Photovoltaic Modules Service Reliability Evaluation Based On Performance Degradation

Based on performance degradation data reliability evaluation and combined with the product failure physical analysis, This Paper established reliability evaluation models and technical methods from the natural degradation and accelerated degradation in both directions, and improving the quality of the products, ensuring the rapid development of strategic emerging industries, promoting advanced manufacturing technology, test theory and application technology development. This work has important academic value and practical significance. The research work is supported by The State Administration of Quality Supervision Inspection and Quarantine project(No.2014QK050) and Guangdong provincial quality and technical supervision key research project(No.2012ZZ05).The dissertation researches identified research content from the PV module reliability influence factors and failure analysis, reliability assessment technology based on performance degradation, reliability evaluation under accelerated degradation test methods and design optimization. Primary work includes:⑴ Further study on failure risk evaluation and failure analysis of PV modules. Based on product features, combined with the physical failure analysis, a new method based on risk measurement and risk sort of P-W-O fuzzy comprehensive evaluation is presented. Designing the new evaluation index set(key performance indicators P, maintenance cost W and probability O), the method achieves objective and significance of the index evaluation. As the risk evaluation results, the key components PET backplane, EVA film and an electrical connector of PV modules are the object, including PET back, EVA film under hot and humid conditions of the outdoor environment hydrolysis aging, light and heat degradation effects of aging process, reaction mechanism, a photovoltaic module encapsulation materials under hot and humid conditions and aging water vapor infiltration process model, and exploring electrical connector contact failure mechanism, influencing factors, establishing contact failure under stress temperature physics equations to guide PV module component reliability modeling, performance degradation test design.⑵ Proposed a modeling method of PV module performance degradation based on β-distribution uniform(β-PDVD). It showed the true reflection of performance degradation quantity distribution fitting at each time, without subjective assuming distribution types in advance, then the genetic algorithm and fuzzy comprehensive judgment to solve the PV module reliability distribution function parameters, preferably optimum reliability distribution, and finally, taking performance degradation data of mono-crystalline silicon PV modules made by BP solar, for example, access to PV modules reliability function under natural degradation. In considering only the performance of PV modules output power parameters under natural degradation, the batch of PV module began to fail from work after 17 years, the median life and characteristic life, respectively 20.39 years and 20.84 years, the entire failure in 24.89 years, basically same as the 25 years life of the quality assurance provided by the manufacturer.⑶ Study reliability evaluation method of PV modules pseudo-failure lifetime distribution estimate based on accelerated degradation data. First, the wavelet analysis combined stationary time series are applied for accelerated degradation test data preprocessing, reducing monitoring data random noise, increasing the degradation paths and life prediction accuracy, and using R2(coefficient of determination test method) to prefer the best track of accelerated degradation and estimating pseudo-failure life of each sample. Then a method for estimating the pseudo failure life distribution of PV modules based on GLD(The Generalized Lambda Distribution) is proposed, using the bootstrap method generate bootstrap samples and expand the sample group, building pseudo failure life distribution model based on GLD. No need to preset the prior distribution, a true reflection of the pseudo failure life distribution of PV modules under different acceleration conditions. Finally, study on the Deep Learning Forecast for accelerated degradation modeling method, exploring build problems of the more stress, the complex structure product accelerated degradation model, researching the corresponding relationship between accelerated condition and normal conditions to estimation under the condition of normal use photovoltaic component reliability and service life.⑷ Study on Photovoltaic module accelerated degradation test design and optimization. Considering PV modules comprehensive influence stress factors, time-consuming test period, and the great samples in the practical application difficult to accept, module accelerated degradation test design and optimization are studied to greatly reduce the number of tests and test costs, shorten the test cycle. The maximum likelihood(ML) estimation method is applied for establishing PV modules accelerated degradation optimization model, optimizing test program and improving testing accuracy. Optimizing results show: After optimization of the test program, asymptotic variance AVar((?)(0.5))(28)0.2079 is reduced 0.1272 than before, and estimation accuracy is increased by 38.1%, the numbers of test samples required are decreased by 37.9% from n’=35.9939 to n=22.3346.Finally, the integrated test environment simulation platform of PV modules are built, developing PV modules accelerated degradation Reliability Assessment software on the data pre-processing(DPP), selecting the optimal trajectory degradation(OTDS), pseudo-life distribution failure prediction(PFLF) and accelerated degradation modeling(ADM). The test results demonstrated: the failure life, characteristic life and median lifetime values of testing PV modules were 22.1 years, 21.6 years and 21.1 years, relatively manufacturer error of 13.1%, 10.7%, 9.5%, meeting the engineering prediction accuracy requirements.