Simultaneous identification of building dynamic model and disturbance using sparsity-promoting optimization
Simultaneous identification of building dynamic model and disturbance using sparsity-promoting optimization
复制标题
使用稀疏性促进优化同时识别建筑动力模型和扰动
DOI：
--
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
P. Barooah
中科院分区：
文献类型：
--
作者：
Tingting Zeng;J. Brooks;P. Barooah
我们提出了一种在存在大的、未测量的扰动的情况下识别用于暖通空调控制的热建筑模型的方法。此外，该方法还识别那些未测量的干扰对输出的影响。我们的方法使用`1-正则化来鼓励识别的尺度扰动的导数稀疏，其动机是物理上有意义的。我们使用来自开环和闭环仿真的训练数据来测试我们的方法。结果表明，即使在存在较大的不可测量扰动的情况下，所辨识的模型也能准确地辨识出两种情况下从流量和送风温度到室温的传递函数，这使得该模型具有很高的应用价值。
We propose a method for identifying thermal building models for HVAC control in the presence of large, unmeasured disturbances. In addition, the method also identifies the effects of those unmeasured disturbances on the output. Our method uses `1-regularization to encourage the derivative of the identified scaled disturbance to be sparse, the motivation of which is physically meaningful. We test our method using training data from both open-loop and closed-loop simulations. Results show that the identified model can accurately identify the transfer functions from flow rate and supply-air temperature to room temperature in both cases, even in the presence of large, unmeasured disturbances, which makes it valuable for MPC applications.