Self-adaptive mixed rime optimization public building cooling load prediction method and related device

The invention provides an adaptive mixed rime optimization public building cold load prediction method and a related device. The method comprises the following specific steps: acquiring cold load data of a public building; inputting the cold load data of the public building into the public building cold load prediction network model to obtain a cold load prediction result; wherein the public building cold load prediction network is constructed through a random forest-based feature selection algorithm, a convolutional neural network, a bidirectional long-short-term memory neural network and a multi-head self-attention mechanism which are connected in sequence; a rime optimization algorithm is improved through adaptive tent chaotic mapping, adaptive t distribution perturbation and an elitism strategy fused with a dandelion optimizer algorithm to obtain an adaptive mixed rime optimization algorithm, and the adaptive mixed rime optimization algorithm is utilized to perform parameter optimization on a public build