The current forecasting models for wind power time series have limitations in handling irregularly sampled time series and need to pay more attention to the piecewise control characteristics of wind speed on wind power output. To address these issues

this paper proposes a short-term wind power forecasting method based on a piecewise controlled hybrid differential neural network. First

the wind speeds at discrete timesteps are converted into a continuous path using cubic spline interpolation. The hybrid differential neural network is employed to model the temporal inertia of wind power and the regulatory effects of continuous wind speed on wind power

thereby capturing its dynamic evolution process in full. Then

based on the differentiated control modes for wind power output when wind speed falls into different numerical intervals

a multilayer perceptron is employed to dynamically output continuous weighting values representing the affiliation of wind speed to distinct control intervals. Finally

the short-term power forecasts for the wind farm are output

which can simultaneously satisfy the temporal inertia of wind power and the piecewise controlling patterns of wind speed. Practical case studies demonstrate that this method exhibits significant advantages in handling irregularly sampled time series

with the proposed piecewise control hybrid differential neural network outperforming the popular RNN networks in terms of forecasting accuracy and reliability.