GB2552967A

GB2552967A - Improved control device

Info

Publication number: GB2552967A
Application number: GB1613965.1A
Authority: GB
Inventors: Ratnakaran Nambiar Krishnan; Abdulshakur Memon Mohammadshahid
Original assignee: Individual
Current assignee: Individual
Priority date: 2016-08-15
Filing date: 2016-08-15
Publication date: 2018-02-21
Also published as: GB201613965D0; WO2018033715A1

Abstract

A timer is initiated (310) when a user operates one or more electronic devices, such as home appliances, by providing an input (302). A subsequent input is received (316), and it is determined whether or not it was received within a threshold time period of the first input (312). A usage pattern, comprising all inputs received within the time period, is generated (320, 314). Afterward, the electronic devices are preferably operated automatically according to the generated usage pattern. The usage pattern may be analysed to determine the associated power consumption, and then optimised to reduce energy usage. Ambient data may be received from sensors and correlations may be identified between the usage pattern, the power consumption and the ambient data. The usage pattern may be amended on the basis of the identified correlations to reduce power consumption. The usage patterns may be stored in a graph data structure.

Description

(54) Title of the Invention: Improved control device

Abstract Title: Method and control device for determining a behavioural usage pattern of electronic devices by a user (57) A timer is initiated (310) when a user operates one or more electronic devices, such as home appliances, by providing an input (302). A subsequent input is received (316), and it is determined whether or not it was received within a threshold time period of the first input (312). A usage pattern, comprising all inputs received within the time period, is generated (320, 314). Afterward, the electronic devices are preferably operated automatically according to the generated usage pattern. The usage pattern may be analysed to determine the associated power consumption, and then optimised to reduce energy usage. Ambient data may be received from sensors and correlations may be identified between the usage pattern, the power consumption and the ambient data. The usage pattern may be amended on the basis of the identified correlations to reduce power consumption. The usage patterns may be stored in a graph data structure.

Fig. 3

This print takes account of replacement documents submitted after the date of filing to enable the application to comply with the formal requirements of the Patents Rules 2007.

1/11

IIG. 1¼

2/11

FIG. 2

3/11

4/11 ,20

Start

402

404 .406

408 / Received pattern /-..........Graph • : \ eXiSiS >..........

Yes

Create directed graph

410

No Z^z Cycle^s>asts \ m pattern

Directed graph

.................*

Yes

412

204

Remove cycles

System state Analyser

Update graph by adding each element as node

Update all the edges between the nodes based on the provided information

418

420

5/11

Fetch current state or the system

504

If state \ and pattern are

506

.....................................$........../.....................

Label pattern with the state and update association variable

6/11

234

602

Fig. 6

7/11

220

Power consumption

A.

/

218

Load patterns
...... ..........
1	ί r.............../	784
/ J / Extract energy / / consumption Information /
	r......

Sort user inputs in the pattern based on energy consumption

210

Refined frequent patterns

706

Compare values of user inputs in the pattern with the sensor data

708

710

Optimise the values of user inputs in the pattern based on the sensors and pre - optimised values fig. 7

8/11 ,416

210

816

Fig.

9/11

918

210,

V.

Refined frequent patterns curr_state = current state of the system I

902

Fetch refined patterns '1

904

216

Z temp_pattern Created?

906

Extract patterns which affect ail the states in the system ^::z<

914 ii Append user input in the temp_pattern

912

920

908 j4hreshold_state = current state of the system

AC

I .£

Wait for user input and sta r t/restart timer

Create temp.„pattern and set ι user input as first element i threshald„state = curr„state reset temp...pattern

SB

/ Difference /

/ between cun..state _vand threshold_state greater/ than or equal to

924

No

.zz threshold _?Z	..............................................4.................................. ί Ask user to execute pattern which ;
i 922 % Ζ f 1	I matches most with temp pattern

FIG. 9

10/11 .,.4002

L1-0N

-* L2-0N

..................I

-> L3-0N

-» L4-0FF I ,-rl L2-GN

L1-0N

L4-0FF

I.....................

