Accurate　prediction　of　water　quality　indicators　can　effectively　predict　sudden　water　pollution　events　and　reveal　them　to　water　users　for　reducing　the　impact　of　water　quality　pollution.　Neural　networks,　e.g.,　Long　Short-Term　Memory　(LSTM)　and　encoder-decoder,　have　been　widely　used　to　predict　time　series　data.　However,　as　the　water　quality　data　increases,　it　becomes　unstable　and　highly　nonlinear,　and　therefore,　its　accurate　prediction　becomes　a　big　challenge.　To　solve　it,　this　work　proposes　a　hybrid　prediction　method　called　VBAED　to　predict　the　water　quality　time　series.　VBAED　combines　Variational　mode　decomposition　(VMD),　a　Bidirectional　input　Attention　mechanism,　an　Encoder　with　bidirectional　LSTM　(BiLSTM),　and　a　Decoder　with　a　bidirectional　temporal　attention　mechanism　and　BiLSTM.　The　definition　of　VBAED　is　an　Encoder-Decoder　model　that　uses　VMD　as　mode　decomposition,　combining　BiLSTM　with　a　bidirectional　attention　mechanism.　Specifically,　VBAED　first　adopts　VMD　to　decompose　historical　data　of　a　predicted　factor,　and　its　decomposed　results　are　adopted　as　the　input　along　with　other　features.　Then,　a　bidirectional　input　attention　mechanism　is　adopted　to　add　weights　to　input　features　from　both　directions.　VBAED　adopts　BiLSTM　as　an　encoder　to　extract　hidden　features　from　input　features.　Finally,　the　predicted　result　is　obtained　by　a　BiLSTM　decoder　with　a　bidirectional　temporal　attention　mechanism.　Real-life　data-based　experiments　demonstrate　that　VBAED　obtains　the　best　prediction　results　compared　with　other　widely　used　methods.