US-20260127062-A1 - METHOD AND SYSTEM FOR HEALTH-AWARE DYNAMIC CONTINGENCY MANAGEMENT FOR AUTONOMOUS UNMANNED AIRCRAFT SYSTEMS

US20260127062A1US 20260127062 A1US20260127062 A1US 20260127062A1US-20260127062-A1

Abstract

A method and system for dynamic, health-aware contingency management for an autonomous Unmanned Aircraft System (UAS). A Fault Detection Module monitors a command-and-control (C2) link. A Prognostic Health Management (PHM) Module continuously calculates a Predicted Remaining Flight Time (RFT_Available) based on component health. In response to detecting a critical fault in the C2 link, a Contingency Decision Module (CDM) is initiated. The CDM accesses a Trajectory Option Set (TOS) of contingency trajectories (e.g., Return to Home, Divert) and dynamically calculates a Required RFT (RFT_Required) for each. The CDM selects an optimal trajectory by using the RFT_Available as a hard constraint, ensuring the RFT Available ≥RFT Required +SafetyMargin. This couples an external operational fault (C2 loss) with an internal health constraint (RFT) to ensure a fail-operational response

Inventors

Richard Joseph Mitchell

Assignees

Richard Joseph Mitchell

Dates

Publication Date: 20260507
Application Date: 20251108

Claims (10)

1 . A method for dynamic, health-aware contingency management for an autonomous Unmanned Aircraft System (UAS) operating with a command-and-control (C2) link, the method executed by an onboard computing unit (OCU) and comprising: a. Continuously monitoring a C2 link status via a Fault Detection Module; b. Continuously calculating, independent of the C2 link status, a Predicted Remaining Useful Life (RUL) of a mission-critical component via a Prognostic Health Management (PHM) Module, wherein the RUL is translated into a dynamic Predicted Remaining Flight Time (RFT_Available) for the UAS; c. Detecting a non-health-related critical fault condition related to the C2 link status, wherein the critical fault condition is either a total lost link or an incipient fault characterized by link degradation exceeding a threshold; d. In response to detecting the non-health-related critical fault condition, initiating a Contingency Decision Module (CDM), wherein the RFT_Available calculated in step (b) is not the trigger for initiating the CDM; e. Accessing a Trajectory Option Set (TOS) comprising a plurality of predefined contingency trajectories, including at least a Return to Home (RTH) trajectory and a Divert to Alternate Recovery Site (ARS) trajectory; f. Dynamically calculating a required Remaining Flight Time (RFT_Required) for each trajectory in the TOS based on current flight parameters and expected energy consumption; g. Comparing the RFT_Available against the RFT_Required for each trajectory using constraint satisfaction logic that incorporates a predefined safety margin, wherein the RFT_Available is used as a determinative, hard constraint for the trajectory selection and wherein the constraint satisfaction logic dictates selection only if the RFT Available ≥RFT Required +SafetyMargin; h. Selecting an optimal contingency trajectory from the TOS only if the RFT_Available satisfies the constraint against the RFT_Required for that trajectory; and i. Executing the selected optimal contingency trajectory autonomously.
2 . The method of claim 1 , wherein the mission-critical component is a lithium-ion battery system, and the PHM Module utilizes hybrid prognosis algorithms, including at least one data-driven regression technique and one state estimation technique, to calculate the RUL.
3 . The method of claim 2 , wherein the data-driven regression technique is selected from the group consisting of Support Vector Regression (SVR), Recurrent Neural Networks (RNN), and Multiple Linear Regression (MLR).
4 . The method of claim 1 , wherein the RUL calculation further provides confidence bounds, and the safety margin used in the constraint satisfaction logic is dynamically determined based on the provided confidence bounds.
5 . The method of claim 1 , wherein if the RFT_Available is determined to be less than the RFT_Required for the RTH trajectory, the CDM disregards the RTH trajectory and automatically selects the Divert to ARS trajectory corresponding to the lowest RFT_Required among all viable ARS options whose RFT constraint is satisfied.
6 . The method of claim 1 , wherein if the critical fault condition is the incipient fault and the RFT_Available exceeds the RFT_Required for the RTH trajectory by a predetermined high margin, the CDM selects a dynamically calculated Loiter maneuver optimized for link recovery and minimal energy expenditure.
7 . The method of claim 1 , wherein if the RFT_Available satisfies the constraint for the RTH trajectory but with a marginal safety margin, the selected optimal contingency trajectory is an Optimized RTH Trajectory, wherein the RUL input is utilized by the UAS flight controller to enforce energy-conserving flight parameters throughout the trajectory.
8 . A health-aware autonomous flight system for an Unmanned Aircraft System (UAS), configured to provide fail-operational contingency management upon a non-health-related C2 link loss, the system comprising: a. A C2 Communications Subsystem configured to transmit and receive command and control data; b. A Prognostic Health Management (PHM) Module hosted on an Onboard Computing Unit (OCU), the PHM Module configured to receive sensor data from a mission-critical component and calculate, independent of the C2 link status, a Predicted Remaining Useful Life (RUL) with associated confidence bounds, and to translate the RUL into a dynamic Predicted Remaining Flight Time (RFT_Available); c. A Fault Detection Module (HRFSA) configured to continuously monitor the integrity of the C2 link and generate a critical fault trigger upon detecting a non-health-related critical fault, said fault being either a total link loss or an incipient link degradation fault; and d. A Contingency Decision Module (CDM) hosted on the OCU and communicatively coupled to the PHM Module and the HRFSA, the CDM configured to: i. Receive the non-health-related critical fault trigger; ii. Wherein the RFT_Available is not the trigger for initiating the CDM; iii. Query the PHM Module for the RFT_Available; iv. Access a Trajectory Option Set (TOS) of predefined contingency maneuvers, each having a dynamically calculable RFT_Required; v. Execute constraint satisfaction logic to select an optimal contingency trajectory from the TOS by comparing the RFT_Available against the RFT_Required for each maneuver, ensuring the selected trajectory maintains a prescribed safety margin defined by the condition RFT Available ≥RFT Required +SafetyMargin; and vi. Interface with a flight control system to execute the selected optimal contingency trajectory autonomously.
9 . The system of claim 8 , wherein the PHM Module comprises a processor executing machine learning algorithms selected from the group consisting of Gaussian Process Regression (GPR) and Support Vector Regression (SVR) to enhance the accuracy and confidence bounds of the RUL calculation.
10 . The system of claim 8 , wherein the CDM's constraint satisfaction logic prioritizes selecting a Divert to Alternate Recovery Site (ARS) trajectory over a Return to Home (RTH) trajectory when the RFT_Available is insufficient to meet the RFT_Required of the RTH trajectory.

