Any natural action creates lost energy which could be exploited to supply our electrical and mobile appliance. Our physical environments have a high number of micro-energy sources. Admittedly, each one provides low power but their multiplicity could be significant, in particular within the framework of the microsystem operation. The previous observation guided our works towards the problematic of autonomous microsystem design. Thus, to be innovative, microsystems engineering must lean on electronic, mechanical and energy domains. The design process is highly multidisciplinary and its efficiency depends on the ability to implement methods and tools: - of collaborative design - of capitalization of technical knowledge - of multiphysic engineering - of integrated design. Based on these fundamentals, we developed a design support tool. The underlying methodology enables: 6- the design problem analysis and structuring of an autonomous microsystem: this phase leads to the identification and functional and environmental description of the system and its environment 7- the knowledge modelling: an architectural analysis gives the description of components and interactions related to the microsystem (directly or indirectly). Then, it leads to a behaviour modelling. 8- the energy qualification and physical coupling: the structured reuse of knowledge models is guided to couple physical models and describe the sources, sinks and the energy mechanism of the environment. 9- the control of innovative concept search: the knowledge base, qualification criteria and functional description, previously constructed, are combined in an unique virtual design approach dedicated to search innovative concepts as a solution 10- the predimensioning: this phase ensures the integration of specific simulation tools (finite elements method and functional simulation). The predimensioning of autonomous microsystems is supported by a synthetic scheme based on an abductive reasoning (bottom-up). The combination of physical reasoning, the integration of methods and engineering domains, the virtual exploration of solution spaces and the modelling represent a new way to support autonomous microsystem design. This approach was applied to the design of an autonomous piezoelectric sensor.
