Towards Fast Bayesian inference of equivalent circuit parameters of perovskite solar cell

Interpreting current–voltage data through an equivalent circuit model offers a rapid, real-time method for analyzing the internal changes in perovskite solar cells. Recent studies have utilized Bayesian inference to solve this inverse problem, but the inference time remains a bottleneck for practical applications. In this work, we propose a modified approach that simplifies the likelihood function within the Bayesian inference process. By leveraging the inherent characteristics of the equivalent circuit, we reduce the five-parameter model to a more efficient three-parameter approximation, greatly enhancing parameter estimation efficiency. Our results demonstrate that the simplified model retains the same level of computational accuracy. In addition, it reduces inference time by a factor of over 15, from hundreds of seconds to just a few seconds on Google Colaboratory using a CPU with two Xeon cores (2.2 GHz). The number of inference steps also decreases from tens of thousands to only a few hundred. This significant improvement accelerates the overall inference process, enabling faster, real-time monitoring and aging analysis of perovskite solar cells. Our method offers an intelligent solution for efficiently analyzing solar cell performance through Bayesian inference, advancing both research and engineering applications.