Self-adaptive learning system realized by fusing IV detection and defect detection

Abstract

The invention discloses a self-adaptive learning system realized by fusing IV detection and defect detection in the field of self-adaptive learning, which comprises the following steps of: S1, inputting a photovoltaic module design model, inputting a designed photovoltaic module model and environmental parameters of a photovoltaic module mounting position into the system, and inputting initial parameters for a system algorithm; according to the scheme, a small light transmittance detection device, an IV detection system and a defect detection system are combined with power data fed back by an IV detection and light transmittance detection technology system and a qualified power condition of a product of the model is input to carry out self-adaptive learning on an over-parameter IoU, confidence and the like in defect detection; the model is intelligently adjusted so that the model can adapt to various different personalized customizations, distinguishing and personalized optimization can be conducted according to products of different gears, and intelligence and generalization are improved.

Description

Technical Field

[0001] The present invention relates to the technical field, and in particular to an adaptive learning system that integrates IV detection and defect detection. Background Art

[0002] With the popularity of photovoltaic power generation, photovoltaic system designers are increasingly in need of a tool that can predict the power generated by photovoltaic modules. Photovoltaic module modeling can not only provide designers with a reference tool for evaluating module performance to assist their decision-making, but also has vital significance for the simulation, design, evaluation, control and optimization of photovoltaic systems. Currently, the single diode model derived from the module material is widely used, including the ideal model, 4-parameter model and 5-parameter model. Among them, the 5-parameter model has the highest accuracy. Compared with the ideal model and the 4-parameter model, the equivalent series-parallel resistance is added to ensure the integrity of the physical structure of the single diode model. It is suitable for simulating photovoltaic modules composed of multiple battery cells, or photovoltaic arrays composed of multiple photovoltaic modules.

[0003] The existing photovoltaic module simulation algorithms have the following problems:

[0004] 1. A larger site is required to accommodate IV testing and defect detection.

[0005] 2. When new defects appear, a new model must be trained. Model training takes a long time and cannot be used directly for new products. When new defects are discovered, a completely new model must be trained for testing. Summary of the invention

[0006] The purpose of this section is to summarize some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of the present invention to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0007] The purpose of the present invention is to address the technical problems existing in the background technology. The present invention proposes an adaptive learning system that integrates IV detection and defect detection. This scheme adaptively learns the hyper-parameters IoU and confidence in defect detection by combining small-scale transmittance detection equipment, IV detection system and defect detection system with power data fed back by IV detection and transmittance detection technology systems and inputting qualified power conditions of products of this model. The intelligent adjustment model enables the model to adapt to a variety of different personalized customizations, and can make distinctions and personalized optimizations for products of different gears, thereby improving intelligence and generalization.

[0008] The present invention proposes an adaptive learning system that integrates IV detection and defect detection, including the following steps:

[0009] S1 inputs the photovoltaic module design model step, inputs the designed photovoltaic module model and the environmental parameters of the photovoltaic module installation location into the system, and inputs the initial parameters for the system algorithm;

[0010] S2 performs IV testing on the model, and simulates and calculates the design model and the installation environment by using the IV prediction algorithm through the input initial data, so as to obtain the predicted lighting efficiency of the photovoltaic module in the target installation environment;

[0011] S3 performs defect detection on the model, performs structural detection on the design model alone, compares the size of the design model with the actual installation requirements, and detects design defects or unrealizable structural positions;

[0012] S4 performs a light transmittance detection step on the model, assembles a miniature design model through the photovoltaic module design model, uses a small light transmittance detection device to perform light transmittance detection on the miniature design model, and summarizes the light transmittance results of the miniature design model, thereby calculating the light transmittance data of the photovoltaic module model of the original size, and then calculating the lighting efficiency of the photovoltaic module model of the original size;

[0013] S5 proposes improvement and adjustment steps for the model. By combining the structural defects detected by the model with the IV prediction algorithm for the model, the algorithm is used to propose improvement plans for the structural defects, and the daylighting efficiency of the improved model is predicted again, and the daylighting data of the prediction is statistically analyzed;

[0014] S6 is a step of comparing IV data with test data, comparing the IV prediction result with the experimental data tested by a small light transmittance test instrument, so as to analyze the values ​​with deviations in the prediction algorithm and the reasons for the errors between the algorithm and the actual test data;

[0015] S7 is a correction step for the IV algorithm. By comparing the experimental data with the predicted data, the calculation logic in the algorithm is adjusted in reverse, so that the results output by the algorithm are more in line with the actual experimental data.

