Cloud edge-end integrated intelligent agent collaborative control method and system

Abstract

The application provides a cloud-edge-end integrated intelligent agent collaborative control method and system. The method comprises: the edge end receives original data collected by a terminal. The original data is preprocessed to extract a feature vector representing the running state of the terminal device or the environmental state. The edge end inputs the feature vector into a local decision model, and the local decision model generates a local decision according to the feature vector and calculates the decision confidence of the local decision; the local decision model is a lightweight machine learning model or a rule engine deployed on the edge end. The edge end determines whether the decision confidence reaches a preset threshold: if the preset threshold is reached, the edge end sends the local decision to the terminal for execution; if the preset threshold is not reached, the edge end uploads the feature vector to the cloud end. The cloud end receives the feature vector, processes the feature vector by using a global optimization algorithm or a large-scale machine learning model, generates a global decision, and sends the global decision to the edge end or the terminal for execution.