Neural network for learning and approximation in configuration of spatial combinatorial cam applied to double regulating hydraulic turbines

The present paper describes the use of a VLSI neural network chip in double regulating speed turbines for hydraulic turbine. The chip is used in low head hydropower plants, with the aim of increasing hydraulic-mechanical efficiency. The idea is to replace the mechanical device named the combinatorial spatial cam with this electronic chip, which realises much more accurately the correlation between the two regulating organs of a double regulating hydraulic turbine: the guide vane opening and the rotor blade rotation angle. This correlation is given by the hydraulic turbine producer only experimentally in the form of sets of triplets (points in status space). This approach is to improve the interpolation between points, which is best realised with spleen functions by the neural network for learning and approximation VLSI chip.