Blueberry matrix temperature and humidity prediction model combining long-term and short-term memory and attention mechanism

The invention discloses a blueberry matrix temperature and humidity prediction model combining long-term and short-term memory and an attention mechanism, and the model comprises a preprocessing module which carries out the preprocessing of greenhouse blueberry growth environment data, obtains a blueberry environment data set, adds a timestamp system into the blueberry environment data set, and carries out the normalization processing, and segments the blueberry environment data set; an attention mechanism fusion module which uses an LSTM feature extraction mode to obtain blueberry growth environment long-term features and blueberry growth environment short-term features, and combines the blueberry growth environment long-term features with a ProbSparse self-attention mode; the feature extraction module is used for respectively inputting the segmented blueberry environment data into an encoder and a decoder, and applying a fusion attention mechanism to respectively generate blueberry environment parameter fea