Nonlinear Hydrological Time Series Forecasting Based on the Relevance Vector Regression
As long leading-time hydrological forecast is a complex non-linear procedure, traditional methods are easy to get slow convergence and low efficiency. The basic relevance vector machine (BRVM) and the developed sequential relevance vector machine (SRVM) are employed to forecast multi-step ahead hydrological time series. The relevance vector machine is a sparse approximate Bayesian kernel method, and it provides full probabilistic forecasting results, which is helpful for hydrological engineering decision. BRVM and SRVM are respectively applied to the annual coming runoff forecast of Three Gorges hydropower station as case study. When compared with auto regression moving average models, BRVM exhibits high model efficiency and provides satisfying forecasting precision. SRVM is potential for its increased freedom and adaptive model selection mechanism. Comparison is also made within direct forecast and iterative one-step ahead forecasting for multi-step ahead forecasting, and the latter shows the ability of highlighting the model performance.
KeywordsMean Absolute Error Forecast Result Hydropower Station Relevance Vector Machine Error Index
Unable to display preview. Download preview PDF.
- 1.Fan, X.Z.: Mid and Long-term Hydrologic Forecast. Hehai University Publishing Company, Nanjing (1999)Google Scholar
- 3.Qin, G.H., Ding, J., Li, M.M., Ni, C.J.: Application of ANNs with Sensitive Ability to Hydrologic Forecast. Advances in Water Science 14(2), 163–166 (2003)Google Scholar
- 4.Yuan, J., Zhang, X.F.: Real-time Hydrological Forecasting Method of Artificial Neural Network Based on Forgetting Factor. Advances in Water Science 15(6), 787–792 (2004)Google Scholar
- 5.Liu, F., Zhou, J.Z., Yang, J.J., Qiu, F.P.: The Application of Fuzzy System with Recursive Least Squares Method to Mid and Long-Term Runoff Forecast. In: Proceedings of World Water and Environmental Resources Congress, Anchorage, Alaska, USA, pp. 337–344 (2005)Google Scholar
- 7.NIkolaev, N.: Sequential Relevance Vector Machine Learning from Time Series. In: Proceedings of International Joint Conference on Neural Networks, Montreal, Canadapp, pp. 1308–1313 (2005)Google Scholar
- 8.Tipping, M.E.: The Relevance Vector Machine. In: Proceedings of Advances in Neural Information Processing Systems 12, pp. 652–658. MIT Press, Cambridge (2000)Google Scholar
- 10.Tipping, M.E., Faul, A.: Fast Marginal Likelihood Maximisation for Sparse Bayesian models. In: Proceedings of the Ninth International Workshop on Social Intelligence Statistics, Key West (2003)Google Scholar
- 14.Faul, A., Tipping, M.E.: Analysis of Sparse Bayesian Learning. In: Proceedings of NIPS (2001)Google Scholar
- 15.Candela, J.Q., Girard, A., Larsen, J., Rasmussen, C.E.: Propagation of Uncertainty in Bayesian Kernel Models – Application to Multi-step Ahead Forecasting. In: Proceedings of ICASSP (2003)Google Scholar
- 17.Candela, J.Q., Hansen, L.K.: Time Series Prediction Based on The Relevance Vector Machine with Adaptive Kernels. In: Proceedings of IEEE International Conference on Acoustics, Speech, and Signal Processing, pp. 985–988 (2003)Google Scholar