L3-0N

L1-0FF

L2-0FF ··> L3-0FF .1008

L2-0N ί L1-0N



	' Si

_______t____

L3-0N

L4-0FF

L3-QFF

Λ.

............·«·............

V

I L2-0FF I /

....................

. L1-0FF

User input link Edge of a directed - -- Directed graph

Vertex of directed fig. 10(A)

11/11

L2-0N

-> L4-0N „1012

J L2-0N k

LI-ON

Ή L4-0N

H L4-QFF ..........·>...........

L3-QN l· ,2-OFF

-> L4-0FF

L4-0N

L2-QN ·.

·< j ' < i ί

> ί

L1-0N I

..........-*.........-s

L3-0N I > L4-0FF

-...................

L2-0FF

Key

L2-0FF

.,.1016

L3-0FF ^; .........................

L1-0FF

User input link ·► Edge of a directed g

- - Directed graph | Vertex of directed g

FIG. 10(B)

Improved Control Device

09 16

Description

1. Field of invention

The present disclosure relates to the field of home automation, and in particular relates to a control device arranged to be operatively coupled to one or more electronic devices, such as one or more home appliances and/or systems, and which is configured to learn the usage behaviour of a user’s usage of the operatively connected electronic device by monitoring and analysing the user’s usage of the device. Once behavioural usage patterns have been learned, operation of the associated one or more electronic devices may be automated. This can result in reduced energy consumption.

2. Background of the invention

Home automation systems are becoming ever more popular. However, a shortcoming of known home automation systems is that their operation is either pre-configured or requires customised user configuration or programming to suit the user's requirements, and thus known systems often rely on trained installers to configure the system. This increases the costs forthe user, and also results in an increase in configuration time to run the installer. Preconfigured systems are inflexible, and often do not capture a user’s desired usage pattern, which may change based on season, family circumstances, finance, additional orsubtractional member to the family, interests, change of job, age etc. Since the system is pre-configured, it will always behave in the same manner regardless of any changes in its environment. In this regard existing home automation systems are not adaptable. Another shortcoming is that the majority of home automation systems are centralised systems, and thus inherently suffer from the drawbacks of such systems; eg. a single point of failure can result in a fatal failure forthe entire system.

Summary of the invention

An objective of the present invention is to address at least some of the shortcomings of the prior art, and in particular to provide a control device, such as an autonomous home automation system, which is intuitive to use and does not require any special expertise, such as programming skills or technical skills, to configure and use by the non-expert user.

Aspects of the invention provide for a control device suitable for coupling to an existing electronic home device. The control device may be configured to monitor a user’s usage of the coupled electronic home device, to generate a user usage profile on the basis of the monitored usage, and to automate operation of the coupled home electronic device on the basis of the generated user profile.

In certain embodiments the electronic home device may relate to any one or more of: a home lighting system comprising light switches; a centralised heating system comprising a thermostat; a centralised locking system; a sound system; a local communication network; any other existing home system.

In certain embodiments, the control device may be coupled to any one or more electronic household appliances, such as any one or more of: a washing machine; a dishwasher; an oven; a stove; a refrigerator; a telephone; and any other white goods appliance.

In certain embodiment, the control device comprises communication means enabling it to communicate with other remotely located devices.

In certain embodiments, a plurality of control devices each coupled to an existing electronic home device may be configured to communicate with each other to form a networked system, with each control device representing one node in the network.

09 16

An advantage associated with the networked system of control devices is that, using swarm intelligence, the system provides the user with the ability to remotely control any electronic device within the property, that is connected to this decentralised system, and the system can automate operation of the different operatively coupled electronic devices.

The control device may be configured to learn the usage behaviour of one or more users and over time, carry out tasks on the basis of the learned usage behaviour.

In certain embodiments, the control devices may be designed as a retrofit, and can be connected to an existing home electrical layout, using an existing backbox in the wall, to hold the device safely in place, for example to the existing light switches. Once the devices are connected, the user may be prompted to create a device network, and to create a name for the network of devices, and to configure the network. The networked devices form a decentralised network, wherein each node can act as a terminal for configuring the device network. Network configuration may comprise creating a name for the room or space where each device is installed, and names for operatively connected electronic home device. Each control device may comprise an input module to facilitate network configuration. The input module may comprise a simple and intuitive user interface.

