Abstract: How to effectively integrate various sensor measurement information in complex systems to form the basis for intelligent control of the system is currently a hot research topic in the field of intelligence. Once the traditional multi-sensor data fusion structure is determined, it cannot be changed which lacks sufficient elastic computing power to ensure the addition of other sensors, and also hinders the construction of the system’s multi-sensor network. In order to form a sensor network to gather comprehensive system data, firstly, a data flow structure based on a data pool is designed on the architecture. So the sensor measurement information is aggregated in the data pool, and related algorithms are treated as services, solving the problem of elastic computing power in data processing; Secondly, the data processed by the service is pushed to the control system according to demand, separating data processing from control, forming a star shaped sensor network centered on the data pool; Finally, an universal non-model data fusion algorithm framework based on detection signals was designed. Which the coefficients after signal decomposition were fitted, and the reconstructed signals were used for state correction, completely eliminating the dependence on the system model and greatly reducing the impact of sensor entering control system. At last, summarized the feasibility and effectiveness of the method from the perspective of design philosophy, structural evolution process, and characteristic analysis. The proposal of this method can provide a new approach for the research of intelligent control of complex systems.

Key words: data pool, intelligent control, non-model, multi-sensor, data fusion