GB2593197A - System and method for classifying vessels - Google Patents

Abstract

Disclosed is a method of classifying a maritime vessel by detecting S100 data from the vessel and determining S102 whether the detected data comprises identification information. If it is determined that the detected data does not comprise identification information, a classification for the maritime vessel is automatically determined S112 using historical pattern of life data S108. A confidence level associated with the estimated classification is also generated. The classification may comprise the vessel identity, type, predicted track or next port of call. Bearing information may be detected using an acoustic array. Also disclosed is determining the activity of a maritime vessel by identifying the maritime vessel from detected data and obtaining a state diagram of permitted activities associated with the vessel. The activity of the vessel is determined, and verified using the obtained state diagram.

Description

SYSTEM AND METHOD FOR CLASSIFYING VESSELS BACKGROUND

The present invention relates generally to a method and system for classifying vessels, particularly maritime surface vessels which have been detected.

Maritime vessels often transmit a periodic signal in the form of an automatic identification system (AIS) signature or a maritime mobile service identity (MMSI). A problem with these signals is that they can be interrupted or simply not detected. Accordingly, the detected vessel cannot be identified.

It is also noted that traditional maritime analysis is track based. However, the present applicant has also recognised the need for an additional or complementary framework to analyse maritime activity.

It is an example aim of example embodiments of the present invention to at least partially overcome or avoid one or more problems of the prior art, whether identified herein or elsewhere, or to at least provide an alternative to existing systems and related methodologies.

SUMMARY OF INVENTION

According to a first aspect of the present invention, there is provided a method for classifying a maritime vessel, the method comprising: detecting data from a maritime vessel which is to be classified; determining whether the detected data comprises identification information for the maritime vessel; in response to a determination that the detected data does not comprises identification information, automatically estimating a classification for the maritime vessel using historical pattern of life data and generating a confidence level associated with the estimated classification.

According to a second aspect of the present invention, there is provided a system for classifying maritime vessels, the system comprising: at least one detector for detecting data from a maritime vessel which is to be classified, and a processor which is configured to: detect data from the maritime vessel; determine whether the detected data comprises identification information for the maritime vessel; and in response to a determination that the detected data does not comprises identification information, automatically estimate a classification for the maritime vessel using historical pattern of life data and generate a confidence level associated with the estimated classification.

The following features apply to both aspects of the invention.

The identification information may be a maritime mobile service identity (MMSI) or other identification information issued by the International Maritime Organisation (IMO) or similar organisation. The system may comprise at least one detector in the form of a receiver for receiving a signal comprising identification information. For example, the signature may be an automatic identification system (AIS) signature which is emitted from a transponder on the vessel. The receiver may be an AIS transceiver.

The historical pattern of life data may comprise historical AIS data or other identification information together with associated tracks and ports of call. For example, the historical tracks of individual vessels may be stored in the historical pattern of life data. The historical pattern of life data may include the changing locations of a plurality of vessels over relatively short periods, e.g. at different times of day and/or the seasonal patterns of change in locations for a plurality of vessels. The historical pattern of life data may also include shipping routes and known maritime activity areas (e.g. fishing areas, anchorage, etc.).

Any other useful information may also be stored, such as non-kinematic data or other voyage related data. The historical pattern of life data may be stored in one or more databases which are accessible by the processor of the system.

As set out above, the at least one detector may comprise an AIS receiver. Additionally, or alternatively, the at least one detector may comprise a sensor, e.g. an acoustic array or other suitable sensor, for detecting bearing information. It will be appreciated that there may be other sensors as needed. For example, the method may be used on a platform, e.g. a ship, another vessel which is itself moving either on the surface of the water or under water. Thus, in addition to detecting data regarding the vessel, the method may comprise obtaining internal data relating to the moving platform and using the internal data when estimating the classification. The internal data may comprise navigational information, e.g. bearing and location, of the moving platform. The sensor(s) may capture live measurements which typically update regularly, e.g. every second.