An advantage associated with the networked configuration of control devices is that this enables a user to control any operatively connected electronic home device within the network, from any node within the network.

The usage patterns generated by the control devices (i.e. by the nodes) may be shared between the different nodes in the network and behavioural usage patterns may be generated in accordance with an algorithm. The generated behavioural usage patterns are associated with the user’s electronic home device usage behaviour. The generated behavioural usage patterns can then be used to automate operation of the associated electronic home device. Since the procedure of learning and generating the behavioural usage pattern is dynamic, a change in the user’s usage of the associated electronic home device is also learnt as a new usage pattern and can be used to update an existing generated behavioural usage pattern, without requiring any specific action of the user.

Brief description of the drawings

The foregoing and other objects, aspects and advantages will be better understood from the following detailed description of embodiments of the invention, with reference to the drawings, in which:

FIG. 1a is a schematic illustration of a node in accordance with an embodiment.

FIG. 1b is a schematic illustration of a networked configuration of a plurality of nodes in accordance with an embodiment.

FIG. 2 is a systems diagram illustrating the architecture and data flow within the various blocks in the system.

FIG. 3 is a flowchart illustrating the methodology for extracting the behavioural pattern from the system usage.

FIG. 4 is a flowchart illustrating the methodology to refine and optimise the behaviour extracted behavioural patterns.

FIG. 5 is a flowchart illustrating the methodology to associate behavioural patterns with the current state of the control units of the system and optimise the behavioural patterns.

FIG. 6 is a flowchart illustrating the methodology to classify regular behavioural patterns based on time, trigger and user.

FIG. 7 is a flowchart illustrating the methodology to cap electricity consumption by optimising the regular usage patterns.

FIG. 8 is a flowchart illustrating the methodology used to execute regular behavioural patterns based on user, time and trigger.

FIG. 9 is a flowchart illustrating the methodology used to perceive the user’s behaviour based on the current usage ofthe system, predict the user’s behaviour and to then execute the predicted behaviour.

09 16

Detailed description of an embodiment ofthe invention

Referring to FIG. 1a, which shows a schematic illustration ofthe modular components comprised in a control device 102, which will also be interchangeably referred to as a node henceforth, and FIG. 1b which shows a system diagram of a plurality of nodes in a networked configuration 112. 102 is an individual node that is operatively coupled to an electronic home device (not shown), and is arranged when coupled to the electronic home device to control its operation. A plurality of such nodes 102(1), 102(2), 102(3), 102(4), 102(n), are shown in a networked configuration 112, where each node is operatively coupled to a different electronic home device. For example, each node may be operatively coupled to a different light switch located within the user’s home. The nodes 102(1), 102(2), 102(3), 102(4), 102(n) are retrofit and may rely on a wireless mesh network to communicate between themselves.

In certain embodiments, each node may have a user defined location ID 104 such as the room name, room type and name ofthe operatively coupled electronic home device. Each node 102 may comprise a unique ID 106 which identifies the node within the network 112. Each node comprises hardware 108 and software components 110.

The hardware 108 components may comprise any one or more ofthe following modules 1. User Interface - which may be a touch screen or a digital switch through which the user can select a desired command.

2. Motherboard - where principle electronic components are embedded.

3. Energy monitoring device - which provides energy usage information for each electric home device connected to the node.

4. Communication module - provides the wireless mesh networking capability to the node.

5. Control module - provides the node with control over the operatively connected electronic home device. For eg. relays, dimmer etc.

Furthermore, each node 102 may also comprise any one or more of the following software components 110 —

1. User Interface - which may be a GUI for a touch screen, or embedded script in a digital switch or similar.

2. Communication - Software stack to manage the required communication protocols.

3. Behaviour Learning Algorithm - which may be an algorithm, configured to learn the usage behaviour of the user’s usage of the operatively coupled electronic home device, and autonomously control operation of the operatively coupled electronic home device in accordance with the learned usage behaviour, to enhance the user’s lifestyle. Based on the observed usage behaviour, the system 112 may cap the total electricity usage ofthe coupled electronic home device with minimal or no disruption to the user’s lifestyle.

