System Design 1 - Chapter 8

08 July 2025, Bastian Luettig

Platform Management: Actor

Requirements

The platform management actor is suitable for the redundant computer, if it gurantees that the redundant computer ($M_{\text{rfcc}}$) correctly controls the actuators in all valid operating scenarios. This is achieved if:

PLAMA:ACTOR ensures actuator consensus\

Fundamentals

Challenges and concepts

Actuators may have components that exist individually for each commanding computing lane (Singe Actuator Components). e.g. position sensors, current sensors, solenoids

An actuator will have components that are common to all commandung computing lanes (Common Actuator Components). e.g. direct drive valve, mode valve, control piston

See previous lecture for more details on Fundamentals and Tasks

Software Architecture

There are four main software tasks:

Drift Correction and Pseudo-Discretes: A similar concept to PLAMA:LAW: the actuator control algorithm is itself a control law, that may contain a BIBO-instable parts and pseudo-discretes, that PLAMA:ACTOR must compensate for.

Cross-Single-Computer-Monitoring: After PLAMA:Actor ensured computer-replica-determinism, the redundant system can now check for individual lane faults: each lane should command the same value. If it does not, it is faulty. Here we see the command as sent by the software, not the actual electrical command that we measure using sensors

Handling of Failure Indications set by Models: Whenever the actual movement diverges too much from the model: PLAMA:ACTOR suspects an actuator failure.

Consistency / Resource Manager: The resource manager will passivate an actuator if it does not meet the required minimum components (i.e. one common component off or too many single components off)

Actor Control Law and Digital Twins

Model for each common actuator component - The system measures $y_\text{ddv}$ which the actuator control function can access and use

Mathematic model for calculating actuator lane interctions: simplifications

  1. Magnetic flux for each coil independent
  2. Resulting anchor force is simply the sum of anchor forces
  3. Very little anchor movement: anchor force independent from position
  4. negligible magnetic induction within coil as a result fromm anchor movement
  5. oil is in-compressible

Example: coil 1 does not work anymore

graph

Reaction: all intact $sf_o$ detect that current for coil 1 differs too much and thus passivate the current sensor and the coil

$$s_\text{off, sac, coild, 1} = true; s_\text{off, ss, coil, 1} = true$$