The Attitude Systems Operation of Computer Sciences Corporation's System Sciences Division provides attitude analysis and operations support for the Attitude Determination and Control Section of NASA's Goddard Space Flight Center. In contrast to the general literature survey, the recent analytic work of this group falls into five major areas: (1) development of more sophisticated procedures for obtaining high reliability with star sensors for missions such as HEAO or MAGSAT, (2) improvements in understanding attitude geometry, primarily for spin-stabilized spacecraft such as AE, SMS/GOES, CTS, SIRIO, IUE, and ISEE; (3) development of procedures for obtaining increased accuracy from Earth horizon sensors, both for spin-stabilized spacecraft and Earth-oriented spacecraft such as SEASAT, HCMM, DE, or MAGSAT; (4) detailed planning for attitude acquisition maneuvers for spacecraft which undergo major changes in determination and control procedures between launch and the initiation of normal mission operations, such as RAE, CTS, GEOS, or HCMM; (5) development of procedures for processing the increased volume of attitude data that is anticipated with the increased launch potential of the Space Shuttle in the 1980s; and (6) satisfying the increasingly stringent attitude determination and control requirements and ensuring the quality of computed attitude solutions.

The remainder of this section discusses developments which are necessary for the continued evolution of spacecraft attitude determination and control techniques. We have divided these developments into five categories: quality assurance.

sensor design, hardware standardization, software standardization, and basic analysis.

Quality Assurance. A major problem that must be resolved in the Shuttle era is quality assurance—the designing of software systems that will provide accurate attitude information without operator intervention under normal circumstances and that will recognize abnormal circumstances. The need for increased quality assurance comes from four principal changes in future attitude operations: increased autonomy with the use of onboard computers, increased accuracy requirements, increased data volume (both more spacecraft and more data per spacecraft), and the need for reduced costs. Increased data volume and reduced costs require a system with minimal operator intervention. Increased accuracy implies more sophisticated modeling procedures, more potentially adjustable attitude parameters, and more complex analysis and filtering procedures. Finally, increased autonomy implies software which executes with minimal external intervention and which is capable of recognizing abnormal data.

One solution to quality assurance and increased autonomy may lie in multiple component systems such as that described for the CTS spacecraft in Section 21.2. Here a deterministic processor (or a differential corrector with only a limited number of state vector elements) is used for normal operations either on the ground or on board the spacecraft, and a more sophisticated ground-based differential corrector is used occasionally for bias determination. The bias parameters determined in the latter system are then used as input parameters to the "normal" processor to perform routine operations. The character of the routine operations may be monitored both by internal checks and flags and by summary displays that permit an operator to gain an overview of the system operation and to examine in more detail any abnormal data segments. However, in the CTS system, even the telemetry processor and deterministic attitude component require operator intervention and interactive graphics for effective use.

A second possible solution may lie in the direction of a hybrid or evolutionary attitude system. Initially, such a system behaves like a very flexible data filter as described in Chapters 13, 14, 20, and 21, with operator control at essentially every stage via interactive graphics. As biases are resolved and the data quality and nature of the most common anomalies are determined, specific automatic options are chosen and the system becomes increasingly autonomous. Operator intervention is required only when anomalies are encountered, which are identified by a series of checks and flags in the data. After the most common anomalies have been identified, automatic procedures for handling these are initiated (via, for example, alternative processing parameters when specific anomalous conditions are encountered). At this point, the system is essentially fully autonomous—having been "designed" by the. experience of the operator to handle the particular data characteristics specific to that mission. The system then operates routinely in a noninteractive mode* with occasional interactive runs to ensure that the system is behaving properly or to account for changing mission conditions.

Sensor Design. One of the principal requirements for the future is a need for attitude analysis at earlier phases of mission planning and hardware design— particularly when attitude accuracy requirements approach the limit of sensor

* For complex ground-based systems, a low level of continuous, interactive control may be necessary to ci 'ire the availability of data sets and hardware devices.

accuracies. Characteristic of this need for early analysis is a dichotomy between two distinct procedures for obtaining attitude measurements—the use of simple sensors with complex output versus the use of complex sensors with simple output. This distinction may be illustrated by the Sun sensors used for SIRIO and the SMS/GOES missions (Chen, et «/., [1976]; Chen and Wertz [1975ft. During the transfer orbit to synchronous altitude, both satellites are spin stabilized and use Sun sensors consisting of approximately V-shaped slits as shown in Fig. 22-2. The Sun angle is determined by measuring the time between Sun sightings for the two slits. The relationship between the Sun angle, /?, and ratio of the time between Sun sightings, A/, to the spin period, P (determined by the time between Sun sightings for the vertical slit), for both sensor types is shown in Fig. 22-3. By examination, it is clear that the SIRIO system is a relatively simple sensor with an output that is at least moderately complex to interpret, since it becomes very nonlinear as the Sun moves toward the spin axis. In contrast, the SMS/GOES Sun sensor has a more complex structure, but the output signal is nearly linear over the sensor's range of performance.