4. Control - acts as a software control for the control module.

FIG. 2 is a process flow chart illustrating operation ofthe control device (i.e. the node) 102. Operation ofthe node 102 is broadly divided into 6 major functional modular components namely, user interface 222, power monitoring unit 224, control and actuation unit 226, communication network 228, Sensors 230 and systems core 232. It is to be appreciated that each illustrated functional modular component may be embodied within a stand-alone physical component within the device 102, or alternatively one or more illustrated functional modular components may be embedded within the same physical component within the device 102.

09 16

Each functional modular component is explained below 1. User Interface 222 facilitates system interaction for users. The user can give inputs to the system, receive feedback from the system, get relevant information and can change information like name of location and connected devices, network, etc. ofthe system from the user interface. User Interface 222 interacts with core 232 to exchange information entered by the user.

2. Power monitoring unit 224 monitors the energy consumption of every operatively coupled electronic home device connected to the networked configuration of nodes 112. This means, advantageously, that a user may view the power consumption of any electronic home device that is coupled to at least one node within the network of nodes 112, from any node comprised in the network, even if the node isn’t directly coupled to the relevant electronic home device, as this information can be shared between the networked nodes. Power consumption 218 saves the energy consumption information. Users can view the history of power usage from user interface 222. It can show information in terms of units, kWh and the cost of usage. Core 232 fetches the information from power monitoring unit 224 and sends the relevant information to user interface 222, in which the data is represented in a logical visual format.

3. Control and actuation unit 226 is tasked with controlling the connected devices based on the commands received from core 232. Control and actuation unit shares state change information with core 232 and power monitoring unit 224. A non-limiting, illustrative example of connected electronic home devices may be non-dimmable lights, dimmable lights, fan, thermostat, etc.

4. Communication network 228 acts as an access point enabling the device 102 to interact with other nodes 102(n) in the network. A user can control any ofthe electronic home devices operatively connected to any node 102(n) in the network 112. Other key information for core 232 and power consumption information are also exchanged between nodes using the communication network 228.

5. Sensors 230 provides core 232 with key information about the environment around the node. In certain embodiments it may comprise of motion sensors, temperature and climate sensors, ambient light sensors, etc. Sensor information can be used to optimise the usage of the connected electronic home devices. For example, if the motion sensor does not detect any motion in the room, then the system can turn off the connected electronic home devices or set them to power saving mode. Similarly, a temperature sensor can support the system to optimise the efficiency of connected heating I cooling devices.

6. Core 232 is the heart ofthe system. It interacts with User interface 222, power monitoring unit 224, control and actuation unit 226, communication network 228, and sensors 230. All the components of core 232 and their functionalities are detailed a. Input 214 receives data from Sensors 230, Communication Network 228 and user interface 222. Input 214 refines the received data and feeds the data to Director 216 and Pattern Listener 212, in the correct format.

b. Director 216 is the main decision maker in the core 232 and sends information to Control and actuation unit 226, communication network 228 and user interface 222. It obtains all the information that is received by input 214. Director 216 keeps track of usage behaviour learnt by the device 102, data derived by the power consumption module 218 and data derived by the power usage capping 220 module. Based on all the information received by director 216, it then decides the appropriate action and communicates to the required component comprised in the device 102 to carry out the appropriate action. It also feeds pattern analyser 208 with the current state of all the operatively connected electronic home devices.

c. Pattern Listener 212 gathers the user’s usage information. For example, turning to FIG. 1b, consider node 102(1) has two non-dimmable lights connected to its control and actuation unit 226. Assume that both lights are in the OFF state. If the user

09 16 switches ON both the lights, then Pattern Listener 212 will observe that user behaviour and feed the information to Pattern Analyser 208. It will capture information of usage pattern as well as the time and information of the user who was responsible for that behaviour. It also maintains the information of the time taken between two events in the pattern.

