讨论了马尔科夫链状态划分的黄金分割率法和“马氏性”检验法,并针对BP神经网络预测和马尔科夫预测的优缺点,提出了BP神经网络与马尔科夫相耦合的BP神经网络马尔科夫模型,以石泉水库年入库径流苗为例,验证了该方法的可行性.
This paper discusses the methods of dividing of state space with Golden Ratio and Markov property test. Aiming at the advantages and disadvantages of BP neural network and Markov prediction model, a new method is presented based on analyzing BP neural network theory and Markov theory. This prediction model is identified by taking prediction of annual runoff variation into the Shiquan Reservoir.