Anomalous system state identification

1. A real-time method for identifying an anomalous state of a system, the anomalous state being a difference in operation of the system away from a normal mode of operation, the system including a sensor outputting time series data items relating to a property of the system, the method comprising:

receiving a current data item from the sensor;

recursively calculating an estimate of a current data density for the time series data items using the current data item, wherein said data density is a measure of the total distances in a data space between data items of said time series data;

recursively calculating at least one statistical property of the estimate of the current data density, wherein the at least one statistical property comprises the current mean of the estimate of the current data density;

determining from the at least one statistical property whether the current data item indicates an anomalous state of the system, the anomalous state being a difference in operation of the system away from a normal mode of operation; and

outputting a signal if it is determined that the current data item indicates an anomalous state of the system.

2. The method of claim 1 , wherein the at least one statistical property further comprises the current variance of the estimate of the current data density.

3. The method of claim 2 , wherein the current data item is determined to be anomalous based on the degree of difference between the estimate of the current data density and the current mean of the current data density.

4. A real-time method for identifying an anomalous state of a system, the system including a sensor outputting time series data items relating to a property of the system, the method comprising:

receiving a current data item from the sensor;

recursively calculating an estimate of a current data density for the time series data items using the current data item, recursively calculating at least one statistical property of the estimate of the current data density;

determining from the at least one statistical property whether the current data item indicates an anomalous state of the system, wherein the at least one statistical property includes the current mean of the current data density and the current variance of the current data density, and wherein the current data item is determined to be anomalous based on the degree of difference between the estimate of the current data density and the current mean of the current data density and wherein the current data item is determined to be anomalous if the difference between the estimate of the current data density and the current mean of the current data density is greater than three standard deviations of the current mean of the current data density; and

outputting a signal if it is determined that the current data item indicates an anomalous state of the system.

5. The method of claim 1 , further comprising applying a further test to determine whether the current data item is anomalous.

6. The method of claim 5 , wherein the further test is selected from: a temporal based test; and an event based test.

7. The method of claim 1 , and further comprising:

recursively calculating a mean value of the data item using the current data item; and

recursively calculating a scalar product for the data item using the current data item.

8. The method of claim 7 , and further comprising:

using the mean value of the data item and the scalar product of the data item to recursively calculate the estimate of the current data density.

9. The method of claim 1 , wherein the system includes a plurality of sensors each outputting time series data items relating to a different property of the system and wherein the method is applied to current data items respectively received from each of the plurality of sensors.

10. The method of claim 9 , wherein determining comprises determining from the at least one statistical property whether a subset of current data items of the plurality of data items indicate an anomalous state of the system.

11. The method of claim 9 , wherein determining comprises determining from the at least one statistical property whether all current data items of the plurality of data items indicate an anomalous state of the system.

12. The method of claim 1 , wherein the signal is selected from: a control signal; a feedback signal; an alarm signal; a command signal; a warning signal; an alert signal; a servo signal; a trigger signal; a data capture signal; and a data acquisition signal.

13. The method of claim 1 , wherein the system is an electrical or electro-mechanical system.

14. A real-time method for identifying an anomalous state of a system, the system including a sensor outputting time series data items relating to a property of the system, the method comprising:

receiving a current data item from the sensor;

recursively calculating an estimate of a current data density for the time series data items using the current data item, recursively calculating at least one statistical property of the estimate of the current data density;

determining from the at least one statistical property whether the current data item indicates an anomalous state of the system, wherein the at least one statistical property includes the current mean of the current data density and the current variance of the current data density, and wherein the current data item is determined to be anomalous based on the degree of difference between the estimate of the current data density and the current mean of the current data density and wherein the current data item is determined to be anomalous if the difference between the estimate of the current data density and the current mean of the current data density is greater than three standard deviations of the current mean of the current data density; and

outputting a signal if it is determined that the current data item indicates an anomalous state of the system;

wherein the system is an electrical or electro-mechanical system, and wherein the system is a video system, the sensor is an image sensor and the time series data is colour video data or greyscale video data.

15. The method of claim 14 , wherein the property is a sub-region of a frame of video data.

16. A data processing apparatus for identifying an anomalous state of a system in real time, the anomalous state being a difference in operation of the system away from a normal mode of operation, the system including a sensor outputting time series data items relating to a property of the system, the apparatus comprising:

a data processing device; and

a storage device in communication with the data processing device, the storage device storing computer program code executable by the data processing device to carry out steps for identifying the anomalous state of the system, the steps comprising:

receiving a current data item from the sensor;

recursively calculating an estimate of a current data density for the time series data items using the current data item, wherein said data density is a measure of the total distances in a data space between data items of said time series data;

recursively calculating at least one statistical property of the estimate of the current data density, wherein the at least one statistical property comprises the current mean of the estimate of the current data density;

determining from the at least one statistical property whether the current data item indicates an anomalous state of the system, the anomalous state being a difference in operation of the system away from a normal mode of operation; and

outputting a signal if it is determined that the current data item indicates an anomalous state of the system.

17. A system, the system comprising:

at least one operative part;

at least one sensor for measuring a property of the operative part, the sensor outputting time series data items relating to a property of the system; and

a data processing apparatus in communication with the sensor to receive time series data from the sensor, the data processing apparatus comprising:

a data processing device; and

a storage device in communication with the data processing device, the storage device storing computer program code executable by the data processing device to carry out steps for identifying an anomalous state of a system being a difference in operation of the system away from a normal mode of operation, the steps comprising:

receiving a current data item from the sensor;

recursively calculating an estimate of a current data density for the time series data items using the current data item, wherein said data density is a measure of the total distances in a data space between data items of said time series data;

recursively calculating at least one statistical property of the estimate of the current data density, wherein the at least one statistical property comprises the current mean of the estimate of the current data density;

determining from the at least one statistical property whether the current data item indicates an anomalous state of the system, the anomalous state being a difference in operation of the system away from a normal mode of operation of the system; and

outputting a signal if it is determined that the current data item indicates an anomalous state of the system.

18. A system as claimed in claim 17 , wherein the data processing apparatus has an output which is in communication with the system to output the signal to the system.

19. A computer readable medium storing in non-transitory form computer program code executable by a data processing device to carry out real-time steps for identifying an anomalous state of a system, the anomalous state being a difference in operation of the system away from a normal mode of operation of the system, the system including a sensor outputting time series data items relating to a property of the system, the real-time steps comprising:

receiving a current data item from the sensor;

recursively calculating an estimate of a current data density for the time series data items using the current data item, wherein said data density is a measure of the total distances in a data space between data items of said time series data;

recursively calculating at least one statistical property of the estimate of the current data density, wherein the at least one statistical property comprises the current mean of the estimate of the current data density;

determining from the at least one statistical property whether the current data item indicates an anomalous state of the system, the anomalous state being a difference in operation of the system away from a normal mode of operation; and

outputting a signal if it is determined that the current data item indicates an anomalous state of the system.