[0016] S8 is an adaptive model correction step, in which the design model corrected by the algorithm is input into the system as the initial data for calculation and analysis again, so as to realize the adaptive iterative correction effect of the system, and the design defects in the design model can be gradually reduced by analyzing and correcting the design model multiple times;

[0017] S9 is an adaptive algorithm correction step, in which the algorithm after comparison and correction is used again as the IV prediction algorithm of this system, so that the prediction algorithm can be iteratively corrected during the repeated iteration of the design model.

[0018] By adopting the above technical solution, this solution can achieve the effect of adaptive adjustment through automatic iterative correction algorithm and correction model.

[0019] Preferably, in step S1, the installation environment parameters include the environment height and the sunshine time and angle, and the installation environment parameters are improved by adding the specific installation direction and installation method of the photovoltaic components.

[0020] By adopting the above technical solution, this solution can correct and supplement the prediction results by supplementing the environmental parameters, and more complete environmental parameters can ensure the accuracy of the prediction results.

[0021] Preferably, in step S2, a predicted light conversion efficiency of the photovoltaic module design model is obtained in an IV prediction algorithm by performing a prediction operation on the design model. The prediction algorithm combines the size model of the photovoltaic module model with the installation environment parameters to derive the prediction data.

[0022] By adopting the above technical solution, the present solution corrects the detailed parameters of the photovoltaic components through the prediction algorithm, thereby avoiding the influence of the position parameters on the prediction results, so that the present system has a hyperparameter effect.

[0023] Preferably, in the step S4, a miniature model of equal scale is produced according to the parameters of the photovoltaic module design model, and then the photovoltaic module design model is tested by a non-predictive algorithm. By setting a small transmittance detection device at different distances from the photovoltaic module miniature model, the effect of the photovoltaic module's light transmittance under different lighting environments can be simulated.

[0024] By adopting the above technical solution, the present solution corrects the algorithm through data obtained by the non-predictive algorithm to avoid repeated iterations of erroneous parameters.

[0025] Preferably, in step S3, the algorithm for performing defect detection on the structure of the photovoltaic module is independent of the IV prediction algorithm, the defect detection algorithm does not change during the iterative use process, and the defect detection algorithm only includes the detection of the basic logic of the photovoltaic module structure.

[0026] By adopting the above technical solution, this solution can ensure the logical smoothness of the basic structure of the photovoltaic module through an independent defect detection algorithm.

[0027] Preferably, in step S5, the defective partial structure is analyzed specifically by using an IV prediction algorithm, and an improved structure corresponding to the defective part can be found from the previously predicted structural data model, thereby correcting the defective structure.

[0028] By adopting the above technical solution, this solution summarizes the previously predicted design model and corrects the defective parts.

[0029] Preferably, in the step S7, the prediction algorithm is corrected by a reverse recursive method, with the measured data as a fixed result and the design model as the input initial data, so that the algorithm can adjust and fit itself according to the fixed input and output results, and multiple iterations are used. Through multiple fittings, the internal parameters of the algorithm can gradually fit the actual situation.

[0030] By adopting the above technical solution, this solution avoids the error existing in single correction prediction through the effect of multiple iterations, so that the final result is more consistent with the actual situation.

[0031] Preferably, in step S8, the revised design model is re-entered as input data into the internal algorithm of the system to achieve an iterative effect. During the iterative process, both the design model and the prediction algorithm are adaptively adjusted.

[0032] By adopting the above technical solution, the present solution can achieve synchronous progress by adaptively adjusting the algorithm while the design model is changing.

[0033] In summary, the present invention includes at least one of the following beneficial effects:

[0034] The design method of this system can improve the accuracy of the prediction of the lighting efficiency of photovoltaic modules after actual installation, and the system has the function of automatically improving the design scheme, so that it can automatically correct the design defects in the design model, thereby improving the lighting efficiency of the final model. The system is corrected with a miniature model, thereby improving the accuracy of the algorithm while reducing the actual production cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0036] Figure 1 A front view of an embodiment of the present invention that integrates IV detection and defect detection to realize an adaptive learning system; DETAILED DESCRIPTION

[0037] The following is combined with Figure 1 The present invention is described in further detail.

