面向复杂场景的大型公共建筑需求冷负荷智能感知方法

The refrigeration process in large public buildings consumes the most energy, with the required cooling load being a crucial factor influencing this energy use. Accurately and intelligently sensing the cooling load demand online in such buildings is vital for enhancing the quality and efficiency of refrigeration. However, current methods face challenges like heavy reliance on data, limited accuracy, and predictions limited to single scenarios. An intelligent perception approach for cooling load demand in large public buildings under complex conditions was introduced. The approach began by classifying cooling scenarios and central air conditioning operating states using a density trajectory space clustering method. Next, the correlation coefficient method was employed to quantitatively assess the relationship between each variable and the cooling load demand across different scenarios. Finally, a dynamic variable hidden layer neural network was processed that used variables with strong correlations as inputs and cooling load demand as the output. By adaptively adjusting and iteratively optimizing the number of hidden layer nodes, the method achieved trend prediction of cooling load demand. Experimental results based on real operational data demonstrate that this method can accurately detect cooling load demand online and intelligently forecast short-term trends, offering valuable data support for energy-saving and consumption reduction in central air conditioning operations.