Method and apparatus for generation of multi-branched diagnostic trees
First Claim
1. A method for automated generation of a multi-branched diagnostic tree for isolation of faults in a system comprising the steps of storing a model of the system to be diagnosed in a memory, said model including terminals indicating measurement points in the system to be diagnosed, inputting into the system measurements of taken at the measurement points in the system,propagating the variables through the model for automated generation of a diagnostic tree;
- determining an operating value for each of said terminals;
selecting a granularity for each node of said diagnostic tree;
establishing at least one terminal as a root node and establishing the root node as a selected node;
generating a plurality of branches extending from said selected node, the number of branches extending from said selected node corresponding to each selected granularity, and selecting a branch value to correspond to each of said generated branches;
selecting one of said branches of said selected node not terminating in a leaf node for analysis;
propagating the branch value of said selected branch through said model thereby generating a propagated value for said terminals;
identifying conflicts between propagated values at each terminal and generating conflict candidate sets;
terminating the selected branch with a leaf node denoting no further fault isolation is possible if no conflict set exists;
terminating the selected branch with a node denoting multiple faults if multiple, nonintersecting conflict sets exits;
terminating the selected branch with a leaf node denoting a single fault if one conflict set having a single component exists;
attempting to identify a next measurement to be made if at least one conflict set exists,terminating the selected branch with a leaf node denoting no further fault isolation is possible if no next measurement can be identified,terminating the selected branch with a measurement node corresponding to an identified next measurement and establishing said measurement node as the selected node and returning to the step of generating a plurality of branches if a next measurement can be identified;
finding a branch not terminated with a node and establishing the found branch as the selected branch and returning to the step of propagating the branch value of said selected branch;
finding a measurement node without established branches, establishing said measurement node as the selected node, and returning to the step of generating a plurality of branches if a measurement node is established; and
stopping tree generation if all branches are terminated with leaf nodes.
3 Assignments
0 Petitions
Accused Products
Abstract
A method and apparatus for automated generation of multi-branched diagnostic trees is described. The present invention generally is practiced on model-based diagnostic systems and utilizes a candidate generator, constraint propagator and best measurement generator, along with a model of a system to be diagnosed, to generate nodes and branches of a diagnostic tree.
-
Citations
36 Claims
-
1. A method for automated generation of a multi-branched diagnostic tree for isolation of faults in a system comprising the steps of storing a model of the system to be diagnosed in a memory, said model including terminals indicating measurement points in the system to be diagnosed, inputting into the system measurements of taken at the measurement points in the system,
propagating the variables through the model for automated generation of a diagnostic tree; -
determining an operating value for each of said terminals; selecting a granularity for each node of said diagnostic tree; establishing at least one terminal as a root node and establishing the root node as a selected node; generating a plurality of branches extending from said selected node, the number of branches extending from said selected node corresponding to each selected granularity, and selecting a branch value to correspond to each of said generated branches; selecting one of said branches of said selected node not terminating in a leaf node for analysis; propagating the branch value of said selected branch through said model thereby generating a propagated value for said terminals; identifying conflicts between propagated values at each terminal and generating conflict candidate sets; terminating the selected branch with a leaf node denoting no further fault isolation is possible if no conflict set exists; terminating the selected branch with a node denoting multiple faults if multiple, nonintersecting conflict sets exits; terminating the selected branch with a leaf node denoting a single fault if one conflict set having a single component exists; attempting to identify a next measurement to be made if at least one conflict set exists, terminating the selected branch with a leaf node denoting no further fault isolation is possible if no next measurement can be identified, terminating the selected branch with a measurement node corresponding to an identified next measurement and establishing said measurement node as the selected node and returning to the step of generating a plurality of branches if a next measurement can be identified; finding a branch not terminated with a node and establishing the found branch as the selected branch and returning to the step of propagating the branch value of said selected branch; finding a measurement node without established branches, establishing said measurement node as the selected node, and returning to the step of generating a plurality of branches if a measurement node is established; and stopping tree generation if all branches are terminated with leaf nodes. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13)
-
-
14. An automated diagnostic computer system, comprising:
-
input means for entering and editing information in the system and generating a model of a system to be diagnosed, said model including terminals indicating measuring points in the system to be diagnosed, a memory in the system for storing said model; means for making predictions about terminal values using said model and assigning predicted values to terminals of said model; means for generating conflicts between said assigned values and actual measured values and between different sets of assigned values; means for identifying a next measurement to be made for isolating a fault; means for determining an operating value for each of said terminals; means for selecting a granularity for each node of said diagnostic tree; means for establishing a first terminal as a root node and establishing the root node as a selected node; means for generating a plurality of branches extending from said selected node, the number of branches extending from said selected node corresponding to said selected granularity, and selecting a branch value to correspond to each of said generated branches; means for selecting one of said branches of said selected node not terminating in a leaf node for analysis; means for propagating the branch value of said selected branch through said model thereby generating a propagated value for said terminals; means for identifying conflicts between propagated values at each terminal and generating conflict candidate sets; means for terminating the selected branch with a leaf node denoting no further fault isolation is possible if no conflict set exists; means for terminating the selected branch with a node denoting multiple faults if multiple, nonintersecting conflict sets exists; means for terminating the selected branch with a leaf node denoting a single fault if one conflict set having a single component exists; means for attempting to identify a next measurement to be made if at least one intersecting conflict set exists, terminating the selected branch with a leaf node denoting no further fault isolation is possible if no next measurement can be identified, terminating the selected branch with a measurement node corresponding to an identified next measurement and establishing said measurement node as the selected node and if a next measurement can be identified generating a plurality of branches extending from said selected node, the number of branches extending from said selected node corresponding to said selected granularity, and selecting a branch value to correspond to each of said generated branches; means for finding a branch not terminated with a node and establishing the found branch as the selected branch and propagating the branch value of said selected branch through said model thereby generating a propagated value for said terminals; and means for finding a measurement node without established branches, establishing said measurement node as the selected node, and if a measurement node is established generating a plurality of branches extending from said selected node, the number of branches extending from said selected node corresponding to said selected granularity, and selecting a branch value to correspond to each of said generated branches; and means for stopping tree generation if all branches are terminated with leaf nodes. - View Dependent Claims (15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29)
-
-
30. A method for automatically assigning a granularity to an internal node of a diagnostic tree of a model-based diagnostic reasoning system, said diagnostic tree including root nodes, integral nodes and branches, said method comprising the steps of:
-
a) establishing at least one terminal of the model as a root node and establishing the root node as a selected node; b) determining a value for said one terminal; c) selecting a granularity for said root node; d) generating a plurality of branches extending from said root node, the number of branches extending from said root node corresponding to the selected granularity and selecting a branch value to correspond to each said generated branches; e) selecting one of said branches of said selected node not terminating in a node for analysis; f) propagating the branch value of said selected branch; g) attempting to identify a next measurement to be made at a terminal and if no next measurement can be identified then terminating said selected branch in a leaf node; h) terminating the selected branch with a node identifying the terminal to be measured if a next measurement is identified; i) determining a number (N) of numeric ranges using the intervals propagated to the terminal to be measured; j) generating N branches from said node identifying the terminal to be measured; k) assigning each said branch one of said interval ranges; l) establishing a node having at least one branch not terminating in a node as the selected node and returning to step (e); and m) stopping tree generation when all paths of the tree terminate in leaf nodes. - View Dependent Claims (31, 32, 33, 34, 35, 36)
-
Specification