d. Pattern Analyser 208 receives the observed usage pattern from the Pattern Listener 212. It then analyses the usage pattern and refines the data of the usage pattern. It removes the cycles from the usage pattern. There is a possibility where usage pattern may have frequent state toggling of the operatively coupled devices or usage pattern might have repeated sub patterns. If the user switches ON and switches OFF the lights repeatedly, then that behaviour is not an intended usage. Such behaviour can be considered as a child playing with the switch or an unintentional usage. These types of observed repeated patterns are removed. The Pattern Analyser also analyses the usage pattern with the current system state to decide if the usage pattern needs to be modified based on the current state or not. The Current system state may be defined as the current operating state of all the operatively coupled devices. In certain circumstances the user may want the system to be in a particular state, but there could be a plurality of ways of attaining the desired state. For example, if the user wants to switch OFF all the operatively coupled devices then he/she will switch OFF all the devices which are not already OFF. Alternatively, it could be a case that the user might have to switch OFF only one device of a plurality of devices if the other devices are already OFF, or the user might have to switch OFF all devices if all of them are ON. In both the usage patterns, the intended system state is the same. Hence, in such situations, it is necessary to analyse the usage pattern of the current state of the system. Once the pattern is refined, the Pattern Analyser 208 saves the pattern in the Directed Graph 204. Directed Graph 204 is a database which stores all the usage pattern’s Directed Graph related information.

e. Frequent pattern learning 202 applies an association rule learning algorithm, such as a frequent itemset mining algorithm on the directed graph and sends the results to the Analysing and Labelling frequent patterns module 206.

f. Analysing and Labelling frequent patterns module 206 classifies the behavioural patterns based on parameters comprising any one or more of: trigger, user and time. The trigger can be user triggered or sensor triggered. For example, if the room has a motion sensor, then the system will sense motion when the user enters the room and uses the node 102. In this example the trigger would be the motion sensor data. There is scope of incorporating a user identification device to add security to the system. In such a case, Analysing and Labelling frequent pattern 206 may classify patterns based on user identification. This is advantageous in that it means the behavioural usage patterns of different users may be monitored. Usage of the connected electronic home devices may be different for different users. Similarly, usage patterns for each user may also differ depending on the day and time, so the observed behavioural patterns may be labelled based on time as well. The Analysing and Labelling frequent pattern module 206 may also receive data from the power usage capping module 220, to refine and optimise behavioural pattern controls on request. It stores the refined patterns into the refined frequent pattern 210.

g. Power usage capping module 220, is responsible for optimizing and refining observed behavioural usage patterns to reduce electricity consumption with no or minimal disruption to the user’s comfort. If a user requests power capping as an option via the user interface 222 during configuration, then the power usage capping module 220 analyses all the user’s behavioural usage patterns and tries to identify all possible ways in which electricity consumption associated with the operatively connected electronic home device can be reduced, and determines the best way to reduce the electricity consumption. For example, consider the example where the control device (i.e. the node 102) is operatively coupled to the central heating system within a house, and specifically to the boiler. If the user sets the desired temperature at X degrees Celsius, the node 102 will analyse the available history of central heating temperature information for that particular time of day from data previously collected by the sensors 230. If the device 102 determines that X degrees Celsius is too high, on the basis of the previously collected temperature information, then the device 102 will automatically refine the temperature setting to a lower value to reduce electricity consumption.

FIG. 3 illustrates the methodology adopted by the pattern listener module 212. Let us assume the user is using the user interface on the node 102(1). At step 302 the user input into the user interface module 222 of the node 102(1) is read. At step 304, it is determined if the user’s control input is for an electronic home device operatively connected to the same node 102(1). If it does not belong to the same node, then the method loops back to the start point at step 302. If instead it is determined that the input relates to the operatively connected electronic home device, then at step 306 a new behavioural usage pattern is created, and at step 308 the user control input is assigned as the first element of the pattern. In the next step, at step 310, a threshold timer is initiated, which times the interval between consecutively received user control inputs. For example, if the user switches ON a first light at time t1, and switches ON a second light at time t2, and the time lapsed is less than a predefined threshold timer, such that the condition t2 -11 < threshold timer