Description

BACKGROUND OF THE INVENTION 1. Technical Field The present invention resides in the technical fields of autonomous systems control, fault-tolerant flight systems, Unmanned Aircraft Systems (UAS) operating in complex airspace, Beyond Visual Line of Sight (BVLOS) operations, and advanced Prognostic and Health Management (PHM). More specifically, the invention relates to systems and methods that enable safe, scalable, and fail-operational autonomy by coupling the internal health state of the aircraft with external operational failure management. 2. Background Art The integration of UAS into the National Airspace System (NAS) requires robust regulatory frameworks demanding comprehensive procedures for handling contingencies, particularly the loss of the Command and Control (C2) link. Prior art systems generally address a lost link event by executing a predetermined, static contingency procedure. This typically involves either loitering in place for a pre-set time or initiating a fixed Return to Home (RTH) trajectory. Existing concepts for communicating contingency trajectories to Air Traffic Management (ATM) often rely on standardized messages, such as those leveraging the Trajectory Option Set (TOS), which communicate the contingency based primarily on the problem type (e.g., lost C2 link) without dynamic operational data. The critical deficiency of this static approach is its failure to account for dynamic, real-time variables that define the actual safety margins of the aircraft. Crucially, reliance on a fixed, pre-calculated endurance assumption—often based solely on initial State of Charge (SoC)—exposes the aircraft to catastrophic failure. If a UAS component, such as a lithium-ion battery, is degraded or if unpredicted external factors, such as high headwinds, significantly reduce the Remaining Flight Time (RFT), a fixed RTH trajectory may exceed the actual endurance of the aircraft, leading to a forced landing in an undesignated and potentially unsafe area. These static procedures fulfill a “fail-safe” requirement by terminating the mission in a planned manner, but they do not achieve the necessary “fail-operational” resilience required for truly scalable, complex operations. Prognostic Health Management (PHM) systems represent the state of the art in assessing component degradation and wear, facilitating Condition-Based Maintenance (CBM) by predicting the Remaining Useful Life (RUL). Significant advancements in RUL estimation utilize sophisticated analytical techniques, including hybrid prognosis models that incorporate data-driven methods like Gaussian Process Regression (GPR), Recurrent Neural Networks (RNN), and Support Vector Regression (SVR). Despite the high fidelity and maturity of onboard PHM systems, this RUL information is conventionally siloed. It is treated primarily as a logistical metric, used to inform long-term maintenance scheduling or provide simple alerts. The RUL data, while accurate and essential for anticipating future component failures, remains disconnected from the immediate, real-time flight control decision loop governing contingency resolution. Prior art systems for contingency path selection may impose non-health-related constraints, such as ensuring communication availability, but they fail to integrate the dynamic, predicted component degradation state (RUL) as a primary constraint. Prior art, such as patents on prognostics-enhanced automated contingency management for vehicles and PHM for electro-mechanical systems (e.g., US8306778B2), focuses on general health monitoring or static responses without dynamic coupling to operational failures in UAS. Similarly, research on battery RUL prediction (e.g., LSTM-GPR hybrids in MDPI articles) emphasizes prognostic accuracy but does not apply RUL as a real-time constraint for trajectory selection in autonomous flight. The non-obvious inventive step is the establishment of a synergistic process that dynamically couples the internal health state (RUL/RFT) with the external operational failure management (C2 lost link detection) to produce a validated, fail-operational response. This integration elevates RUL from a maintenance indicator to a safety-critical, hard operational constraint for instantaneous decision-making, differentiating the disclosed technology from existing systems. BRIEF SUMMARY OF THE INVENTION The present invention provides a Method and System for Health-Aware Dynamic Contingency Management, establishing a truly fail-operational autonomous agent through the integrated deployment of three key functional modules: the Prognostic Health Management (PHM) Module, the Fault Detection Module (referred to herein as the HRFSA Reliability Platform), and the Contingency Decision Module (CDM). Upon the HRFSA detecting a C2 link fault—which may include an incipient fault signifying pre-failure signal degradation—the CDM is immediately activated. The CDM executes automated reasoning by querying the PHM Module for the current, dyn