Abstract The decentralized fuzzy inference method (DFIM) was applied to estimate the time-dependent heat flux of 1D participating medium. The direct problem concerned on coupled radiation and conduction heat transfer in the medium was solved by the finite volume method and discrete ordinate method. The simulated boundary temperature was served as input for the inverse analysis. The inverse problem was formulated as an optimization approach. Three improved decentralized fuzzy inference methods (IDFIMs) were developed to accelerate the convergence rate and enhance the estimation accuracy. Five kinds of time-dependent heat fluxes were considered to test the performance of the present inverse technique. No prior information on the functional forms of the unknown boundary conditions was needed for the inverse analysis. All retrieval results showed that the incident heat flux of participating medium can be accurately estimated by DFIMs. The proposed IDFIMs achieved better performance than the original DFIM in terms of computational accuracy and efficiency. Moreover, a comparison between the IDFIM and other optimization techniques was conducted. The proposed IDFIM was proved to be more efficient and accurate than conjugate gradient method, Levenberg-Marquardt method, stochastic particle swarm optimization algorithm and genetic algorithm.

中文翻译：

---

改进分散模糊推理方法在估计参与介质热边界条件中的应用

摘要 应用分散模糊推理方法(DFIM)估计一维参与介质的瞬态热通量。通过有限体积法和离散纵坐标法解决了介质中耦合辐射和传导传热的直接问题。模拟边界温度用作逆分析的输入。逆问题被表述为一种优化方法。开发了三种改进的分散模糊推理方法 (IDFIM) 以加快收敛速度​​并提高估计精度。考虑了五种随时间变化的热通量来测试本逆技术的性能。逆分析不需要关于未知边界条件的函数形式的先验信息。所有反演结果表明，DFIMs可以准确估计参与介质的入射热通量。所提出的 IDFIM 在计算精度和效率方面取得了比原始 DFIM 更好的性能。此外，还对 IDFIM 和其他优化技术进行了比较。所提出的IDFIM被证明比共轭梯度法、Levenberg-Marquardt法、随机粒子群优化算法和遗传算法更有效和准确。