When the identification information is not detected, the classification which is estimated may be to classify at least one of vessel identity, vessel type, predicted track and next port of call. The vessel type may be defined using any suitable terminology. Merely as an example, the types of vessels may be the vessel class such motor boats, sailing boats with sails up, sailing boats with sails down, small ferries, offshore patrol vessel, fishing vessel, merchant or other maritime vessels. A track is illustrative of the location, movement and identity of the vessel.

When the identification information is detected, the vessel may be identified from the identification information. However, such information may be interrupted, for example by the vessel diving under water. Thus, the method may comprise in response to a determination that the detected data does comprises identification information, monitoring for an interruption in receiving the identification information, and when an interruption is detected, estimating a classification for the maritime vessel using the identification information and the historical pattern of life data.

When the identification information was detected but has been interrupted, it may not be necessary to estimate the vessel identity and/or type because it may be possible to determine this information from the identification information. Accordingly, estimating the classification may comprise estimating at least one of predicted track and next port of call. Estimating the classification may comprise determining whether at least part of the identification information matches identification information in the historical data.

The estimating is done automatically, e.g. by an automated system without user intervention and/or input. Estimating the classification may comprise using a correlation algorithm. For example, expected bearing rates for a plurality of vessels may be derived from historical tracks for the plurality of vessels, the detected bearing data may be compared with the expected bearing rates and the vessel having the expected bearing rate with the maximum correlation with the detected bearing data is output as the classification. The classification may be the identity (and/or type) for the vessel with the maximum correlation. The track and next port of call may also be output based on the historical track for the vessel with the maximum correlation. The expected bearing rates may be calculated and compared using any suitable technique. For example, the expected bearing rates may be plotted on a graph with each of the plurality of vessels represented by a central mean line bounded by a number of standard deviations and the comparison may be done by plotting the detected bearing data on the same graph.

Automatically estimating the classification may comprise using a machine learning algorithm. For example, the machine learning algorithm may be trained on the historical pattern of life data. The detected data may be input to the machine learning algorithm to determine the classification.

The confidence level may also be considered when deciding whether and/or how to output the estimated classification. For example, a plurality of classifications may be estimated, each having different associated confidence levels. The confidence level may be output with the or each classification and/or each output estimated classification may be ranked. Alternatively, only the estimated classification with the highest confidence level may be output. When it is not possible to estimate the vessel identity with a confidence above a threshold, the estimated classification may be the type of vessel. The output may be on any suitable user interface or display, Traditionally, maritime analysis is track based. However, the present applicant has also recognised the need for an additional or complementary classification method.

According to another aspect of the invention, there is provided a method for determining the activity of a maritime vessel, the method comprising: detecting data from a maritime vessel; identifying the maritime vessel from the detected data; obtaining a state diagram of permitted activities associated with the identified maritime vessel, determining an activity of the maritime vessel and verifying the determined activity using the state diagram.

The data which is detected may the same as the data described above using the detectors (including sensors and/or receivers) or may be just the identification information detected by the receiver. Identifying the maritime vessel may comprise classifying the maritime vessel as described above. Thus, the two methods may be used together or in isolation from one another.

Each state diagram may be associated with a recognised maritime activity, e.g. passenger ferry service, fishing or other activities, for example as defined in European maritime activities. The state diagram may comprise a plurality of states with each state associated with a discrete activity associated with the recognised maritime activity. For example, for the overarching activity of "passenger ferry service", the discrete activities may be port operation and transit. The state diagram may also comprise the transitions between states and any associated probabilities of transition.

Determining an activity may comprise determining a current activity of the maritime vessel based on the detected data. For example, the location of the maritime vessel may be detected and the location may be used to determine the current activity, e.g. if the location is a port, the current activity may be determined as port operation. Alternatively or additionally, determining the activity may comprise determining a change of activity by detecting a transition event which causes the maritime vessel to change between a first state and a second state on the state diagram and determining the current activity to be the activity which corresponds to the second state. Any suitable method for detecting the transition event may be used. For example, detection of a port exit event is indicative of a change of state from "port operation" to "transit".

The method may further comprise predicting a future activity of the maritime vessel based on the determined current activity and the associated state diagram. For example, each probability of transitioning from a first state representing the current activity to a second, different state may be determined and the second state with the highest probability may be estimated as the future activity. Activity of the maritime vessel may be monitored to determine a next activity of the maritime vessel. The predicted future activity (i.e. the next state) may be compared with the next determined current activity. When it is determined that there is no match between the predicted future activity and the next current activity, an anomaly may be flagged, e.g. by issuing an alert.

