改進(jìn)LSTM在金沙江流域中長(zhǎng)期徑流預(yù)報(bào)研究
水利水電技術(shù)(中英文)
頁(yè)數(shù): 11 2024-03-20
摘要: 徑流非平穩(wěn)性及超參數(shù)率定容易陷入“維數(shù)災(zāi)”的問題限制了神經(jīng)網(wǎng)絡(luò)(Long Short-Term Memory, LSTM)在中長(zhǎng)期徑流預(yù)報(bào)中的應(yīng)用,針對(duì)上述問題,采取“分解-預(yù)測(cè)-重構(gòu)”的思路,在數(shù)據(jù)預(yù)處理階段采用變分模態(tài)分解(Variational Mode Decomposition, VMD)進(jìn)行分解重構(gòu),耦合PSO(Particle Swarm Optimization...