This paper proposes a novel anomaly detection method for spacecraft systems based on data-mining techniques. This method automatically constructs a system behavior model in the form of a set of rules by applying pattern clustering and association rule mining to the time-series data obtained in the learning phase, then detects anomalies by checking the subsequent on-line data with the acquired rules. A major advantage of this approach is that it requires little a priori knowledge on the system.

View full abstract

The zooplankton culture system is used to raise live food for fish and is essential for the fish culture in space. A trial model of this system was first manufactured in 1998. A problem was found in the filter which has the role of separating nitrogen waste from the rearing water, so it was not suitable for long-term culture. Subsequently, a new model , the airtight recirculating zooplankton culture device, was designed and manufactured based on experience gained from the trial model. A culture of Moina macrocopa was successfully maintained for ten days using this device, In this experiment M. macrocopa was cultured at high densities, up to 12.0 ind./ml. However, improvements to the waste solution separation device were insufficient and there were still obvious problem. Additionally, there were problems with preventing fluctuations of pH. Further improvements are warranted for this device.

View full abstract

When an upward magnetization force F_m is imposed on a liquid, the force acting on the liquid is the difference between gravitational and magnetization forces, Thus, the magnitude of vertical acceleration acting on the liquid can be varied from 1g to 0g by changing the upward F_m. To verify this vertical acceleration-control method. behavior of bubbles by injecting gas from a nozzle was examined. By decreasing the magnitude of vertical acceleration, the number of bubbles formed per unit time decreased and their size increased. These results confirm that vertical acceleration can be changed from normal gravity to near zero-gravity. The method to generate low-gravity conditions will be useful to study the effects of gravitational fields.

View full abstract

GPS attitude determination system (ADS) uses carrier phase measurements observed at multiple antennas, and it contributes especially to simplification of the GN&C sub-system. The major problem of GPS attitude determination is how to improve reliability and accuracy. In particular, to establish a reliable attitude determination algorithm, which resolves integer ambiguity of carrier wave length, is the most critical subject. A usual integer search resolution is vulnerable to measurement error source with only one epoch data of a single carrier frequency, and also requires a high performance on-board computer. This paper presents the preliminary design result of a reliable and accurate algorithm determination. In the algorithm, the cost function to be evaluated for initial ambiguity search is accumulated double-difference (DD) measurement residuals in order to mitigate the effects of measurement errors. And it has a function of bias error estimation in single-difference(SD) equations to improve attitude accuracy. Besides, attitude propagation using Gyro rate data is adopted in order to reduce search space, i.e., computational load. This GPS attitude determination algorithm was verified by aircraft flight tests, which were conducted in November 2001. The accuracy was evaluated using IMU(Inertial Measurement Unit) data. Ambiguity resolution and attitude accuracy better than 0.5deg were obtained even during dynamic attitude maneuvers in a reasonable computation time.

View full abstract