Control of a machine using relaxed abduction method

Patent metadata

Field	Value
Publication number	US-9897983-B2
Application number	US-201114353039-A
Country	US
Kind code	B2
Filing date	Oct 28, 2011
Priority date	Oct 28, 2011
Publication date	Feb 20, 2018
Grant date	Feb 20, 2018

Abstract

A method and apparatus for actuating a machine is provided in which a relaxed abduction problem is determined in order to explain the greatest possible part of the observations with the fewest possible assumptions. Based upon two preference orders over a subset of observations and a subset of assumptions, tuples are determined so that the theory together with the subset of assumptions explains the subset of observations. On the basis of the formal validity of the approach certain characteristics of the set of results (such as correctness, completeness, etc.) are checked. By the choice of underlying representational language and the preference relations, the complexity of the problem-solving process is influenced and thus flexibly adapted with regard to domain requirements. The invention may be used for any machines, e.g. gas turbines or steam turbines.

First claim

The invention claimed is:

1. A method for actuating a machine, comprising:
determining, by a processing unit, a solution using relaxed abduction for a defined problem related to incorrect information or incomplete models associated with diagnosis of the machine using automatically joint optimization of sets of explained observations and required assumptions to determine a relaxed abduction problem,
solving, by a processing unit, the relaxed abduction problem so that the machine is actuated,
wherein the relaxed abduction problem is determined by description logic, and
actuating, by a processing unit, the machine for the diagnosis,
wherein two orders of preference over a subset of the observations and a subset of the assumptions are taken as a basis for determining tuples, so that the theory together with the subset of the assumptions explains the subset of the observations;
wherein the relaxed abduction problem is solved by transforming the relaxed abduction problem into a hypergraph, so that tuples (A,O) are encoded by pareto-optimal paths in the hypergraph;
wherein hyperedges of the hypergraph are induced by transcriptions of prescribed rules; and
wherein a weighted hypergraph H RAP =(V,E), which is induced by the relaxed abduction problem, is determined by

V
=

{

(

A
⊑
B

)

,

(

A
⊑

∃

r
·
B

)

|
A

,

B
∈

N
C
T

,

r
∈

N
R

}

,

⁢
wherein

V
T

=

{

(

A
⊑
A

)

,

(

A
⊑
T

)

|

A
∈

N
C
T

}

⊆
V

denotes a set of final states and E denotes a set of the hyperedges

e
=

(

T
⁡

(
e
)

,

h
⁡

(
e
)

,

w
⁡

(
e
)

)

,

so that the following holds: there is an axiom aεT∪A that justifies the derivation h(e)εV from T(e) ⊂ V on the basis of one of the prescribed rules, wherein the edge weight w(e) is determined according to

A
=

{

{
a
}

if
⁢

⁢
a

⁢

∈

⁢
A

,

⁢

∅

otherwise

⁢

⁢
O

=

{

{

h
⁡

(
e
)

}

Assignees

• Grimm Stephan
• Hubauer Thomas
• Roshchin Mikhail
• Stender Holger
• Watson Stuart
• Siemens Ag

Inventors

• Grimm Stephan
• Hubauer Thomas
• Roshchin Mikhail
• Stender Holger
• Watson Stuart

Classifications

• G05B13/04 Primary

  involving the use of models or simulators · CPC title

• G05B13/0265 Primary

  the criterion being a learning criterion · CPC title

• G05B23/0278

  Qualitative, e.g. if-then rules; Fuzzy logic; Lookup tables; Symptomatic search; FMEA · CPC title