Backward-Curved Centrifugal Fan Flow Control Without Sensors

The patent introduces an intermediary computational model that processes multiple measurable parameters (motor current, voltage, frequency, temperature) to indirectly determine volume flow rate. This mediator translates ambiguous individual measurements into accurate flow rate estimation through mathematical relationships, eliminating the need for direct flow sensors while maintaining measurement precision.

The patent replaces mechanical/physical flow measurement systems (flow sensors, differential pressure sensors) with an electronic computational system. By substituting physical measurement devices with algorithmic processing of electrical parameters, the system reduces device complexity while maintaining or improving measurement accuracy through the use of trained neural networks or lookup tables.

The patent performs preliminary action by pre-training neural networks or pre-calculating lookup tables during the manufacturing phase. Extensive training data is collected and processed beforehand to create highly accurate models that are then deployed in the operating system. This preliminary preparation enables precise control accuracy during operation without requiring complex real-time computational resources.

The patent transforms the control approach by changing from simple mathematical formulas to sophisticated parameter relationships captured in trained neural networks. By incorporating multiple parameters (current, voltage, frequency, temperature) and their interactions, the system achieves high precision control. The trained models capture non-linear relationships and cross-parameter dependencies that simple algorithms miss.