The BPNN learning rate and momentum coefficient are found to be 0.01 and 0.95–1.0, respectively (Table 2). Reasonable model-field measurement matches (see below for a description of the model performance statistics) are obtained from the Hukou model after 3,000 iterations (Table 2). Hence, the same number of iterations is used for training other BPNN models, and convergence (based on MSE) to an optimal set of parameters was tested in each case. The closeness of the learning rate, momentum coefficient, and number of iterations for all models (and scenarios) demonstrates that the model results are not especially sensitive to these parameters. The number of neurons in the hidden layer is the main parameter that varies between models, ranging from 21 to 33 neurons to predict lake water levels with acceptable accuracy (Table 2). The sensitivity of the model to the optimal parameters in this study is consistent with previous BPNN modeling by Chen *et al.* (2012a, b), who found that the optimal number of nodes in the hidden layer was important for obtaining the best network architecture.

Table 2

. | . | Model parameters . | ||||
---|---|---|---|---|---|---|

BPNN model . | Model scenario . | Learning rate . | Momentum coefficient . | Iteration number . | Input nodes . | Hidden nodes . |

Hukou | S1 | 0.01 | 1.0 | 3,000 | 5 | 22 |

S2 | 0.01 | 1.0 | 3,000 | 6 | 22 | |

S3 | 0.01 | 1.0 | 3,000 | 6 | 21 | |

Xingzi | S1 | 0.01 | 1.0 | 3,000 | 5 | 28 |

S2 | 0.01 | 1.0 | 3,000 | 6 | 25 | |

S3 | 0.01 | 1.0 | 3,000 | 6 | 25 | |

Duchang | S1 | 0.01 | 0.98 | 3,000 | 5 | 22 |

S2 | 0.01 | 0.98 | 3,000 | 6 | 22 | |

S3 | 0.01 | 0.98 | 3,000 | 6 | 21 | |

Tangyin | S1 | 0.01 | 0.95 | 3,000 | 5 | 22 |

S2 | 0.01 | 1.0 | 3,000 | 6 | 24 | |

S3 | 0.01 | 1.0 | 3,000 | 6 | 21 | |

Kangshan | S1 | 0.01 | 0.98 | 3,000 | 5 | 22 |

S2 | 0.01 | 0.98 | 3,000 | 6 | 33 | |

S3 | 0.01 | 0.98 | 3,000 | 6 | 23 |

. | . | Model parameters . | ||||
---|---|---|---|---|---|---|

BPNN model . | Model scenario . | Learning rate . | Momentum coefficient . | Iteration number . | Input nodes . | Hidden nodes . |

Hukou | S1 | 0.01 | 1.0 | 3,000 | 5 | 22 |

S2 | 0.01 | 1.0 | 3,000 | 6 | 22 | |

S3 | 0.01 | 1.0 | 3,000 | 6 | 21 | |

Xingzi | S1 | 0.01 | 1.0 | 3,000 | 5 | 28 |

S2 | 0.01 | 1.0 | 3,000 | 6 | 25 | |

S3 | 0.01 | 1.0 | 3,000 | 6 | 25 | |

Duchang | S1 | 0.01 | 0.98 | 3,000 | 5 | 22 |

S2 | 0.01 | 0.98 | 3,000 | 6 | 22 | |

S3 | 0.01 | 0.98 | 3,000 | 6 | 21 | |

Tangyin | S1 | 0.01 | 0.95 | 3,000 | 5 | 22 |

S2 | 0.01 | 1.0 | 3,000 | 6 | 24 | |

S3 | 0.01 | 1.0 | 3,000 | 6 | 21 | |

Kangshan | S1 | 0.01 | 0.98 | 3,000 | 5 | 22 |

S2 | 0.01 | 0.98 | 3,000 | 6 | 33 | |

S3 | 0.01 | 0.98 | 3,000 | 6 | 23 |

This site uses cookies. By continuing to use our website, you are agreeing to our privacy policy.