[0038] Embodiment 1

[0039] like Figure 1 As shown, in order to solve the existing problems, in this embodiment, the present invention discloses an adaptive learning system that integrates IV detection and defect detection, including the following steps:

[0040] S1 inputs the photovoltaic module design model step, inputs the designed photovoltaic module model and the environmental parameters of the photovoltaic module installation location into the system, and inputs the initial parameters for the system algorithm;

[0041] S2 performs IV testing on the model, and simulates and calculates the design model and installation environment by using the IV prediction algorithm through the input initial data, so as to obtain the predicted lighting efficiency of the photovoltaic module in the target installation environment;

[0042] S3 performs defect detection on the model, performs structural detection on the design model alone, compares the size of the design model with the actual installation requirements, and detects design defects or unrealizable structural positions;

[0043] S4 performs a light transmittance detection step on the model, assembles a miniature design model through the photovoltaic module design model, uses a small light transmittance detection device to perform light transmittance detection on the miniature design model, and summarizes the light transmittance results of the miniature design model, thereby calculating the light transmittance data of the photovoltaic module model of the original size, and then calculating the lighting efficiency of the photovoltaic module model of the original size;

[0044] S5 proposes improvement and adjustment steps for the model. By combining the structural defects detected by the model with the IV prediction algorithm for the model, the algorithm is used to propose improvement plans for the structural defects, and the daylighting efficiency of the improved model is predicted again, and the daylighting data of the prediction is statistically analyzed;

[0045] S6 is a step of comparing IV data with test data, comparing the IV prediction result with the experimental data tested by a small light transmittance test instrument, so as to analyze the values ​​with deviations in the prediction algorithm and the reasons for the errors between the algorithm and the actual test data;

[0046] S7 is a correction step for the IV algorithm. By comparing the experimental data with the predicted data, the calculation logic in the algorithm is adjusted in reverse, so that the results output by the algorithm are more in line with the actual experimental data.

[0047] S8 is an adaptive model correction step, in which the design model corrected by the algorithm is input into the system as the initial data for calculation and analysis again, so as to realize the adaptive iterative correction effect of the system, and the design defects in the design model can be gradually reduced by analyzing and correcting the design model multiple times;

[0048] S9 is an adaptive algorithm correction step, in which the algorithm after comparison and correction is used again as the IV prediction algorithm of this system, so that the prediction algorithm can be iteratively corrected during the repeated iteration of the design model.

[0049] This system can realize the prediction of the light conversion rate of photovoltaic modules and can also correct and improve its own prediction algorithm. Therefore, compared with the traditional prediction algorithm, this system has a higher degree of fitting and more accurate prediction results. The combination of algorithm prediction and transmittance experimental instrument detection can effectively avoid the cyclic iteration of erroneous data, thereby ensuring that the data is always supported by real experimental data during multiple iterations.

[0050] Embodiment 2

[0051] like Figure 1 As shown, in order to solve the existing problems, in this embodiment, based on the same concept as the above-mentioned embodiment 1, the adaptive learning system realized by integrating IV detection and defect detection also includes: in the step S1, the installation environment parameters include the environmental height and the sunshine time and angle, and the installation environment parameters are improved by increasing the specific installation direction and installation method of the photovoltaic components.

[0052] In the step S2, a predicted light conversion efficiency of the photovoltaic module design model is obtained in the IV prediction algorithm by performing a prediction operation on the design model. The prediction algorithm combines the size model of the photovoltaic module model with the installation environment parameters to derive the prediction data.

[0053] In the step S4, a miniature model of equal scale is produced according to the parameters of the photovoltaic module design model, and then the photovoltaic module design model is tested by a non-predictive algorithm. By setting a small transmittance detection device at different distances from the photovoltaic module miniature model, the effect of the photovoltaic module's light transmittance under different light environments can be simulated.

[0054] In the step S3, the algorithm for performing defect detection on the structure of the photovoltaic module is independent of the IV prediction algorithm. The defect detection algorithm does not change during the iterative use process. The defect detection algorithm only includes the detection of the basic logic of the photovoltaic module structure.

[0055] In the step S5, the defective part of the structure is analyzed specifically by the IV prediction algorithm, and the improved structure corresponding to the defective part can be found from the previously predicted structural data model, thereby correcting the defective structure.

[0056] In the step S7, the prediction algorithm is modified by a reverse recursive method, with the measured data as a fixed result and the design model as the input initial data, so that the algorithm can adjust and fit itself according to the fixed input and output results, and multiple iterations are used. Through multiple fittings, the internal parameters of the algorithm can gradually fit the actual situation.

[0057] The step S8 achieves an iterative effect by re-entering the modified design model as input data into the internal algorithm of the system. During the iterative process, both the design model and the prediction algorithm are adaptively adjusted.

[0058] Compared with traditional algorithms, this system has the function of adaptive correction. Compared with traditional fixed prediction algorithms, which only have fixed output results for single input data, this system can perform self-correction after multiple uses, thereby improving the accuracy of this system.

