Management zones for precision farming

• US 6,058,351 A
• Filed: 09/10/1998
• Issued: 05/02/2000
• Est. Priority Date: 09/10/1998
• Status: Expired due to Term

• Current Assignee: Blue Leaf IP Incorporated (CNH Industrial N.V.), CNH Industrial America LLC (CNH Industrial N.V.)
• Original Assignee: Case Corporation (CNH Industrial N.V.)
• Inventors: McCauley, James D.
• Primary Examiner(s): McElheny, Jr., Donald E.
• Application Number:
  US09/151,125
• Time in Patent Office: 600 Days
• Field of Search: 702/2, 702/5, 706/925, 706/930, 706/931, 706/929
• US Class Current: 702/5
• CPC Class Codes: A01B 79/005 Precision agriculture
• CPC Class Codes: G06N 3/02 Neural networks
• CPC Class Codes: Y02A 40/10 in agriculture

Abstract

A method and apparatus for determining management zones in a field for precision farming. The method and apparatus use a self-organizing network, such as a Kohonen neural network, to perform the classification of site-specific farming data into management zones in a farm field. The self-organizing network is configured to learn the proper classifications through a learning or training process.

Claims

1. A method of determining management zones in a field for precision farming, the method comprising the steps of:

obtaining site-specific farming data;

providing the site-specific farming data to a self-organizing network, the self-organizing network configured to classify locations in the field into management zones based upon the site-specific farming data;

obtaining classification outputs from the self-organizing network; and

correlating the classification outputs of the self-organizing network with positions in the field.

2. The method of claim 1, further comprising the step of displaying a map of the field showing the management zones throughout the field.

3. The method of claim 1, further comprising the step of receiving field location data from a georeferenced digital map of the field.

4. The method of claim 1, further comprising the step of training the self-organizing network to make adjustments to the self-organizing network so that the classification characteristics of the self-organizing network for classifying the site-specific farming data improve.

5. The method of claim 4, wherein the step of obtaining classification outputs includes post processing data output from the self-organizing network.

6. The method of claim 4, further comprising the step of choosing a training rate value.

7. The method of claim 6, wherein the training rate value is variable during training.

8. The method of claim 4, wherein the site-specific farming data includes a plurality of data layers and the method further comprises the step of determining which layers of site-specific farming data to provide to the self-organizing network to generate the management zones.

9. The method of claim 8, further comprising the step of dividing the field into a multiplicity of discrete areas.

10. The method of claim 4, further comprising the step of preprocessing the site-specific farming data.

11. The method of claim 10, wherein the preprocessing step includes the step of normalizing the site-specific farming data being provided to the self-organizing network.

12. The method of claim 1, wherein the self-organizing network is selected from a group consisting of a Kohonen Network, an Adaptive Resonance Theory (ART) network, a radial basis function network, an Adaptive Kohonen Network (AK), and a hybrid comprised of a Kohonen Network and a network selected from a group consisting of an ART and an AK.

13. The method of claim 1, wherein the self-organizing network is a Kohonen neural network.

14. The method of claim 13, further comprising the step of choosing a distance metric to be used by the Kohonen neural network.

15. The method of claim 14, wherein the distance metric used is a Euclidean distance metric.

16. The method of claim 14, further comprising the step of displaying the management zones on a cathode ray tube map display.

17. An apparatus for determining at least two management zones in a field, based upon field characteristics at a multiplicity of field positions, the apparatus comprising:

an information input device for receiving site-specific field characteristic data;

a self-organizing network coupled to the information input device, wherein the self-organizing network classifies each of the field positions into one of the at least two management zones; and

an information output device coupled to the self-organizing network, wherein the information output device communicates the management zone classifications.

18. The apparatus of claim 17, further comprising a postprocessor coupled to the information output device, wherein the post processor manipulates the management zone classifications.

19. The apparatus of claim 18, wherein the postprocessor correlates the management zone classifications with the field positions.

20. The apparatus of claim 17, further comprising a preprocessor coupled to the information input device, wherein the preprocessor manipulates the field characteristic data for utilization by the self-organizing network.

21. The apparatus of claim 20, wherein the preprocessor normalizes the field characteristic data.

22. The apparatus of claim 17, further comprising an information database coupled to the information input device, wherein the information database stores the field characteristic data.

23. The apparatus of claim 17, wherein the self-organizing network is a Kohonen neural network.

24. The apparatus of claim 17, wherein the self-organizing network is selected from a group consisting of a Kohonen Network, an Adaptive Resonance Theory (ART) network, a radial basis function network, an Adaptive Kohonen Network (AK), and a hybrid comprised of a Kohonen Network and a network selected from a group consisting of an ART and an AK.

25. The apparatus of claim 17, further comprising an information database coupled to the information output device, wherein the information database stores the management zone classifications.

26. The apparatus of claim 25, further comprising a management zone display, wherein the management zone display shows the management zone classifications in a user-readable format.

27. The apparatus of claim 17, wherein the self-organizing network has a multiplicity of weighted interconnections and has a multiplicity of nodes, each node having a transfer function designed to provide desired management zone data.

28. A computing apparatus for determining at least two management zones in a field, based upon field characteristics at a multiplicity of field positions, the apparatus comprising:

an input device;

a microprocessor coupled to the input device;

an output device coupled to the microprocessor; and

a memory device coupled to the microprocessor, wherein the memory device stores a Kohonen neural network program, the program being read and executed by the microprocessor to implement a Kohonen neural network, the Kohonen neural network receiving field characteristic data from the input device, and communicating management zone data to the output device.

29. The computing apparatus of claim 28, wherein the input device receives the field characteristic data, the field characteristic data being communicated to the microprocessor.

30. The computing apparatus of claim 28, wherein the output device receives management zone data from the microprocessor and communicates the management zone data to a user.

31. The computing apparatus of claim 28, wherein the memory device also stores a preprocessing algorithm that conditions the field characteristic data to be communicated to the Kohonen neural network.

32. The computing apparatus of claim 28, wherein the memory device also stores a post processing algorithm that conditions the management zone data to be communicated to a user.

33. The computing apparatus of claim 28, wherein the output device is a visual display, the visual display showing a map of the field and the management zones within the field.

34. The computing apparatus of claim 28, wherein the memory device stores a training program, the training program being configured to make adjustments to the Kohonen neural network program, so that the Kohonen neural network is trained to properly classify field characteristic data into management zone data.