The invention provides a heterogeneous building air conditioner group quick
response control method based on
physical information deep learning, and the method comprises the steps: 1, obtaining a real-time discrete step scheduling instruction issued by a
power grid, obtaining the thermodynamic parameters and equipment
inertia time constants of all building units in a heterogeneous building group, obtaining the
outdoor temperature, and carrying out the real-time discrete step scheduling instruction; the solar
radiation intensity and the
indoor air temperature of each
building unit; step 2, extracting spatial-temporal characteristics of each
building unit based on a multi-head self-attention mechanism, and generating
initial distribution target power of the air conditioner; 3, inputting the initial allocation target power into a pre-trained cascaded
physical information neural network (PINN), and predicting power and temperature evolution tracks in a future scheduling period; 4, calculating an out-of-limit
risk index of each unit, judging whether to execute a quantile risk migration strategy or not, and generating a final distribution target power of the air conditioner; and 5, calculating the transient climbing deviation between the actual aggregation power and the target power of the
power grid, and carrying out deviation compensation.