SAAPS, Version 1.01

Anomaly Prediction Models Help


Based on the anomaly data from a few selected satellites neural network models have been developed to predict the anomalies using a daily geomagnetic disturbance index (sum of Kp) as input.

The models predict whether there will be anomalies or not during a one day interval. With each prediction the probability that the prediction will be correct is also computed. The anomaly prediction plot shows the probability for anomalies over a 14 day interval. E.g., having a 10% probability for anomalies for a certain day means that it is likely that the specific satellite will not experience any anomalies during that day. On the other hand, a 90% probability means that it is likely that the satellite will have anomalies during the day. A probability around 50% means that the model can not make a reliable prediction.

User interface

In the Model List the available models are shown. At startup the first model in the list is by default selected. Other models can be selecting it by clicking on it. Pushing the "Inspect model" button will give a brief description of the model.

In the Year, Month, and Day fields a date for which the prediction should be made can be entered. The fields are initiated with the current date. The date is the last date in the 14 day interval that is plotted.

The "Today" button will set the date fields to the current date.

When the "Plot" button is pushed the model will be run for a 14 day period ending on the slected date and the plot will be displayed.

About the models

The models consist of standard feed-forward neural networks with one input layer, one hidden layer, and one output layer. The input is the daily sum Kp value, which is the sum of eight 3-hour Kp values. As the time history of Kp is important the input consist of several sum Kp values delayed in time.

The output is the prediction whether there will be an anomaly or not. Both one day forecast and nowcast models have been developed. Generally to one day forecast only work for internal charging type of anomalies.

The network has been trained to predict the anomalies for a specific satellite. During the optimization procedure the length of the time history of sum Kp and the number of hidden neurons have been varied.

When the optimal network has been found the prediction accuracy can be examined. The prediction of an anomaly is a binary yes/no problem. However, the network output is a real value which is related to how accurate the prediction is. By examining the network output on the satellite anomaly set probabilities can be assigned to the network output. The figure below illustrates this for one model. The solid lines are the probabilities for correct predictions while the dashed line shows the fraction of events in each bin. The bins are marked with the vertical dotted lines. There are five bins and we see that the events are approximately evenly distributed with 20% in each bin. We see that there is a clear correlation between the network output (|y| in the figure) and the probability. Based on the output value from the network we can thus give the probability that the prediction is correct in the plots.

Reference

Wintoft, P., Predicting charging induced anomalies, Technical Note 3, Development of AI methods in spacecraft anomaly predictions, ESA/ESTEC Contract No. 11974/96/NL/JG(SC), 2001.


Peter Wintoft
Tuesday, October 23, 2001