How to combine both numerical data and categorial data for numeric prediction in time series LSTM

Question

I am working on power prediction using time series LSTM. I have implemented the model and my predictions were okay. Now i want to consider adding the weekday for each time series so i'll get a better power prediction, because the each day (Mon to Fri) has it own effect on the power even at different time. I one hot encoded the week days and i now have 8 columns (Power and 7 days of the week). I have generated a sequence but i get alot of errors.

I'm yet to get the best practise on solving this type of question.

Here is a snippet of my scaled data, numeric and one hot encoding.

        0.        ],
       [0.21494554, 0.        , 1.        , ..., 0.        , 0.        ,
        0.        ],
       [0.21770302, 0.        , 1.        , ..., 0.        , 0.        ,
        0.        ],
       ...,
       [0.2135668 , 0.        , 0.        , ..., 0.        , 0.        ,
        0.        ],
       [0.20681097, 0.        , 0.        , ..., 0.        , 0.        ,
        0.        ],
       [0.21549704, 0.        , 0.        , ..., 0.        , 0.        ,
        0.        ]])```