Similarly, the method may comprise determining whether the determined current activity (e.g. the activity determined from the detected data and/or for the change of activity) corresponds to a state on the associated state diagram, and when it is determined that the determined activity does not correspond to a state, outputting an alert. For example, if it is determined that the activity (i.e. state) is "loitering" or "rendezvous" and these activities not part of the state machine, an alert may be output.

According to another aspect of the invention, there is a non-transitory computer-readable medium comprising processor control code which when running on a 30 system causes the system to carry out the method described above.

It will be appreciated that any one or more features described in relation to an aspect of the invention may replace, or be used in combination with, any one or more features described in relation to another aspect of the invention, unless such replacement or combination would be understood by the skilled person as mutually exclusive, after a reading of this disclosure. In particular, any features described in relation to apparatus-like aspects may be used in combination with, or in place of, any features described in relation to method-like aspects.

FIGURES

For a better understanding of the invention, and to show how embodiments of the same may be carried into effect, reference will now be made, by way of example, to the accompanying diagrammatic figures in which: Figure 1 is a schematic illustration of the classifying system; Figure 2 is a flowchart setting out the steps of a first method carried out by the classifying system of Figure 1; Figure 3 is a graph plotting historical tracks by bearing against time and indicating the current measurements for a detected vessel; Figure 4 is a flowchart setting out the steps of another method carried out by the classifying system of Figure 1; and Figures 5a and 5b are illustrative state diagrams which may be used in the method of Figure 4.

DETAILED DESCRIPTION

Figure 1 shows a schematic illustration of the classifying system 10 which may be used as described below to classify a detected vessel 12. The classifying may include estimating the unique identification for the vessel, estimating the type of vessel, predicting the track (i.e. location and movement) for the vessel and/or predicting the most likely route to the next port of call (NPOC) for the vessel. The unique identification may be a maritime mobile service identity (MMSI) or other identification information issued by the International Maritime Organisation (IMO) or similar organisation. The type of vessel may be defined using any suitable terminology, e.g. the open architecture radar interface standard (OARIS) pre-recognition type with semantics defined by NATO's Standardization Agreement STANAG 5516. Merely as an example, the types of vessels may be the vessel class such motor boats, sailing boats with sails up, sailing boats with sails down, small ferries, offshore patrol vessel, fishing vessel, merchant or other maritime vessels. Alternatively, the type of vessel may be defined by providing a cohort of possible maritime mobile service identities.

The vessel may be broadcasting a periodic signal using a transponder, transceiver or the like. A common format for such a periodic signal is an automatic identification system (AIS) signature, although it will be appreciated that other formats can be used. Information provided in such a signal typically comprises some or all of a unique identification, position, course and speed.

Such a signal may be detected by the classifying system 10, for example by using a detector in the form of one or more receivers 20 within the system. The vessel 12 may also send a signal in response to a query from the classifying system 10 and the response may be detected by the same or a different receiver. Such a response signal may be a response to an identification friend or foe (IFF) interrogation.

It is possible that the vessel 12 is not transmitting an AIS signal or the transmission of the signal to the classifying system 10 is interrupted, e.g. when the vessel moves underwater. The classifying system 10 may additionally detect the vessel 12 using a detector in the form of one or more other sensors 22. For example, the sensor(s) may be any suitable sensor for detecting bearing measurements, e.g. an acoustic array. The sensor(s) may capture live measurements which typically update regularly, e.g. every second. These measurements typically do not provide a direct observation of the quantities (e.g. identity and/or type) and thus the sensor(s) may be termed passive sensors. As an example comparison, a passive sensor may measure bearings and an active sensor may measure both bearings and range. Passive sensors may be considered to provide poor quality live measurements because not all the desired information is available.

