Internal Name: State Space Controller

What’s New: An optimum path to soft landing is mapped out in advance, tailored to the specific characteristics of the system being controlled. This “roadmap,” specifying magnetic flux linkage as a function of position and velocity, resides in controller memory as feed-forward information, keeping the system on a dynamic trajectory that maximizes latitude for course correction in the presence of strong external perturbations such as vibration and turbulent gas flow past a valve.

Status: issued U.S. August 29, 2006

Why: In a valve actuation solenoid, feasible dynamic trajectories are strongly constrained by magnetic saturation, by inductive limitation of the rate-of-change of magnetic force, and by the small fraction of a solenoid stroke over which a strong magnetic force is possible. “Reasonable” control choices made early in a trajectory can easily leave a solenoid in a condition from which soft landing with latching is impossible. Unpredictable gas flow forces can also prevent successful landing. This controller achieves near-optimum results under the widest possible range of external path perturbations, in an approach that translates into fast, efficient digital code.

How: We have a detailed dynamic computer model of our valve solenoid – and can readily develop such a model for a client’s solenoid. This model takes account of inertia, spring forces, friction, nonlinear magnetic forces, magnetic hysteresis, and eddy currents. Working with our “Sensorless Position Measurement” method, our system tracks position, velocity, and magnetic flux, yielding a complete description of the instantaneous electromagnetic and mechanical state of the system at each controller time step. Our model delineates the boundaries of this three-variable state space, where “out-of-bounds” means dynamic states from which successful soft landing of the solenoid is impossible. Within these boundaries we design a trajectory “roadmap” that maximizes the latitude for course corrections responding to possible perturbations of the system: from external vibrations, gas-flow forces acting on the valve, temperature-dependent friction, etc. This target trajectory is defined by two functions of position: magnetic flux linkage and velocity. These functions are time-skewed so that next-time-step course corrections in flux linkage and velocity are defined as functions of previously-sampled data. We store these two time-skewed functions as digital lookup tables. As the solenoid travels toward closure, solenoid current is sampled; magnetic flux linkage is tracked by integration of a corrected voltage (inferred from the current and the PWM drive signal); position is computed from current and flux linkage; and velocity is computed from recent positions. A next-time-step target flux linkage is computed by table lookup for a currently-measured position. The measured velocity is compared to the time-skewed velocity target function of position, generating a velocity error signal, which is added to the target flux linkage. Finally, a PWM is computed that will bring the system from its recently-measured flux linkage to the velocity-corrected target flux linkage over the coming time interval. This procedure is robust, achieving soft landings in the presence of large perturbations.

Download: State space control of solenoids U.S. 7,099,136  Joseph Seale, Gary Bergstrom


Magnesense LLC Gorham,ME (207) 839-8637

©2009 Joseph Seale