Hello,

I am using a Kalman-Filter for a System with a watertank . Theres a known voltage input signal U_pump, that is used to control a waterpump. The waterpumps flowrate is proportional to the voltage U_Pump, so the relationship between U_Pump and the flowrate Q is Q = K * U_Pump.

As you can see, this is just a very easy model and probably not very accurate. So I have a Sensor, and want to decrease the uncertainty in this model by using the sensor to update the flowrate Q.

Therefore I have set Q as a state.

So the Differential Equation for the Flowrate Q is.

Q_dot = 0* Q + 0*U_Pump;

with A = 0; B = 0;

Because Q_dot is just the Rate at which Q changes over time, I cant put the K * U_Pump in there.

This means, in the discrete form:

Q_k = Q_k-1, which is a static model for the Flowrate Q, although I have the model Q = K*U_Pump

So my predicted Measurement of Q with the Kalman-Filter is just

Q_k_predicted = 1 * Q_k-1; which is not correct when I am adjusting the Voltage U at this timestep, therefore I can only use this for a static Voltage.

So is there a way to update U_Pump with the sensor value, or is there non?