Traitement en cours

Veuillez attendre...

Paramétrages

Paramétrages

1. US20100030711 - Method and apparatus for evolving overlays to operate an extended analog computer as a classifier or a controller

Note: Texte fondé sur des processus automatiques de reconnaissance optique de caractères. Seule la version PDF a une valeur juridique

Claims

1. A method for configuring an extended analog microcontroller for a control application comprising:
selecting input pins from among a plurality of pins in a continuous sheet processor;
selecting an arrangement of intermediate and output pins from among the remaining pins in the plurality of pins in the continuous sheet processor;
applying a pattern data set to the input pins;
using an evolutionary algorithm, couple current sources and sinks to the intermediate and output pins;
measuring an error between an output and its expected value; and
continuing to select intermediate and output pin arrangements, apply pattern data sets, and measure errors until a configuration threshold is met.
2. The method of claim 1, the coupling of current sinks and sources to the intermediate and output pins is implemented using a particle swarm optimization process.
3. The method of claim 1, the measured error is one of percentage error and a sum squared error.
4. The method of claim 1 further comprising:
coupling fuzzy logic and Lukasiewicz logic functions to the intermediate and output pins with the use of the evolutionary algorithm.
5. The method of claim 1, the coupling of the current sources and sinks further comprising:
generating a group of intermediate current source locations and values; and
treating the group of intermediate current source locations and values as a group of particles for a particle swarm algorithm.
6. The method of claim 5, the application of the pattern data set further comprising:
applying a pattern data set from a plurality of elevator brake fuzzy logic controller training pattern data sets;
reading a response from an output pin; and
measuring a difference between the response and an expected response for the training pattern data set applied to the intermediate pins.
7. The method of claim 6 further comprising:
applying the remaining pattern data sets in the plurality of elevator brake fuzzy logic controller training pattern data sets;
reading a response from the output pin for each application of a pattern data set;
measuring a difference between each response and its expected response; and
adding the differences to generate a fitness value for the group of intermediate current source locations and values to which the plurality of elevator brake fuzzy logic controller training pattern data sets were applied.
8. The method of claim 7 further comprising:
adjusting the group of intermediate current source locations and values in accordance with a particle swarm formula; and
reapplying the plurality of elevator brake fuzzy logic controller training pattern data sets, reading the responses, measuring the differences, and generating a fitness value for the adjusted group of intermediate current source locations and values.
9. The method of claim 8 further comprising:
comparing a fitness value for a group of intermediate current source locations and values to the fitness values for the other groups of intermediate current source locations and values that were generated; and
selecting the group of intermediate current source locations and values having a best fitness value determined from the comparisons of fitness values.
10. A method for configuring an extended analog microcontroller for a classifier application comprising:
selecting input pins from among a plurality of pins in a continuous sheet processor;
selecting an arrangement of intermediate and output pins from among the remaining pins in the plurality of pins in the continuous sheet processor;
applying a pattern data set having pattern data inputs and a corresponding class for the pattern data inputs to the input pins;
using an evolutionary algorithm, couple current sources and sinks to the intermediate and output pins;
measuring an error between an output and its expected class; and
continuing to select intermediate and output pin arrangements, apply pattern data sets, and measure errors until a configuration threshold is met.
11. The method of claim 10, the coupling of current sinks and sources to the intermediate and output pins is implemented using a particle swarm optimization process.
12. The method of claim 10, the measured error is a percentage error or a sum squared error.
13. The method of claim 10 further comprising:
coupling fuzzy logic and Lukasiewicz logic functions to the intermediate and output pins with the use of the evolutionary algorithm.
14. The method of claim 10, the coupling of the current sources and sinks further comprising:
generating a group of intermediate current source locations and values; and
treating the group of intermediate current source locations and values as a group of particles for a particle swarm algorithm.
15. The method of claim 14, the application of the pattern data set further comprising:
applying a pattern data set from a plurality of classifier training pattern data sets;
reading a classification response from an output pin; and
measuring a difference between the classification response and an expected classification response for the training pattern data set applied to the intermediate pins.
16. The method of claim 15 further comprising:
applying the remaining pattern data sets in the plurality of classifier training pattern data sets;
reading a classification response from the output pin for each application of a pattern data set;
measuring a difference between each classification response and its expected classification response; and
adding the differences to generate a fitness value for the group of intermediate current source locations and values to which the plurality of classifier training pattern data sets were applied.
17. The method of claim 16 further comprising:
adjusting the group of intermediate current source locations and values in accordance with a particle swarm formula; and
reapplying the plurality of classifier training pattern data sets, reading the responses, measuring the differences, and generating a fitness value for the adjusted group of intermediate current source locations and values.
18. The method of claim 17 further comprising:
comparing a fitness value for a group of intermediate current source locations and values to the fitness values for the other groups of intermediate current source locations and values that were generated; and
selecting the group of intermediate current source locations and values having a best fitness value determined from the comparisons of fitness values.