The AIS signal (or similar) and sensor detected data is sent to an estimation module 20 which provides an estimated classification for the vessel together with associated confidence levels where appropriate. The estimation module 20 may also be termed a central module because it also communicates with various other modules or components of the system. The estimation module comprises components such as a processor 34 for carrying out the steps described below and a communication module 36 for facilitating the communication with the other modules/components. It will be appreciated that the functionality of the various modules could be combined in to fewer modules or separated in to more modules.

The processor 34 may classify a detected vessel as described below. Use of a processor 34 to generate an automated or automatic classification differs from known systems in which a user determines classifications, e.g. by visually associating live sensor data with shipping density plots that reflect the historical information. The historical information also known as pattern of life information which is used in the classification process may be received from a database 26 which may be external to the system (i.e. remote from, perhaps cloud-based). The database 26 may comprise historical AIS data. For example, the historical tracks of individual vessels may be stored in the database. The database may include the changing locations of a plurality of vessels over relatively short periods, e.g. at different times of day and/or the seasonal patterns of change in locations for a plurality of vessels. The database may also include shipping routes and known maritime activity areas (e.g. fishing areas, anchorage, etc.).

-10 -Any other useful information may also be stored, such as non-kinematic data or other voyage related data. It will be appreciated that the database may be separated across several discrete databases.

The classification system 10 may itself be mounted on a ship or other moving vessel which may be termed a moving platform. Passive sensors on such moving platforms typically only receive partial information about the detected vessel. For example, passive sensors may only measure an angle of contact on the detected vessel. Accordingly, internal information about the moving platform such as position and ship's head data may also be used in the classification process. This data may be termed internal data because it is received from one or more internal sensor(s) 28 which may form part of an internal system such as the navigation system. The internal sensor(s) may capture live measurements which typically update regularly, e.g. every second.

The internal data may be used to derive more detailed information, e.g. bearing information, from the sparser detected data, e.g. angle of contact data, which is captured by the passive sensor.

The output from the estimation module 20 may be transmitted to a console 24, 20 e.g. for viewing by a user of the system. The console 24 may be any suitable user interface or display. The output from the estimation module 20 may be also be transmitted to a management system 32.

The management system 32 may generate alerts, e.g. when classifying a vessel as an object of interest. Alternatively, or additionally, the management system 32 may comprise an internal performance monitoring module which can be used to log or record every time a classification of a track or a vessel is made (and also when an alert is issued, if appropriate). A threshold for the correct identification of vessels and/or tracks can be set at any suitable value between 0 to 100% and a typical confidence threshold may be 90% or 95%. The internal performance monitoring module could also be used to check how quickly the system is identifying tracks and/or classifying vessels. For example, the acceptable threshold for identifying a new track or classifying a vessel may be set at a few seconds from initial detection If this time frame is not met, an alert may be issued.

The management system 32 may optionally include an activity module 40 (or this may be a separate module within the system). Traditionally maritime analysis is track based (i.e. illustrative of the location, movement and identity of the vessel). The activity module may be used to estimate which activity is being performed by the vessel. Examples of activities are transiting, fishing, port operations, loitering and rendezvous. The list of activities may be derived from any authoritative source. The activity module may be used to determine whether the estimated activity is consistent with the cohort of vessels of the same type, within the same region and time window. The activity module may be used to output an alert when the estimated activity is not permitted, perhaps because it is in violation of a policy framework or is not consistent with the neighbouring cohort).

The activity module may be used provide an estimation or prediction of the next expected activity or series of activities. The activity module may use any suitable algorithm to make the estimations. For example, when predicting or estimating the next activity, the activity module may use a state diagram which includes the activities (e.g. port operations, transit etc) as states and events (e.g. port entry, port exit) as state transitions. The state diagram may include probabilities for each event. Examples of state diagrams are shown in Figures 5a and 5b below and are relatively compact but still useful in decision making as explained below.

Figure 2 is a flowchart showing the steps of the process which may be carried out by the estimation module of Figure 1. In a first step, data from a vessel is detected (step S100). The data may be detected by a receiver, e.g. an AIS receiver or by a passive sensor as described above. A determination is then made (step 5102) as to whether or not there is AIS data (or similar identifying -12 -data). When the AIS data is detected, the system is able to identify the detected vessel and can thus classify the detected vessel. However, if there is no AIS data or the AIS data stream is interrupted as determined in step 5104, the process proceeds to a classification phase.

