Thruster-based control typically requires a substantial engineering effort in order to mitigate an important issue: the control system tends to use many thruster pulses to accomplish what a single thruster pulse could do instead. This fundamental problem arises because the standard controller’s feedback response is (by definition) proportional to the spacecraft’s current tracking error state alone, without regard for the time-integrated tracking error. By ignoring this particular signal, it becomes especially difficult to build a control system that realizes the following performance capabilities (PCs):
PC1: The control system should minimize the number of pulses needed to maintain a specified level of absolute tracking error accuracy.
PC2: The control system should nullify the time-averaged tracking error signal while maintaining the specified absolute tracking accuracy.
State-of-the-art thruster-based control systems do not currently offer both performance capabilities (PC1, PC2) in a unified, single-mode controller package. Yet, the importance of providing PC1 control capability is clearly underscored by the following factors: (a) Hydrazine thrusters are commonly used in the aerospace industry; (b) these actuators are inherently pulse-count limited due to valve-actuator life; and (c) this ongoing issue continues to make a significant impact on the spacecraft’s mission life.
Regarding PC2, this feature plays an important role in many spacecraft pointing and proximity operations where disturbance rejection (via integral control action) is needed to ensure that the spacecraft is able to track mission-specific command trajectories in the presence of severe/unpredictable disturbances (e.g., comet outgassing, rendezvous/ docking/landing contact dynamics, etc.). Also, spacecraft trajectory-correction maneuvers (TCMs) incur execution errors that are effectively minimized whenever the average spacecraft pointing error is zero during the delta-V burn phase; that is, when PC2 holds. In this work, a novel thruster-based control algorithm that ensures PC1 and PC2 is proposed.
Although the standard proportional-integral-derivative (PID) control strategy has historically been applied to this problem, the standard approaches have proven to be problematic because the thrusters must remain off while the tracking error is within the absolute accuracy limits, i.e., the dead-zone. (Dead-zones are needed to accommodate the thruster’s minimum ON-time hardware limitation.) Hence, with standard PID, time-integrated tracking error biases can grow unbounded (while inside the dead-zone), causing the control system to become unstable. Moreover, this issue cannot be mitigated by the inclusion of (industry-standard) anti-windup. Indeed, when the tracking-error signal is inside the dead-zone, standard PID (with or without anti-windup) merely accelerates the growth-rate of this signal towards the edge of the dead-zone (without regard for adjusting the minimum thruster pulse width accordingly), resulting in excessive pulse firings.
In the proposed approach, the aforementioned issues are eliminated because the minimum thruster pulse width is adjusted dynamically for purposes of reducing the pulse count while ensuring closed-loop stability. To do so, the disturbance torque on the spacecraft must be learned in real time, and a novel integral control action is developed to achieve this end. Effectively, the control system remains idle inside the dead-zone, and uses single-pulse thruster events to realign (and stabilize) the spacecraft in a manner that minimizes the time-averaged tracking error.