State Estimation for Electromagnetic Actuators

In order to realize advanced control or enable enhanced diagnostic opportunities, knowledge of the system state is a key requirement. The use of additional sensors is, in general, prohibited due to cost or space restrictions, or at least undesired due to reliability reasons. To this end research in the area of sensorless position estimation, or more precisely, state observation based on quantities that can easily be measured such as voltage and electric current is ongoing since several years. With the necessary computational power at hand, sophisticated algorithms can be executed in real time even in low-cost application. However, a robust and general solution of the sensorless state estimation problem in solenoid actuators does not seem to be available so far.

In the literature the problem has been approached in numerous ways. It can be readily understood that the types of algorithm that can be used have to be selected according to the application. In certain areas, e.g. in linear magnetic actuators and proportional valves, the solenoid is driven by a more or less slowly varying current and the task is to determine the position of the armature even when it is at rest at an arbitrary position. We refer to this type of application as the slow moving armature case, where inference from electrical signals to position is typically made by parameter estimation, from which the position can be reconstructed.

In this type of applications the pulse width modulated voltage signal is used for driving the desired current through the coil and concurrently provides a persistent excitation in order to evoke certain features. These can be used to estimate electrical parameters. Suitable parameters for position estimation are the position dependent main inductance and the also position dependent eddy current resistance, or jointly the impedance of the circuit. In other suggested solutions, features of the current ripple wave form, e.g. the slopes are used directly. It has also been suggested to provide an additional carrier signal to support the extraction of the electrical features required for position estimation.

In contrast to the slow motion case, in a vast quantity of applications, e.g. in digital hydraulics, the armature is moving very fast for a time span in the millisecond range or even below. Although in these applications the current control is done with a pulse width modulated voltage signal as well, the use of parameter estimation based methods seems to be much more difficult. This is on the one hand due to the required time, a parameter estimation based scheme needs to track the changing parameters. On the other hand, and more important, the fast motion of an armature in the magnetic field leads, according to Faraday's law, to an induction of a voltage that superimposes the voltage induced by the change of the driving current. Therefore, the parameter changes based on current, position and, in particular, the velocity. Thus, a unique solution for position estimation just based on the typical parameters mentioned above is infeasible. Therefore, for applications with fast moving armatures it is appropriate not to estimate parameters but to use a different model based approach. In the deployed model the interaction between the electrical quantities like energizing and eddy currents and the mechanical quantities position and velocity has to be reflected with sufficient accuracy. Then, the not measured mechanical quantities can be estimated with reasonable accuracy.

Publikationen

• Sensorless Position Estimation of Magnetic Actuators, J. Reuter, T. Braun, F. Straußberger, 2015
• State Estimation for Fast-Switching Solenoid Valves: a Study on Practical Nonlinear Observers and New Experimental Results, T. Braun, F. Straußberger, J. Reuter, 2015
• Position Estimation in Electro-Magnetic Actuators Using a Modified Discrete Time Class A/B Model Reference Approach, F. Straußberger, M. Schwab, T. Braun, J. Reuter, 2014
• New Results for Position Estimation in Electro-Magnetic Actuators Using a Modified Discrete Time Class A/B Model Reference Approach, F. Straußberger, M. Schwab, T. Braun, J. Reuter, 2014
• Intermediate Linear Voltage Control for Motion State Detection in Solenoid Valves: a Lyapunov Approach, M. Schwab, T. Braun, F. Straußberger, J. Reuter, 2014
• State Estimation for Fast-Switching Solenoid Valves - A Nonlinear Sliding-Mode-Observer Approach, T. Braun, M. Schwab, F. Straußberger, J. Reuter, 2014
• Nonlinear Observer For A Dual Coil Solenoid Valve, M. Jäkle, J. Reuter, 2012

Zurück zu Forschung