During the classification phase, both internal data (i.e. navigational data from the moving platform on which the estimation module is located) and historic data (i.e. pattern of life data) are obtained (step 5106 and 5108). These are shown as sequential steps but it will be appreciated that the data can be obtained simultaneously or in any order.

The obtained data and the detected data are then input into an estimation algorithm (step S110). The detected data is compared with the historic data to output a classification for the vessel. In the instance where the AIS data was received but has been interrupted, the classification includes one or both of the predicted track and/or the next port of call. This is because the identity and type of the vessel are already known from the previously received AIS data. If the AIS data has never been received, in addition to the predicted track and/or the next port of call, the classification may include the vessel identity and/or type.

Each classification may be output with an associated confidence level.

The estimation algorithm may use any known technique to compare the detected data and the historic data. For example, known correlation techniques may be used. Alternatively, the estimation algorithm may use a machine learning algorithm (e.g. a deep neural network) which may have been subject to training based on the historic data.

Figure 3 illustrates one technique for estimating the classification of a vessel. As examples, two historical vessel tracks are shown for candidate vessels P and Q respectively. It will be appreciated that two candidate vessels are shown for ease of reference and the historical pattern of life data will typically include the data for many more vessels. The tracks show the changes in bearing -13 -against time for these vessels. The expected bearing rates of vessels P and are derived from the aggregate historical tracks (geo-referenced legs and way-points) for these vessels and the internal data, particularly kinematic parameters such as position (latitude/longitude) and velocity (course and speed). For each candidate vessel, the variation in multiple tracks is indicated by plotting a central mean (dashed line) bounded by a number of standard deviations (dotted lines). The bearing data which has been detected for the vessel which is to be classified is plotted on the graph. It is clear that the correlation is maximised for vessel Q. The degree of correlation may be used to present a confidence level.

Accordingly, the classification which may be output is the identification information (MMS1/1M0) for vessel Q together with the associated confidence level. Furthermore, the track and next port of call for the vessel in question can also be output based on the historical track.

When the identification information has been received, albeit interrupted, the identification information may be compared with the historical data to identify a match. If a match is found, the matching information may be used for the estimation, for example by obtaining a route (i.e. track) associated with the matching identification information in the historical data, obtaining the current bearing of the vessel and comparing the current bearing with the route. Using the example of Figure 3, the identification information would have identified the vessel as vessel Q. The bearing information may then be used to determine where the identified vessel is on the historical track for vessel Q. The historical track may be then used to output as a classification the track and next port of call based on the historical track for vessel Q. Figure 4 illustrates a method which may be used by the activity module of Figure 1 to provide an estimation of the next expected activity or series of activities. In a first step, data is detected from a vessel (step S200). The data which is detected may the same as the data detected in the method of Figure 2 using the detectors (including sensors and/or receivers) or may be just the identification information detected by the receiver. The vessel is then identified -14 - (step S202), e.g. using the identification information if this is received or by using the method describe above. Once the vessel is identified, the state diagram of activities associated with the identified vessel is then obtained, e.g. from a database which may be local or remote (step S204).

As examples, Figures 5a and 5b illustrate two state machines for two maritime activities: passenger ferry services and fishing. A suggested recognised source of maritime activities, which has been used by a range of bodies across Europe is defined in European maritime activities and potential environmental issues published for example in https://wwweeae Each overarching activity may then be decomposed into elemental "activity states". The aim is to establish an abstraction that sits above the "track" for the vessel and thus the activity states are necessarily reasonably broad and wide-ranging. Figure 5a illustrates the state machine for a passenger ferry service which is shown in category 5 -transport and shipbuilding of the list of activities defined above. Just two states are needed to describe the activities of a passenger ferry service: port operations and transit. The transition between states is also indicated on the state machine. Thus, as shown, a passenger ferry will remain in the state "port operations" until it exits the port and thus transitions to the state "transit". The passenger ferry will remain in the state "transit" until it enters the port thus transitions to the state "port operations". The use of such state diagrams (or state machines -the terms may be used interchangeably) provides a basic standardised vocabulary for maritime activities and a reasoning framework which may be used as explained below to highlight activities which are inconsistent with the expected activity and associated vessel type.

