An efficient and accurate uncertainty propagation methodology for mechanics problems with random fields is developed in this paper. This methodology is based on the stochastic response surface method （SRSM） which ha...An efficient and accurate uncertainty propagation methodology for mechanics problems with random fields is developed in this paper. This methodology is based on the stochastic response surface method （SRSM） which has been previously proposed for problems dealing with random variables only. This paper extends SRSM to problems involving random fields or random processes fields. The favorable property of SRSM lies in that the deterministic computational model can be treated as a black box, as in the case of commercial finite element codes. Numerical examples are used to highlight the features of this technique and to demonstrate the accuracy and efficiency of the proposed method. A comparison with Monte Carlo simulation shows that the proposed method can achieve numerical results close to those from Monte Carlo simulation while dramatically reducing the number of deterministic finite element runs.展开更多
A novel method based on time-dependent stochastic orthogonal bases for stochastic response surface approximation is proposed to overcome the problem of significant errors in the utilization of the generalized polynomi...A novel method based on time-dependent stochastic orthogonal bases for stochastic response surface approximation is proposed to overcome the problem of significant errors in the utilization of the generalized polynomial chaos(GPC) method that approximates the stochastic response by orthogonal polynomials. The accuracy and effectiveness of the method are illustrated by different numerical examples including both linear and nonlinear problems. The results indicate that the proposed method modifies the stochastic bases adaptively, and has a better approximation for the probability density function in contrast to the GPC method.展开更多
高比例分布式电源的不确定性给孤岛配电网的稳定运行带来了的巨大的挑战。针对基于传统分布模型的源荷短期预测存在尖峰和重尾的缺点,采用双向长短时记忆(bidirectional long and short-term memory,BiLSTM)神经网络与非参数核密度法(ke...高比例分布式电源的不确定性给孤岛配电网的稳定运行带来了的巨大的挑战。针对基于传统分布模型的源荷短期预测存在尖峰和重尾的缺点,采用双向长短时记忆(bidirectional long and short-term memory,BiLSTM)神经网络与非参数核密度法(kernel density method,KDE)结合的方法,构建了多场景及不同时间尺度下源荷预测误差的分布模型;并在此基础上,系统多时段运行调控过程中,考虑短时气象的不确定性波动,采用混合整数二阶锥规划(mixed-integer second-order cone programming,MISOCP)对潮流模型进行松弛,并由随机响应面(stochastic response surface,SRSM)得到系统的概率潮流;基于随机响应面法改进Sobol’法,建立计及源荷不确定性的孤岛配电网运行风险的全局灵敏度分析模型。基于此提出一种基于Bi LSTM-SRSM法的风险实时风险评估及调控策略。最后,采用IEEE33节点的辐射型配电网系统验证了所提方法的可行性。展开更多
基金The project supported by the National Natural Science Foundation of China(10602036)
文摘An efficient and accurate uncertainty propagation methodology for mechanics problems with random fields is developed in this paper. This methodology is based on the stochastic response surface method （SRSM） which has been previously proposed for problems dealing with random variables only. This paper extends SRSM to problems involving random fields or random processes fields. The favorable property of SRSM lies in that the deterministic computational model can be treated as a black box, as in the case of commercial finite element codes. Numerical examples are used to highlight the features of this technique and to demonstrate the accuracy and efficiency of the proposed method. A comparison with Monte Carlo simulation shows that the proposed method can achieve numerical results close to those from Monte Carlo simulation while dramatically reducing the number of deterministic finite element runs.
基金Project supported by the National Natural Science Foundation of China(Nos.11632011,11572189,and 51421092)the China Postdoctoral Science Foundation(No.2016M601585)
文摘A novel method based on time-dependent stochastic orthogonal bases for stochastic response surface approximation is proposed to overcome the problem of significant errors in the utilization of the generalized polynomial chaos(GPC) method that approximates the stochastic response by orthogonal polynomials. The accuracy and effectiveness of the method are illustrated by different numerical examples including both linear and nonlinear problems. The results indicate that the proposed method modifies the stochastic bases adaptively, and has a better approximation for the probability density function in contrast to the GPC method.
文摘高比例分布式电源的不确定性给孤岛配电网的稳定运行带来了的巨大的挑战。针对基于传统分布模型的源荷短期预测存在尖峰和重尾的缺点,采用双向长短时记忆(bidirectional long and short-term memory,BiLSTM)神经网络与非参数核密度法(kernel density method,KDE)结合的方法,构建了多场景及不同时间尺度下源荷预测误差的分布模型;并在此基础上,系统多时段运行调控过程中,考虑短时气象的不确定性波动,采用混合整数二阶锥规划(mixed-integer second-order cone programming,MISOCP)对潮流模型进行松弛,并由随机响应面(stochastic response surface,SRSM)得到系统的概率潮流;基于随机响应面法改进Sobol’法,建立计及源荷不确定性的孤岛配电网运行风险的全局灵敏度分析模型。基于此提出一种基于Bi LSTM-SRSM法的风险实时风险评估及调控策略。最后,采用IEEE33节点的辐射型配电网系统验证了所提方法的可行性。
