Use of sequential nearest neighbor clustering for instance selection in machine condition monitoring
View Patent ↗A method is provided for selecting a representative set of training data for training a statistical model in a machine condition monitoring system. The method reduces the time required to choose representative samples from a large data set by using a nearest-neighbor sequential clustering technique in combination with a kd-tree. A distance threshold is used to limit the geometric size the clusters. Each node of the kd-tree is assigned a representative sample from the training data, and similar samples are subsequently discarded.
1. A computer implemented sequential nearest neighbor clustering method for selecting a set of training data from a set S of samples from a machine condition monitoring system that includes an array of sensors, the selected set of training data being for use in training a statistical model for evaluating measurements in the machine condition monitoring system, the method comprising the steps of:
performing the following steps for each sample p from the set S:
calculating a distance from the sample p to a node of a kd-tree, wherein each node of the kd-tree is empty or represents another sample removed from the set S;
if the calculated distance is greater than a distance threshold r max , wherein r max indicates the geometric size of a cluster, and the node of the kd-tree has children, calculating a distance from the sample p to a child of the node selected according to a classification of p in the kd-tree;
repeating the previous step until either the calculated distance is below the distance threshold r max , or the node of the kd-tree has no children; and
if no calculated distance is below the distance threshold r max , then populating a next leaf in the kd-tree with p; and using the nodes of the kd-tree as the set of training data.
2. The method of claim 1 , further comprising the step of:
randomizing an order of the samples p in the training set S.
3. The method of claim 1 , wherein the distance r max is determined as
r max =r 0 √{square root over ( d )}
where r 0 is a predetermined constant and d is a number of sensors represented by the sample s.
4. The method of claim 3 , wherein r 0 is set to 1/33.
5. The method of claim 1 , the step of performing the steps for each sample p from the set S further comprises: backtracking in the kd-tree if a nearest cluster center for a sample p is not located in a hypercube of the kd-tree specified by the cluster center.
6. A computer-usable medium having computer readable instructions stored thereon for execution by a processor to perform a sequential nearest neighbor clustering method for selecting a set of training data from a set S of samples from a machine condition monitoring system that includes an array of sensors, the selected set of training data being for use in training a statistical model for evaluating measurements in the machine condition monitoring system, the method comprising the steps of:
performing the following steps for each sample p from the set S:
calculating a distance from the sample p to a node of a kd-tree, wherein each node of the kd-tree is empty or represents another sample removed from the set S;
if the calculated distance is greater than a distance threshold r max , wherein r max indicates the geometric size of a cluster, and the node of the kd-tree has children, calculating a distance from the sample p to a child of the node selected according to a classification of p in the kd-tree;
repeating the previous step until either the calculated distance is below the distance threshold r max , or the node of the kd-tree has no children; and
if no calculated distance is below the distance threshold r max , then populating a next leaf in the kd-tree with p; and using the nodes of the kd-tree as the set of training data.
7. The computer useable medium of claim 6 , further comprising the step of:
randomizing an order of the samples p in the training set S.
8. The computer useable medium of claim 6 , wherein the distance r max is determined as
r max =r 0 √{square root over ( d )}
where r 0 is a predetermined constant and d is a number of sensors represented by the sample s.
9. The computer useable medium of claim 8 , wherein r 0 is set to 1/33.
10. The computer useable medium of claim 6 , the step of performing the following steps for each sample p from the set S further comprises: backtracking in the kd-tree if a nearest cluster center for a sample p is not located in hypercube of the kd-tree specified by the cluster center.