Search International and National Patent Collections

1. (WO2018076475) PHOTOVOLTAIC ASSEMBLY ACCELERATED DEGRADATION MODEL ESTABLISHED BASED ON DEEP APPROACH OF LEARNING, AND METHOD FOR PREDICTING PHOTOVOLTAIC ASSEMBLY LIFETIME

Pub. No.:    WO/2018/076475    International Application No.:    PCT/CN2016/109121
Publication Date: Fri May 04 01:59:59 CEST 2018 International Filing Date: Sat Dec 10 00:59:59 CET 2016
IPC: G06F 19/00
G06N 3/08
Applicants: GUANGDONG TESTING INSTITUTE OF PRODUCT QUALITY SUPERVISION
广东产品质量监督检验研究院
Inventors: YU, Rongbin
余荣斌
LIU, Guixiong
刘桂雄
XU, Huan
徐欢
Title: PHOTOVOLTAIC ASSEMBLY ACCELERATED DEGRADATION MODEL ESTABLISHED BASED ON DEEP APPROACH OF LEARNING, AND METHOD FOR PREDICTING PHOTOVOLTAIC ASSEMBLY LIFETIME
Abstract:
Disclosed are a photovoltaic assembly accelerated degradation model established based on a deep approach of learning, and a method for predicting photovoltaic assembly lifetime. This method establishes a deep neural network (DNN) via a restricted boltzmann machine (RBM), takes different accelerated stress conditions (T i, H i and R ai) and the corresponding pseudo-failure lifetime distribution quantile function Q i(p) as input vectors, uses a CD fast learning algorithm to train the RBM and the DNN, seeks a model optimal parameter set θ *, and constructs a photovoltaic assembly accelerated degradation model, thereby predicting an expected working life of the photovoltaic assembly under normal stress conditions.