基於BP-Adaboost強預測器的瓦斯湧出量預測

摘　要:為了提高傳統BP神經網絡瓦斯湧出量預測模型精度,避免BP網絡容易陷入局部極值、收斂速度慢等問題,將BP神經網絡和Adaboost算法相結合,提出了一種BP-Adaboost強預測器模型.將該模型用於實際瓦斯湧出量預測,並進行了40次仿真實驗.結果表明：該模型預測精度高於傳統的BP神經網絡,且收斂速度快,具有較強的魯棒性,預測精度能滿足實際工程需要,為瓦斯湧出量預測提供了一種新的途徑.

關鍵詞:瓦斯湧出量 BP神經網絡 Adaboost算法 預測

分類號:　TD712.5Prediction for Gas Emission Quantity Based on BP-AdaboostZHANG Tian- xing , BI Jian- wu (Liaoning Technical University a. College of Mining Engineering; b. College of Safety Science and Engineering, Fuxin 123000, China)Abstract:In order to improve the gas emission prediction accuracy of traditional BP neural network model, a prediction model was presented based on combined BP neural network and Adaboost algorithm. With simulation experiment carried out by forty times and compared with other models, the authors concluded that the model prediction accuracy was higher than the traditional BP neural network, and had strong robustness. The result showed that convergence speed was fast. The accuracy could meet the requirements of practical engineering, which provided a new way for the gas emission prediction.

Keyword:gas emission ; BP neural network ; Adaboost algorithm ; forecast