Implementing a Kalman Filter with multiple input sensors

Hello everyone,

I am trying to develop a model in SIMULINK of a discrete Kalman Filter in which 2 different sensors are used as inputs for computing the ‘’innovation’’ (within the correction part of the algorithm).

Each of these two sensors has different sampling periods (0.05 seconds and 1 second respectively) and so, they update the innovation part of the algorithm at these specific time steps.

If there are no measurements available, the algorithm keeps running based on the previous predicted measurement computed by the prediction part of the Kalman filter.

To sum up, the problem I have is that for each different time step, I have a different measurement matrix H in which the number of columns is fixed and the number of rows varies from time step to time step.

Therefore, every 0.05 seconds I have a 2x7 matrix and every second a 5x7 matrix. I have tried to build a simple model containing a clock block, the aforementioned H matrix, a selector and a matlab script that outputs the index of the H matrix for each time step. The problem is that the index variable changes in size which simulink does not support and so, this model crashes.

May anyone help me to figure out how I can implement this?

Answer 1

[John Petersen]

You can keep the same size H matrix, just zero out the rows for which there is no measurement.

Answer 2

[suresh babu chandanapalli]

sir,please send the mat lab code and simulink