[0059] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An adaptive learning system is implemented by integrating IV detection and defect detection, comprising the following steps: S1 inputs the photovoltaic module design model step, inputs the designed photovoltaic module model and the environmental parameters of the photovoltaic module installation location into the system, and inputs the initial parameters for the system algorithm; S2 performs IV testing on the model, and simulates and calculates the design model and the installation environment by using the IV prediction algorithm through the input initial data, so as to obtain the predicted lighting efficiency of the photovoltaic module in the target installation environment; S3 performs defect detection on the model, performs structural detection on the design model alone, compares the size of the design model with the actual installation requirements, and detects design defects or unrealizable structural positions; S4 performs a light transmittance detection step on the model, assembles a miniature design model through the photovoltaic module design model, uses a small light transmittance detection device to perform light transmittance detection on the miniature design model, and summarizes the light transmittance results of the miniature design model, thereby calculating the light transmittance data of the photovoltaic module model of the original size, and then calculating the lighting efficiency of the photovoltaic module model of the original size; S5 proposes improvement and adjustment steps for the model. By combining the structural defects detected by the model with the IV prediction algorithm for the model, the algorithm is used to propose improvement plans for the structural defects, and the daylighting efficiency of the improved model is predicted again, and the daylighting data of the prediction is statistically analyzed; S6 is a step of comparing IV data with test data, comparing the IV prediction result with the experimental data tested by a small light transmittance test instrument, so as to analyze the values ​​with deviations in the prediction algorithm and the reasons for the errors between the algorithm and the actual test data; S7 is a correction step for the IV algorithm. By comparing the experimental data with the predicted data, the calculation logic in the algorithm is adjusted in reverse, so that the results output by the algorithm are more in line with the actual experimental data. S8 is an adaptive model correction step, in which the design model corrected by the algorithm is input into the system as the initial data for calculation and analysis again, so as to realize the adaptive iterative correction effect of the system, and the design defects in the design model can be gradually reduced by analyzing and correcting the design model multiple times; S9 is an adaptive algorithm correction step, in which the algorithm after comparison and correction is used again as the IV prediction algorithm of this system, so that the prediction algorithm can be iteratively corrected during the repeated iteration of the design model.

2. According to claim 1, a self-adaptive learning system is realized by integrating IV detection and defect detection, characterized in that: In the step S1, the installation environment parameters include the environment height and the sunshine time and angle, and the installation environment parameters are improved by adding the specific installation direction and installation method of the photovoltaic components.

3. According to claim 1, a self-adaptive learning system is realized by integrating IV detection and defect detection, characterized in that: In the step S2, a predicted light conversion efficiency of the photovoltaic module design model is obtained in the IV prediction algorithm by performing a prediction operation on the design model. The prediction algorithm combines the size model of the photovoltaic module model with the installation environment parameters to derive the prediction data.

4. According to claim 1, a self-adaptive learning system is realized by integrating IV detection and defect detection, characterized in that: In the step S4, a miniature model of equal scale is produced according to the parameters of the photovoltaic module design model, and then the photovoltaic module design model is tested by a non-predictive algorithm. By setting a small transmittance detection device at different distances from the photovoltaic module miniature model, the effect of the photovoltaic module's light transmittance under different light environments can be simulated.

5. According to claim 1, a self-adaptive learning system is realized by integrating IV detection and defect detection, characterized in that: In the step S3, the algorithm for performing defect detection on the structure of the photovoltaic module is independent of the IV prediction algorithm. The defect detection algorithm does not change during the iterative use process. The defect detection algorithm only includes the detection of the basic logic of the photovoltaic module structure.

6. The adaptive learning system realized by integrating IV detection and defect detection according to claim 1, characterized in that: In the step S5, the defective part of the structure is analyzed specifically by the IV prediction algorithm, and the improved structure corresponding to the defective part can be found from the previously predicted structural data model, thereby correcting the defective structure.

7. The adaptive learning system realized by integrating IV detection and defect detection according to claim 1, characterized in that: In the step S7, the prediction algorithm is modified by a reverse recursive method, with the measured data as a fixed result and the design model as the input initial data, so that the algorithm can adjust and fit itself according to the fixed input and output results, and multiple iterations are used. Through multiple fittings, the internal parameters of the algorithm can gradually fit the actual situation.

8. The adaptive learning system realized by integrating IV detection and defect detection according to claim 1, characterized in that: The step S8 achieves an iterative effect by re-entering the modified design model as input data into the internal algorithm of the system. During the iterative process, both the design model and the prediction algorithm are adaptively adjusted.

Images