Engineering topics
Publications and source records attributed to Vandervelde, W. E..
Progress in robust design of generalized parity relations, design of failure sensitive observers using the geometric system theory of Wonham, computational techniques for evaluation of the performance of control systems with fault tolerance and redundancy management features, and the design and evaluation od control systems for structures having nonlinear joints are described.
Efforts in this reporting period were centered on four research tasks: design of failure detection filters for robust performance in the presence of modeling errors, design of generalized parity relations for robust performance in the presence of modeling errors, design of failure sensitive observers using the geometric system theory of Wonham, and computational techniques for evaluation of the performance of control systems with fault tolerance and redundancy management
Progress in robust failure detection and isolation, control system reconfiguration, and performance evaluation of systems having redundancy management features is briefly summarized.
This report analyzes the reliability of NASA's Ultra-reliable Fault Tolerant Control System (UFTCS) architecture as it is currently envisioned for helicopter control. The analysis is extended to air transport and spacecraft control using the same computational and voter modules applied within the UFTCS architecture. The system reliability is calculated for several points in the helicopter, air transport, and space flight missions when there are initially 4, 5, and 6 operating channels. Sensitivity analyses are used to explore the effects of sensor failure rates and different system configurations at the 10 hour point of the helicopter mission. These analyses show that the primary limitation to system reliability is the number of flux windings on each flux summer (4 are assumed for the baseline case). Tables of system reliability at the 10 hour point are provided to allow designers to choose a configuration to meet specified reliability goals.
Rather than continuing to focus on software development projects per se, the system dynamics modeling approach outlined is extended to investigate a broader set of issues pertaining to the software development organization. Rather than trace the life cycle(s) of one or more software projects, the focus is on the operations of a software development department as a continuous stream of software products are developed, placed into operation, and maintained. A number of research questions are ""ripe'' for investigating including: (1) the efficacy of different organizational structures in different software development environments, (2) personnel turnover, (3) impact of management approaches such as management by objectives, and (4) the organizational/environmental determinants of productivity.
The unreliability of control system components was investigated in our attempt to deal with that problem. This matter is of concern in large space structure control because of the large number of components required to achieve specified performance in some situations, and the long operating period required between maintenance visits. The detection and isolation of component failures during system operation, and algorithms for reconfiguring control systems following detection and isolation of a failure were emphasized.
A methodology has been presented which is intended to assist the designer of a control system for a large space structure to decide how many sensors should be incorporated in the system and where they should be placed on the structure. This approach is intended to be especially useful in the early stages of the evolution of the system, before a complete control system concept has been defined. This methodology is based on a quantitative measure of the observability of the system for a given set of sensors. The effect of possible component failures during the mission period was incorporated in the measure. The question of sensor placement is then resolved by finding the locations which maximize the performance measure. The number of components to use cannot be determined by optimizing the measure because the observability always improves with increased number of sensors if they are optimally located. However, the improvement in the measure with component number can be determined, and this information can be used along with data on cost, power required, etc., to decide how many sensors to use.
Studies are carried out to assist the designer of the control system for a large flexible space structure in his choice of how many actuators and sensors to incorporate in the system, and where to locate them on the structure. The degree of controllability and minimum control energy strategy for driving the system are described.
A decision making methodology is presented which is intended to be useful in the early stages of system design, before a control system is designed in detail. The methodology accounts for the likelihood of failure among the sensors and actuators in a control system. A method to compute the degree of controllability and degree of observability of a system for a given set of actuators and sensors is also presented.
The degree of controllability of a large space structure is found by a four step procedure: (1) finding the minimum control energy for driving the system from a given initial state to the origin in the prescribed time; (2) finding the region of initial state which can be driven to the origin with constrained control energy and time using optimal control strategy; (3) scaling the axes so that a unit displacement in every direction is equally important to control; and (4) finding the linear measurement of the weighted "volume" of the ellipsoid in the equicontrol space. For observability, the error covariance must be reduced toward zero using measurements optimally, and the criterion must be standardized by the magnitude of tolerable errors. The results obtained using these methods are applied to the vibration modes of a free-free beam.
This paper considers reliability in designing control systems for large orbital structures that are expected to have appreciable flexibility. Reliability is considered for both the control configuration (i.e. selecting sensor and actuator locations over an admissible set) and the ultimate operation of the system. The approach presented is to construct a cost function that indicates the absolute goal of the system and, for a given structural design, to determine the lowest achievable costs conditioned on the failure states of the system (i.e. which actuators and sensors are operational). These costs are then weighted and summed to provide an overall system performance measure. This measure is then minimized over the set of admissible control configurations. The approach is illustrated using a beam that has four actuators with the goal of achieving a given parabolic shape.