A more complex state machine is shown in Figure 5b which is for fishing which is shown in category 2 -extraction of living resources of the list of activities defined above for Europe. In this state machine, the activities of a passenger -15 -ferry service are also present, namely port operations and transit. Additionally, there is an activity of "fishing" which is represented as a separate state. The transitions to this state are from the state "transit" and are "start fishing" and "end fishing" as appropriate.

Returning to Figure 4, the next step is to determine the current activity of the vessel by any appropriate means (step S206), e.g. by determining the vessel is in a port from the location of the vessel. It is likely that the detectors will be probabilistic and thus a confidence level associated with the determined activity may also be output. The determined activity may then be compared with the state diagram to determine if there is an anomaly (step S208). For example, if it is detected that the current activity is "loitering" or "rendezvous" which are not part of the state machine for either type of vessel but are more typically associated with large cargo vessels, an alert may be output (step S210). It is noted that an alert that the activity is unexpected or an anomaly is not necessarily indicative of nefarious intent. For example, there are activities such as periods of vessel building, periods of maintenance, vessel rescue at sea and safety of life at sea (SOLAS) operations which may be legitimate activities. Such additional legitimate activities could also be added to the state machine if preferred.

The current activity may also be determined by detecting a change of activity (S212) is to detect a change of the state. Again, the change of activity may be determined by any appropriate means which detects a transition event which causes the maritime vessel to change between a first and second state on the state diagram. For example, detection of a port exit event will signal a change of state from "port operation" to "transit" for both types of vessel. Accordingly, the current activity may be determined to be "transit". A detector of this nature may be based on crossing a pre-defined port area geo-fence and/or looking a change in vessel kinematics from slow to high speed with a few changes in the ships head information. It is likely that the detectors will be probabilistic and thus -16 -a confidence level associated with the detected change of activity (and/or the resulting determined current activity) may also be output.

After a change of state is detected, there may be an optional check to see whether or not the new determined current activity is permitted for the identified vessel or is an anomaly (step 5214). As above, the current activity is compared with the state diagram and if the activity is not within the state diagram, an alert may be output (step S210).

The state machine will typically indicate the probability associated with the next state for the vessel. Thus, regardless of whether an alert is issued, the final stages in the method (step S216 to 5220) are based on the state diagram. For example, a prediction for the next activity may be made (step 5216), e.g. by determining which activity is most likely based on the probabilities of transitioning to different states. The real-time current activity which occurs next may then be determined (step 5218), for example as described above in relation to steps S206 or S214. The predicted next activity may be compared with the detected next current activity (step 5220) to determine whether or not there is another anomaly to be flagged.

The methods described above may be considered more generally as methods to predict activity of vessels within a geographic region. Such predictions may be output, e.g. on a navigational chart of expected activity. These predictions may then be verified with live sensor data as needed.

At least some of the example embodiments described herein may be constructed, partially or wholly, using dedicated special-purpose hardware. Terms such as 'component', 'module', 'processor' or 'unit' used herein may include, but are not limited to, a hardware device, such as circuitry in the form of discrete or integrated components, general processing units (GPUs), a Field Programmable Gate Array (FPGA) or Application Specific Integrated Circuit (ASIC), which performs certain tasks or provides the associated functionality. In -17 -some embodiments, the described elements may be configured to reside on a tangible, persistent, addressable storage medium and may be configured to execute on one or more processors. These functional elements may in some embodiments include, by way of example, components, such as software 5 components, object-oriented software components, class components and task components, processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays, and variables. Although the example embodiments have been described with reference to the components, modules 10 and units discussed herein, such functional elements may be combined into fewer elements or separated into additional elements.

Although a few preferred embodiments have been shown and described, it will be appreciated by those skilled in the art that various changes and modifications might be made without departing from the scope of the invention, as defined in the appended claims. Attention is directed to all papers and documents which are filed concurrently with or previous to this specification in connection with this application and which are open to public inspection with this specification, and the contents of all such papers and documents are incorporated herein by reference.