Fig. 22-3. Sun Angle Versus Spin Angle for SMS and SIRIO Sun Sensors. The spin angle is the azimuthal rotation of the spacecraft between observations of the Sun by the two sensor components.

Fig. 22-2. Shape of SMS Sun Sensor Mask and Relative Orientation of Slits in the SIRIO Sun Sensor

Fig. 22-3. Sun Angle Versus Spin Angle for SMS and SIRIO Sun Sensors. The spin angle is the azimuthal rotation of the spacecraft between observations of the Sun by the two sensor components.

The relative advantages of the two sensor types do not become apparent until we ask what the measurement is to be used for and how it is to be processed. If the Sun angle is to be measured by simply attaching a scale to the sensor output with no analysis possible, as might be required in a simple display device or analog use of the data for onboard attitude control, then the SMS/GOES configuration is superior because of the linear output. However, in any attitude determination system or display device where there is software available for processing, the SIRIO design becomes distinctly superior. First, a straightforward analysis of the spherical geometry involved in the SIRIO sensor (Section 7.1) shows that tan/? = tan 0/sin(A//2wP) (22-1)

where 0 is the angle between the two linear slits in the SIRIO design. In contrast, the SMS/GOES design requires a table of calibration values which must be stored and interpolated to compute the Sun angle. In addition, the SIRIO sensor is very amenable to bias determination and in-flight calibration. If we assume only that the slits have indeed been made linear, then only three bias parameters fully characterize the relative orientation between the two slits and the spin axis (Section 7.1). In principle, these parameters can be established as elements of a state vector or determined manually from several data segments. Thus, it is at least possible to have a very accurate in-flight calibration of the SIRIO Sun sensor parameters. (In practice this may be impossible for any specific mission because of limitations in the amount or quality of available attitude data, the geometry, or mission timeline constraints.) In contrast, there is no analytic procedure for general determination of bias parameters for the SMS/GOES Sun sensor because each segment of the calibration curve would have to be corrected separately. We may carry out a bias determination procedure at any particular Sun angle, but extrapolation of the results to other Sun angles is tenuous at best.

As is clear from the above example, the complex sensor with a simple output may be preferable in systems such as limited-capacity onboard processors or control center displays where the output must be used directly with no algebraic manipulation, bias determination, or state estimation. However, the simple sensor with output which can be analytically modeled is preferred whenever there is software available for processing or whenever bias determination or state estimation techniques are available to increase attitude accuracies. This would include spacecraft using onboard processors for which biases can be telemetered up from the ground, as described for SMM below. Thus, high accuracy requirements would suggest the need for sensors which can be modeled analytically to make the best use of sophisticated data analysis techniques.

Hardware Standardization. Related to the problem of sensor design is the need for software systems to provide greater reliability and reduced costs and to permit more effort to be applied to the new and unique problems which arise. The principal problem in the standardization of hardware is that missions have widely varying requirements and hardware systems have normally been designed to meet specific mission conditions at minimum cost. Therefore, the main precursor to standard sensors or standard software is the development of a hardware system with sufficient flexibility to meet the requirements of many missions. Two basic approaches to this problem are available: to work with combinations of existing hardware or to design new hardware with the specific intent of designing flexibility.

The use of combinations of existing hardware is a major goal of the MultiMission Spacecraft, MMS, series, the first of which will be the Solar Maximum Mission, SMM. As shown in Fig. 22-4, the MMS spacecraft consists of three standard modules (Power, Attitude Control, and Communications and Data Handling) in a triangular frame with space for the payload equipment at one end [GSFC, 1975]. The attitude control subsystem shown in Fig. 22-5 contains the following attitude sensing equipment: a set of three two-degree-of-freedom gyroscopes; two Ball Brothers CT 401 star trackers (described in Section 6.6); one precision digital Sun sensor; and three orthogonal magnetometers. All MMS spacecraft will have a coarse Sun sensor system, and a high-accuracy payload sensor may also be used for some missions. Attitude control will normally be provided by a set of three orthogonal, 20 N-m-s reaction wheels (possibly with a

0 0

Post a comment