A monitoring system with load feature recognition is crucial to the realization of autonomic and intelligent management of power demand side. In order to solve the problem that nonintrusive detection technology is limited to established equipment, to promote the active recognition of all kinds of electrical equipment so that the functions of power consuming behavior analysis and power optimization control are supported in automatic demand side management system, load dactylogram and its composition based on operational characteristics of household appliances are defined. Based on cyber-physical system, a hierarchical technology structure for load dactylogram management is presented, which includes the double layer dactylogram management platform with a local system and a cloud server. Finally, intelligent gathering terminal with socket and intelligent interactive concentrator with multiple communication technologies are developed to provide strong support of hardware for the realization of the technical architecture proposed in this paper.