Inferring Power System Dynamics From Synchrophasor Data Using Gaussian Processes
Inferring Power System Dynamics From Synchrophasor Data Using Gaussian Processes
复制标题
DOI：
10.1109/tpwrs.2022.3144935
复制
发表时间：
2021-05
影响因子：
6.6
通讯作者：
M. Jalali;V. Kekatos;Siddharth Bhela;Hao Zhu;V. Centeno
中科院分区：
文献类型：
--
作者：
M. Jalali;V. Kekatos;Siddharth Bhela;Hao Zhu;V. Centeno
同步相量数据为推断电力系统动态提供了前所未有的机会，例如估计电压角、频率和加速度以及所有总线的功率注入。为了实现这一目标，这项工作提出了一种利用高斯过程（GP）学习小信号干扰后动态的新颖框架。我们将使用 GP 学习线性时不变系统的结果扩展到多输入多输出设置。这是通过将电力系统动态分解为一组具有窄通带的单输入单输出线性系统来实现的。所提出的学习技术捕获连续时间内的时间导数，适应以不同速率采样的数据流，并且可以应对丢失的数据和异构的准确度水平。虽然基于卡尔曼滤波器的方法需要了解所有系统输入，但所提出的框架可以处理从总线的任意子集收集的系统输入、输出、它们的导数及其组合的读数。依靠最小的系统信息，除了系统动力学的点估计之外，它还进一步提供不确定性量化。数值测试验证了该技术可以推断非计量母线的动态，估算和预测同步相量，并在环境和故障扰动下的线性和非线性系统模型下定位故障。
Synchrophasor data provide unprecedented opportunities for inferring power system dynamics, such as estimating voltage angles, frequencies, and accelerations along with power injection at all buses. Aligned to this goal, this work puts forth a novel framework for learning dynamics after small-signal disturbances by leveraging Gaussian processes (GPs). We extend results on learning of a linear time-invariant system using GPs to the multi-input multi-output setup. This is accomplished by decomposing power system dynamics into a set of single-input single-output linear systems with narrow frequency pass bands. The proposed learning technique captures time derivatives in continuous time, accommodates data streams sampled at different rates, and can cope with missing data and heterogeneous levels of accuracy. While Kalman filter-based approaches require knowing all system inputs, the proposed framework handles readings of system inputs, outputs, their derivatives, and combinations thereof collected from an arbitrary subset of buses. Relying on minimal system information, it further provides uncertainty quantification in addition to point estimates of system dynamics. Numerical tests verify that this technique can infer dynamics at non-metered buses, impute and predict synchrophasors, and locate faults under linear and non-linear system models under ambient and fault disturbances.