All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and/or all of the steps of any method or process so disclosed, may be combined in any combination, except combinations where at least some of such features and/or steps are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise. Thus, unless expressly stated otherwise, each feature disclosed is one example only of a generic series of equivalent or similar features.

Claims (22)

1. -18 -CLAIMS1 A method for classifying a maritime vessel, the method comprising: detecting data from a maritime vessel which is to be classified; determining whether the detected data comprises identification information for the maritime vessel; in response to a determination that the detected data does not comprises identification information, automatically estimating a classification for the maritime vessel using historical pattern of life data and generating a confidence level associated with the estimated classification.
2. 2. The method of claim 1, further comprising: in response to a determination that the detected data does comprises identification information, monitoring for an interruption in receiving the identification information, and when an interruption is detected, estimating a classification for the maritime vessel using the identification information and the historical pattern of life data and generating a confidence level associated with the estimated classification.
3. 3. The method of claim 1, wherein estimating the classification comprises estimating at least one of vessel identity, vessel type, predicted track and next port of call.
4. 4. The method of claim 2, wherein estimating the classification comprises estimating at least one of predicted track and next port of call.
5. The method of claim 2 or claim 3, wherein estimating the classification comprises determining whether at least part of the identification information matches identification information in the historical data.
6. 6. The method of any one of the preceding claims, wherein detecting data comprises detecting bearing information for the maritime vessel.
7. 7. The method of claim 6, comprising detecting bearing information using an acoustic array.
8. 8. The method of any one of the preceding claims, further comprising detecting the data using a sensor on a moving platform, obtaining internal data relating to the moving platform and using the internal data when estimating the classification.
9. 9. The method of any one of the preceding claims, wherein estimating the classification comprises using a correlation algorithm
10. 10. The method of any one of claims 1 to 9, wherein estimating the classification comprises using a machine learning algorithm.
11. 11. The method of any one of the preceding claims, further comprising outputting a plurality of different estimated classifications together with their associated confidence levels.
12. 12. A system for classifying maritime vessels, the system comprising: at least one detector for detecting data from a maritime vessel which is to be classified; and a processor which is configured to: detect data from the maritime vessel; determine whether the detected data comprises identification information for the maritime vessel; and in response to a determination that the detected data does not comprises identification information, automatically estimate a classification for the maritime -20 -vessel using historical pattern of life data and generate a confidence level associated with the estimated classification.
13. 13. The system of claim 12, wherein the at least one detector comprises a sensor for detecting bearing information.
14. 14. The system of claim 12 or claim 13, wherein the at least one detector comprises a receiver for receiving a signal comprising identification information.
15. 15. A method for determining the activity of a maritime vessel, the method comprising: detecting data from a maritime vessel; identifying the maritime vessel from the detected data; obtaining a state diagram of permitted activities associated with the identified maritime vessel; determining an activity of the maritime vessel; and verifying the determined activity using the obtained state diagram.
16. 16. The method of claim 15 wherein determining an activity comprises determining a current activity of the maritime vessel based on the detected data.
17. 17. The method of claim 15 or claim 16 wherein determining an activity comprises determining a current activity by detecting a transition event which causes the maritime vessel to change between a first state and a second state on the state diagram and determining the current activity to be the activity which corresponds to the second state.
18. 18. The method of claim 16 or claim 17, further comprising predicting a future activity of the maritime vessel based on the determined current activity and the associated state diagram.
19. 19. The method of claim 18, further comprising monitoring activity of the maritime vessel to determine a next activity of the maritime vessel, comparing the next activity of the maritime vessel with the predicted future activity, and when it is determined that the predicted activity does not correspond to the next activity, outputting an alert
20. 20. The method of any one of claims 15 to 17, further comprising determining whether the determined activity corresponds to a state on the associated state diagram, and when it is determined that the determined activity does not correspond to a state, outputting an alert.
21. 21. The method of any one of claims 15 to 20, wherein identifying the maritime vessel comprises classifying the maritime vessel as claimed in any one of claims 1 to 11.
22. 22. A computer-readable medium comprising processor control code which when running on a system causes the system to carry out the method of any one of claims 1 to 11 or claims 15 to 21.