A　one-step　forward　forecasting　test　for　carbon　market　pricing　is　done　in　this　work,　in　which　data　from　the　first　seven　trading　days　is　used　to　anticipate　the　price　on　the　eighth　trading　day.　The　study　compares　the　MAE,　MSE,　and　RMSE　values　of　several　forecasting　models　and　discovers　that　combining　empirical　mode　decomposition　(EMD)　and　forecasting　models　yields　the　best　results.　It　was　shown　that　the　hybrid　model　can　improve　both　the　durability　and　accuracy　of　carbon　price　estimates.　In　terms　of　forecasting,　the　combined　EMD-BiLSTM-ATTENTION　model　beats　other　comparator　models,　and　carbon　price　forecasting　errors　in　Hubei　and　Fujian　are　smaller　than　those　in　Shenzhen　due　to　their　more　stable　frequency　amplitudes.　Nevertheless,　it　has　been　discovered　that　estimating　the　carbon　price　in　Shenzhen　is　more　difficult　due　to　higher　amplitude　variations,　resulting　in　higher　prediction　errors.　Overall,　the　findings　indicate　that　the　proposed　EMD-BiLSTM-ATTENTION　model　is　appropriate　for　carbon　price　prediction,　and　the　study　includes　carbon　market　price　prediction　maps　for　Shenzhen,　Hubei,　and　Fujian.　©　2023　IEEE.