GEOSTATIONARY STATION-KEEPING BY ION THRUSTERS: GENETIC ALGORITHMS OPTIMIZATION

The orbital control of geostationary satellites requires the periodic execution of manoeuvres for gravitational and solar radiation perturbation compensation. To maintain a fine control of the satellite position, it is usually sufficient to perform orbit corrections every two weeks, firing chemical thrusters. The operational load for the manoeuvre's planning and execution is usually quite low. Innovative technologies for station-keeping are available now, based on high efficiency ion thrusters, which allow a significant propellant saving. They are included in Artemis, a telecommunications satellite developed by the European Space Agency. When low acceleration ion thrusters are adopted, a near continuous orbit control is required; the resulting long firing arcs can be in conflict with some operational or technical constraints. The station-keeping planning is a critical process, since it is not based on periodic impulsive manoeuvres, but rather on quasi-continuous firings. Furthermore the orbit control should be planned more on a seasonal or a yearly cycle rather than on short weekly cycles. Since each manoeuvre is defined by its start time and duration, the total number of parameters involved in a station-keeping optimization process amounts to several tens. The presence of many variables and logical constraints makes the use of classical optimization methods quite complex. The model proposed here is based on genetic algorithms, which allows the convergence to an optimum solution through successive iterations, in which a random set of solutions is progressively selected on the basis of a factor of merit tailored to the target. This paper presents some innovative concepts for station-keeping optimization when ion propulsion is adopted for orbital control; furthermore, the general properties of genetic algorithms are discussed, together with the results obtained in the specific station-keeping application.     

MULTIRESPONSE OPTIMIZATION of ACID CASEIN PRODUCTION

A multivariate skimmilk extrusion process, designed to produce an acid coprecipitate was studied in terms of minimizing residual lactose, ash and fines. an experimental model system was utilized to simulate the extrusion process and evaluated using response surface methodology to assess the relationship between the responses (fines, residual lactose and minerals) and the process variables (concentration, pH, temperature, agitation, washing time and wash water ratio). Compromise optimum conditions were derived using the Generalized Distance Approach (GDA) and an Extended Response Surface Procedure (ERSP) which made use of the SAS RSREG procedure with and without constraints. the GDA procedure produced good results in terms of providing an optimum for a general acid casein process, while the ERSP allowed more extensive analysis of the data in terms of assessing specific processing conditions. Although more computing intensive, the ERSP conferred additional flexibility in determining optimal conditions for special situations such as extrusion processing. Both approaches are useful for process engineering, with the GDA being a more general tool while the ERSP is advantageous when the GDA procedure becomes limiting.     

Persistent thrombin generation in humans during specific thrombin inhibition with hirudin

BACKGROUND: The degree to which antithrombotic drugs suppress thrombin generation is unknown. Because hirudin, unlike antithrombin III, binds intravascular thrombin rapidly and selectively to yield a circulating inactive complex of 3- to 4-hour half-life, we used intravenous hirudin in humans to investigate the course of thrombin generation during and early after anticoagulation with this potent, direct antithrombin. METHODS AND RESULTS: Intravascular thrombin was measured with an ELISA for the thrombin-hirudin complex formed during and for 18 hours after stopping a 6-hour infusion of hirudin at 0.1, 0.2, and 0.3 mg.kg-1.h-1 in three groups of six patients each. With free hirudin in 20- to 10,000-fold molar excess of thrombin and peak activated partial thromboplastin times of 2.3 to 3.0 times baseline, mean plasma thrombin-hirudin complex increased from 794 +/- 85 pg/mL (mean +/- SEM) 15 minutes after the start of the infusion to 1617 +/- 151 pg/mL at 6 hours of infusion to 2667 +/- 654 pg/mL at 24 hours. During the 24-hour observation period, plasma concentration of fragment 1.2 (the peptide released during conversion of prothrombin to thrombin) never fell below baseline but rather increased transiently during the hirudin infusion. Plasma concentrations of thrombin-antithrombin III complex (in ng/mL) decreased from 4.34 +/- 0.40 at baseline to 1.64 +/- 0.13 at 6 hours (P < .001) and gradually increased after stopping the infusion to 5.7 +/- 0.87 at 24 hours (nonsignificant compared with baseline). CONCLUSIONS: Measurement of thrombin-hirudin complex may be used as a marker of thrombin generation in humans. Persistent accumulation of thrombin-hirudin complex and generation of fragment 1.2 during and after completion of potent anticoagulation with hirudin suggest thrombin generation is not blocked by high-affinity thrombin inhibition. The persistent formation of thrombin during declining plasma levels of hirudin may contribute to the pathogenesis of rethrombosis early after antithrombin therapy or during inadequate anticoagulation.