Output Corridor Impulsive Control of First-order Continuous System with Non-local Attractivity Analysis

1. Problem Formulation

The paper addresses the control of a scalar, continuous, time-invariant linear plant (a first-order system), which serves as a fundamental model for chemical kinetics and pharmacokinetics (single-compartment models). The system is described by:
$$\dot{x}(t) = ax(t), \quad a < 0$$
The objective is to design an impulsive feedback controller that confines the system's output $x(t)$ within a predefined corridor $[x^*_{min}, x^*_{max}]$.

Unlike traditional setpoint stabilization, this problem requires maintaining the state within a range rather than at a single point. The controller operates by injecting discrete impulses (doses) at specific time instants $t_n$. The control law is defined by two modulation functions:

• Amplitude Modulation ($F$): Determines the size of the impulse $\lambda_n = F(x(t_n^-))$.
• Frequency Modulation ($\Phi$): Determines the inter-dose interval $T_n = \Phi(x(t_n^-))$.

The resulting closed-loop system is a hybrid system (continuous dynamics between impulses, discrete jumps at impulses). The challenge is to design $F(\cdot)$ and $\Phi(\cdot)$ such that the system converges to a periodic solution (a 1-cycle) whose orbit exactly spans the desired corridor.

2. Methodology

The authors propose a methodology based on the analysis of a discrete return map (Poincaré map) derived from the hybrid dynamics.

A. The Return Map and 1-Cycles

The state of the system immediately before the $n$-th impulse, denoted $x_n = x(t_n^-)$, evolves according to a discrete recursion:
$$x_{n+1} = Q(x_n) = e^{a\Phi(x_n)}(x_n + F(x_n))$$
A 1-cycle corresponds to a fixed point $x^*$ of this map ($x^* = Q(x^*)$). The authors derive explicit formulas for the period $T$ and impulse weight $\lambda$ of a 1-cycle that "spans" the corridor (i.e., the minimum and maximum values of the continuous trajectory between impulses coincide with $x^*_{min}$ and $x^*_{max}$):
$$T = \frac{\ln(x^*_{min}) - \ln(x^*_{max})}{a}, \quad \lambda = x^*_{max} - x^*_{min}$$
The modulation functions $F$ and $\Phi$ are then designed to satisfy the interpolation constraints $F(x^*) = \lambda$ and $\Phi(x^*) = T$.

B. Stability and Attractivity Analysis

The paper investigates both local and global attractivity of the designed 1-cycle:

1. Global Attractivity: Using the Banach Contraction Principle, the authors establish conditions under which the map $Q(x)$ is a contraction ($|Q'(x)| \leq q < 1$). This guarantees that all admissible solutions converge exponentially to the unique 1-cycle.
2. Super-Exponential (Deadbeat) Convergence: The authors show that if the derivative of the map at the fixed point is zero ($Q'(x^*) = 0$), the convergence becomes super-exponential (quadratic). In specific scenarios (e.g., event-triggered initial conditions), this allows for finite-time convergence (deadbeat control), where the system reaches the corridor in a single step.
3. Negative Feedback Case: The paper analyzes the specific case where $F$ is non-increasing and $\Phi$ is non-decreasing (standard negative feedback). For this case, they derive tighter ultimate bounds using an iterative procedure involving a function $\Psi(x)$, proving that the system's state is confined within a specific invariant interval $[m^*, M^*]$.

3. Key Contributions

1. Analytic Solution for Corridor Control: The paper provides an explicit analytic expression for the periodic solution (1-cycle) that tightly satisfies the corridor constraints, solving the problem for the simplest single-compartment kinetics.
2. Global Attractivity Conditions: New conditions are derived to ensure the global attractivity of the 1-cycle, moving beyond the local stability guarantees often found in existing literature.
3. Super-Exponential Convergence: The authors demonstrate that proper design of the feedback gains can yield $Q'(x^*) = 0$, leading to ultra-fast convergence that is practically indistinguishable from finite-time convergence.
4. Application to Pharmacometrics: The theoretical framework is applied to intravenous paracetamol dosing, a critical medical application where maintaining drug concentration within a safe and effective range is vital to avoid toxicity (liver damage) and under-dosing (pain).

4. Results and Simulation

The proposed approach was validated through numerical simulations of a Paracetamol PK/PD model:

• Open-Loop Failure: Standard open-loop dosing (fixed intervals and doses) resulted in the drug concentration frequently falling outside the safe corridor (under-dosing and overdosing).
• Closed-Loop Success: The designed impulsive controller successfully confined the drug concentration to the target corridor $[10, 20]$ mg/L.
• Case Studies:
  • Case 1 (Deadbeat): By selecting specific feedback gains, the controller achieved convergence to the corridor in finite time (one step) after an initial bolus, provided the initial state was within a reachable range.
  • Case 2 & 3: Variations in modulation functions (pure frequency modulation vs. combined amplitude/frequency) showed that while all cases converged, the specific design of the slopes ($F'$ and $\Phi'$) significantly impacted the convergence rate and the tightness of the bounds.
• Bounds Verification: The simulation confirmed that the theoretical bounds derived from the contraction mapping and the iterative $\Psi$-method accurately predicted the system's behavior, with the designed controller achieving the tightest possible bounds (coinciding with the corridor limits).

5. Significance

This paper is significant for several reasons:

• Bridging Theory and Practice: It connects abstract impulsive control theory with practical medical dosing problems, offering a rigorous alternative to Model Predictive Control (MPC) for systems where discrete dosing is the only option.
• Hybrid Dynamics Insight: It advances the understanding of hybrid systems by explicitly solving the "corridor control" problem for first-order systems, a class ubiquitous in engineering and biology.
• Safety and Efficacy: By guaranteeing that drug concentrations stay within a predefined safe corridor, the method directly addresses patient safety concerns in pharmacotherapy, preventing both toxic accumulation and therapeutic failure.
• Efficiency: The demonstration of finite-time (deadbeat) convergence suggests that such controllers can rapidly correct deviations, which is crucial in emergency medical scenarios or acute pain management.

In summary, the paper presents a robust, mathematically rigorous framework for designing impulsive controllers that maintain system states within safe operational corridors, with a specific and successful application to optimizing intravenous drug dosing.