基于改进闭环子空间的水电机组参数辨识方法

以往的开环辨识方法仅适用于水电机组并大网模型,系统并入孤网或小网或空载运行时应采用闭环辨识。具有较好辨识效果的预测形式简约子空间闭环辨识方法(PARSIM-K)充分利用了马尔克夫参数矩阵的Toeplitz结构,通过奇异值分解降阶和线性投影获取模型参数,但需要选择合适的时域参数,目前尚无一般的方法。为此,建立了带有频率噪声的水轮机调速系统模型,提出基于粒子群优化算法参数优化的PARSIM-K。该方法利用粒子群优化算法优化时域参数p、f,提高了辨识精度。与传统开环方法相比,所提方法能够克服噪声的影响,更加简便、安全、实用。仿真结果表明,与未优化参数的方法相比,所提方法辨识的模型参数误差更小、模型精度更高。

Parameter identification method for hydropower generator based on improved closed-loop subspace

The previous open-loop identification methods are only applicable to the model of hydropower generator connected with large power grid. When the system is connected with isolated grid or small power grid or in no-load operation, the closed-loop identification methods should be used. The PARSIM-K(PARsimonious Subspace Identification Method in predictor form),having better identification effects, takes full advantage of the Toeplitz structure of Markov parameter matrix and obtains the model parameters by singular value decomposition order reduction and li-near projection, while it needs to select suitable time-domain parameters and there is no general method now. Therefore, the model of hydraulic turbine governing system with frequency noise is set up and PARSIM-K based on parameters optimized by particle swarm optimization algorithm is proposed. The time-domain parameters p and f are optimized by particle swarm optimization algorithm to improve the identification precision. Compared with the traditional open-loop methods, the proposed method can overcome the influence of noise and it is more convenient, safer and more practical. Simulative results show that model parameters identified by the proposed method have smaller error and the model precision is higher compared with the method with non-optimized parameters.