Method and a control system for monitoring the condition of an industrial robot

1. An industrial robot system condition analysis method, comprising:

monitoring at least two selected input signals, each selected input signal indicating a condition related to a property of the industrial robot system,

using a processor for analyzing any combination of said selected input signals utilizing a multivariate data analysis to identify combinations of input signals and/or relationships between the input signals, wherein said multivariate data analysis of a selection of said signals comprises

formatting the data of n selected signals as data points in an n-dimensional space, classification of state of each data point, wherein said state indicate one of: normal operation or failure, and

determining a direction of each of said data points indicating a direction in said n-dimensional space in which the data point tend to move, and

detecting condition changes in the robot system based on a result of said analysis.

2. The method according to claim 1 , further comprising:

identifying a root cause of said condition changes in the robot system.

3. The method according to claim 1 , further comprising:

predicting a failure.

4. The method according to claim 1 , wherein said analysis of said selected input signals is performed by utilizing a signal modeling of the input signals.

5. The method according to claim 1 , wherein an input signal used in the condition analysis comprises a signal carrying information about the condition of the robot system, said signal being any one from the group of:

a signal comprising more than one sample per measurement, y(t k ,i, m), where k is the sample number in the measurement, i is the signal number and m is the measurement number;

a signal comprising a continuous single value per measurement, y(i, m,), where i is the signal number and m is the measurement number;

a signal comprising a binary single value per measurement, y d (i, m) where i is the signal number and m is the measurement number.

6. The method according to claim 1 , wherein said Multivariate Data Analysis of said signals further comprises:

classification of the state of a data point indicates at least one of: detection of a failure of the robot system and isolation of a component being the root cause of a failure in the robot system; and

said determination of the direction of said data point is used for the prediction of the lifetime of a component causing a failure in the robot system.

7. A computer readable medium embodying a computer program product, the computer program product comprising:

computer program instructions recorded on the computer readable medium and executable by a processor for performing a method comprising monitoring at least two selected input signals, each selected input signal indicating a condition related to a property of the industrial robot system, analyzing any combination of said selected input signals utilizing a multivariate data analysis to identify combinations of input signals and/or relationships between the input signals, wherein said multivariate data analysis of a selection of said signals comprises formatting the data of n selected signals as data points in an n-dimensional space, classification of state of each data point, wherein said state indicate one of: normal operation or failure, and determining a direction of each of said data points indicating a direction in said n-dimensional space in which the data point tend to move, and detecting condition changes in the robot system based on a result of said analysis.

8. The computer program product according to claim 7 , wherein said computer program instructions further comprise algorithms for performing signal modeling of said at least two input signals.

9. The computer program product according to claim 7 , wherein the computer program product is performed as software and/or hardware.

10. An industrial robot system condition analysis method, comprising:

monitoring at least two selected input signals, each selected input signal indicating a condition related to a property of the industrial robot system,

using a processor for analyzing any combination of said selected input signals utilizing a multi layer principal component analysis approach, wherein said multi layer principal component analysis approach comprises

selecting sensor signals of the robot system,

pre-processing said sensor signals by means of calculation of high frequency RMS, peak-to-peak, crest factor, kurtosis and peak-to-mean values of the signals,

performing a principal component analysis model of a normal behavior of the robot system based on said pre-processing of the selected signals, and

determining a severity factor Q, representing a deviation from said normal behavior of the robot system, and

detecting condition changes in the robot system based on the result of said analysis.

11. The method according to claim 10 , wherein said Multi Layer Principal Component Analysis data analysis approach further comprises:

said severity factor Q is a base for: detection of a failure of the robot system, isolation of a component being the root cause of a failure in the robot system, prediction of the remaining lifetime of a component causing a failure in the robot system, prediction of the lifetime for the whole robot system.

12. A computer readable medium embodying a computer program product, the computer program product comprising:

computer program instructions recorded on the computer readable medium and executable by a processor for performing a method comprising monitoring at least two selected input signals, each selected input signal indicating a condition related to a property of the industrial robot system, analyzing any combination of said selected input signals utilizing a multi layer principal component analysis approach, wherein said multi layer principal component analysis approach comprises selecting sensor signals of the robot system, pre-processing said sensor signals by means of calculation of high frequency RMS, peak-to-peak, crest factor, kurtosis and peak-to-mean values of the signals, performing a principal component analysis model of a normal behavior of the robot system based on said pre-processing of the selected signals, and determining a severity factor Q, representing a deviation from said normal behavior of the robot system, and detecting condition changes in the robot system based on the result of said analysis.