Simulation of knowledge enabled monitoring and diagnosis tool for Mars lander payloads

This activity addresses the current limitations, proposing the use of knowledge-enabled technologies to increase the robustness of the mission monitoring and control systems. This will involve the use of a complementary diagnostic system that will use explicit knowledge models to enhance the real-time fault diagnostic of the existing systems developed in the frame of past Aurora activities. The current activity is proposed in the context of a more ambitious roadmap to bring more intelligence and autonomy to deep space missions. In the Aurora context, the project will be integrated within the autonomy domain. It will analyse the feasibility of its application to the Mars Sample Return (MSR) mission, focusing on the MSR lander payload.

In order to balance the stringent requirements for reliable autonomous operation and the constrained resources of the spacecraft, an iterative and phased approach is proposed. The use of knowledge-enabled operations will evolve gradually from non-critical monitoring activities to the mission-critical control, as the robustness of the approach is tested and approved. Real autonomy must then be achieved at payload level first, for example within Pasteur. This itself covers two aspects. The first aspect is to increase onboard diagnosis capabilities by the use of explicit and implicit knowledge and models of subsystems. Therefore, it is firstly essential to develop the ability to effectively recognise that anomaly, malfunction or an unpredicted event is taking place. Tools must be developed in order to create onboard diagnosis of instruments based on unsupervised modelling and hence capable to monitor their evolving behaviour, detecting subtle deviations and thus predicting failure.

Once the critical condition is predicted or recognised it will be possible to re-plan the mission control operations and isolate/protect the ’critical-subsystem’, which refers to a variable subset of the system that currently supports the execution of the most relevant aspects of the mission. It will be possible to create systems capable of continuously and autonomously identifying minimal requirements for mission execution (thus optimising resources on board) and capable, as well, of optimising dynamically the configuration of degraded modes. This aspect is complementary but not a part of the proposed activity. It is being addressed within the context of other Aurora activities.

A number of studies and prototypes have looked at the possibility of increasing the autonomy of the mission onboard control systems. Although a number of positive results have been achieved using different knowledge-based technologies for diagnosis and control tools (for example rule-based reasoning, expert systems, data mining and fuzzy logic), the real strength of these technologies lies in their combined operation. Therefore, this activity is proposed to demonstrate the capability of European industry and academia to build a monitoring and diagnosis software module for a complex Mars lander payload.

The approach is to build a simulation environment including the nominal behaviour of Mars lander instruments, capability of pseudo-fault injection in the form of slight modifications of the behaviour, and a knowledge based module for monitoring and diagnosis. The simulator will use the existing payload model to be refined in the future.

The main tasks foreseen are the following:

• Review and analysis of the currently available Mars Sample Return lander payload model and its instruments
• Assessment of the knowledge based techniques that will support the monitoring and diagnosis software module
• Review, analysis and trade-off on simulations tools; selection of the most appropriate simulation environment for this activity
• Modelling of Mars Sample Return lander payload instruments in the simulation environment
• Implementation of pseudo-fault injection capabilities in the payload model in the simulation environment
• Implementation of the monitoring and diagnosis module in the simulation environment
• Demonstration of the capability of the Monitoring and Diagnosis module to detect abnormal behaviour in Pasteur introduced via pseudo-fault injection
• Study synthesis and recommendations

Executive Summary

Download the Executive Summary (PDF file - 465 KB) »»