Abstract The estimation of wall thermal properties by in situ measurement enables to increase the reliability of the model predictions for building energy efficiency. Nevertheless, retrieving the unknown parameters has an important computational cost. Indeed, several computations of the heat transfer problem are required to identify these thermal properties. To handle this drawback, an innovative approach is investigated. The first step is to search the optimal experiment design among the sequence of observation of several months. A reduced sequence of observations of three days is identified which guarantees to estimate the parameter with the maximum accuracy. Moreover, the inverse problem is only solved for this short sequence. To decrease further the computational efforts, a reduced order model based on the modal identification method is employed. This a posteriori model reduction method approximates the solution with a lower degree of freedom. The whole methodology is illustrated to estimate the thermal diffusivity of an historical building that has been monitored with temperature sensors for several months. The computational efforts is cut by five. The estimated parameter improves the reliability of the predictions of the wall thermal efficiency.

中文翻译：

---

结合模态辨识法与优化试验设计的古建筑墙体热性能估计

摘要 通过原位测量估计墙体热性能可以提高建筑能效模型预测的可靠性。然而，检索未知参数具有重要的计算成本。实际上，需要对传热问题进行多次计算才能确定这些热特性。为了解决这个缺点，研究了一种创新方法。第一步是在几个月的观察序列中寻找最优的实验设计。确定了一个减少的三天观察序列，这保证了以最大精度估计参数。此外，逆问题仅针对这个短序列解决。为了进一步减少计算工作，采用基于模态识别方法的降阶模型。这种后验模型约简方法以较低的自由度逼近解。说明了整个方法来估计已使用温度传感器监测数月的历史建筑的热扩散率。计算工作减少了五个。估计参数提高了壁热效率预测的可靠性。