple monocrystalline silicon component data sets from a historical database; wherein each monocrystalline silicon component data set includes: open circuit voltage, short circuit current, maximum output power, maximum output power point voltage, maximum output power point current and fill factor; Clustering the plurality of monocrystalline silicon component data sets by a clustering algorithm to obtain a plurality of clustering clusters, and determining the component level corresponding to each monocrystalline silicon component data set according to the clustering cluster to which each monocrystalline silicon component data set belongs; wherein each clustering cluster corresponds to a component level; Based on each monocrystalline silicon component data set and the component level corresponding to each monocrystalline silicon component data set, a monocrystalline silicon component performance determination model is trained and obtained; The monocrystalline silicon component data set of the target monocrystalline silicon component whose component grade is to be determined is input into the monocrystalline silicon component performance determination model to obtain the component grade of the target monocrystalline silicon component.
2. The method for determining the performance of a single crystal silicon module according to claim 1, characterized in that: The method of training and obtaining a single crystal silicon component performance determination model based on each single crystal silicon component data set and a component level corresponding to each single crystal silicon component data set includes: Input each monocrystalline silicon component data set and the component level corresponding to each monocrystalline silicon component data set into multiple preset models for training respectively, to obtain multiple pre-output models; wherein the multiple preset models include a decision tree model, a random forest model, a support vector machine, a K nearest neighbor model and a naive Bayes classifier model; Based on the receiver operating characteristic curve, the accuracy of multiple pre-output models was obtained; The model with the highest accuracy among multiple pre-output models is optimized based on K-fold cross validation to obtain the single crystal silicon component performance determination model.
3. The method for determining the performance of a single crystal silicon module according to claim 1, characterized in that: The method further comprises: Obtaining the component grades of multiple target monocrystalline silicon components on a target production line through the monocrystalline silicon component performance determination model, and obtaining the sealing loss rate of all target monocrystalline silicon components in each component grade; For each component level, based on the sealing damage rates of all target monocrystalline silicon components in the component level, a first sealing damage rate range is obtained; For each cluster, based on the sealing failure rate corresponding to the single crystal silicon component data set in the cluster, a second sealing failure rate range is obtained; determining in sequence whether a fluctuation value of the first sealing damage rate range relative to the second sealing damage rate range at each component level is less than a preset threshold; If the fluctuation value of the first sealing damage rate range relative to the second sealing damage rate range at each component level is less than or equal to a preset threshold, the detection result of the target production line status is normal; If the fluctuation value of the first sealing damage rate range relative to the second sealing damage rate range at at least one component level is greater than the preset threshold, the detection result of the target production line status is abnormal.
4. The method for determining the performance of a single crystal silicon module according to claim 1, characterized in that: The method further comprises: Based on a decision tree algorithm, the single crystal silicon component data set and the component level of the target single crystal silicon component, an importance coefficient of each data in the single crystal silicon component data set of the target single crystal silicon component is determined.
5. The method for determining the performance of a single crystal silicon module according to claim 1, characterized in that: The step of obtaining a plurality of monocrystalline silicon component data sets from a historical database includes: Acquire a plurality of first data sets from a historical database; The plurality of first data sets are subjected to standardization processing and data normalization to obtain a plurality of single crystal silicon component data sets.
6. The method for determining the performance of a single crystal silicon module according to claim 5, characterized in that: The step of performing standardization and data normalization on the plurality of first data sets to obtain a plurality of single crystal silicon component data sets includes: Performing standardization processing on the multiple first data sets to obtain multiple second data sets; For each second data set, normalize the second data set by combining the first formula, the second formula, the third formula, the fourth formula, the fifth formula and the sixth formula to obtain a plurality of single crystal silicon component data sets; The first formula is: VOCnorm=VOC×(1+β×(Tref-T)) Wherein, VOCnorm represents the open circuit voltage after normalization, VOC represents the open circuit voltage before normalization, β is the voltage temperature coefficient, Tref is the reference temperature, and T is the actual test temperature; The second formula is: ISCnorm=ISC×Gref / G Among them, ISCnorm represents the normalized short-circuit current, ISC represents the short-circuit current before normalization, Gref represents the reference light intensity, and G represents the actual test light intensity; The third formula is: Pmaxnorm=Pmax×Gref / G×(1+γ×(Tref-T)) Wherein, Pmaxnorm represents the maximum output power after normalization, Pmax represents the maximum output power before normalization, and γ represents the power temperature coefficient; The fourth formula is: VPMnorm=VPM×(1+β×(Tref-T)) Wherein, VPMnorm represents the maximum output power point voltage after normalization, and VPM represents the maximum output power point voltage before normalization; The fifth formula is: IPMnorm=IPM×Gref / G Where, IPMnorm represents the maximum output power point current after normalization, and IPM represents the maximum output power point current before normalization; The sixth formula is: FFnorm=Pmaxnorm / (VOCnorm×ISCnorm) Among them, FFnorm represents the normalized filling factor.
7. A device for determining the performance of a single crystal silicon module, characterized in that: include: A data acquisition module is used to acquire multiple monocrystalline silicon component data sets from a historical database; wherein each monocrystalline silicon component data set includes: open circuit voltage, short circuit current, maximum output power, maximum output power point voltage, maximum output power point current and fill factor; A level discrimination module, used for clustering the plurality of monocrystalline silicon component data sets through a clustering algorithm to obtain a plurality of clustering clusters, and determining the component level corresponding to each monocrystalline silicon component data set according to the clustering cluster to which each monocrystalline silicon component data set belongs; wherein each clustering cluster corresponds to a component level; A model training module, used for training and obtaining a single crystal silicon component performance determination model based on each single crystal silicon component data set and a component level corresponding to each single crystal silicon component data set; The result output module is used to input the monocrystalline silicon component data set of the target monocrystalline silicon component whose component grade is to be determined into the monocrystalline silicon component performance determination model to obtain the component grade of the target monocrystalline silicon component.
8. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 6 when executing the computer program.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.
10. A computer program product, characterized in that The method comprises a computer program, which implements the method according to any one of claims 1 to 6 when being executed by a processor.