09 16 holds, then the pattern listener module 212 proceeds to step 316 where receipt of further data input is awaited. If instead it is determined at step 312 that the threshold timer has expired, then the observed usage pattern is forwarded to the pattern analyser 208 at step 314, and the process loops back to step 302. At step 316, it is determined if further control inputs from the user interface 222 or from the communication network 228 have been received before the threshold timer has expired. The method proceeds to step 318, where the received user control input is appended to the pattern created at step 306, along with the time and user information. The method then at step 320 determines if the pattern length equals to a threshold length. If not then the method loops back to step 310, where the threshold timer is restarted. If instead at step 320 it is determined that the pattern length equals threshold length, then the method proceeds to step 314, where the pattern is sent to the pattern analyser 208, and the method loops back to step 302, where the method is begun afresh.

FIG. 4 is a process flow chart illustrating the methodology adopted by the pattern Analyser module 208 comprised in the device 102. Pattern Analyser receives pattern data from the pattern listener 212, at step 404. At step 406 it is determined if a directed graph exists in the directed graph database 204. A directed graph is set of vertices that are connected together, where all the edges are directed from one vertex to another, and within the present context may graphically display a usage pattern. If the directed graph does not exist, a new directed graph is created at step 408, which is stored in the directed graph database 204. Once the directed graph has been stored, the method proceeds to step 410. If, at step 406, it is determined that the directed graph exists, then the method proceeds to step 410. At step 410, it is determined if any cycles exists in the relevant pattern data. A cycle is defined as a repeated observed usage of an operatively coupled electronic home device. For example, if a received pattern data comprises inputs associated with repeatedly switching the same light switch ON and OFF, this constitutes a cycle. If it is determined at step 410 that a cycle does exist in the received pattern data, then at step 412 a filter is applied to the pattern data to remove the user inputs associated with the identified cycle repeated sub patterns from the usage pattern. For example, the removal of cycles from the pattern data may be envisaged by again considering the scenario where the device 102 is operatively coupled to a light switch. Assume that the user switches the lights ON and followed by an almost immediate OFF and then ON quickly thereafter. Removal of the cycle may comprise preserving the latest user input, which means removing the first ON and OFF inputs from the pattern data, whilst preserving the last ON input for the associated light switch. Once the cycles have been removed, the method proceeds to step 414. At step 414 further analysis of the pattern is carried out which will be explained in detail in FIG. 5. At step 416 the Directed graph is updated by assigning each user input comprised in the usage pattern as a vertex in the directed graph. At step 418, the edges of the graph are updated by creating edges between the user input vertices in the Directed graph. Each Edge contains important information like time, user, count, etc. Detailed description of the graph is explained on FIG 10(A) and FIG (10B).

FIG. 5 is a process flow chart providing further details of step 414 (labelled system state analyser 414) of FIG. 4, which is carried out by the pattern analyser module 208. The object of this process is to analyse the observed usage pattern with the state of operatively connected electronic home devices. This may be understood by reference to the previous illustrative example in which the electronic home devices may relate to a plurality of different light switches controlling a plurality of different lights. Assume that the networked system comprises two nodes 102(1) and 102(2). Each node is operatively connected to a light switch configured to independently control operation of two non-dimmable lights. We will call the connected lights of node 102(1) as L11 and L12 and of node 102(2) as L21 and L22. Each light can be in one of two states, either ON or OFF. Consider an initial state in which all the lights are in the OFF state. The initial state of the lights may therefore be expressed as L11-OFF, L12-OFF, L21-OFF, L22-OFF.

Now assume the user switches ON both the lights of node 102(1) from its user interface. The pattern will then be 09 09 16

L11-ON, L12-ON

Hence, the state of the lights after the pattern execution will be L11-ON, L12-ON, L21-OFF, L22-OFF.

Now consider that the initial state of the lights may therefore be expressed as L11-OFF, L12-OFF, L21-OFF, L22-ON.

Now assume the user switches ON both the lights of node 102(1) from its user interface. The pattern will then be L11-ON, L12-ON

Hence, the state of the lights after the pattern execution will be L11-ON, L12-ON, L21-OFF, L22-ON.

The method commences at step 502, where the pattern analyser 208 receives the information concerning the current state of the operatively coupled electronic home devices from the director 216, and the current system of the electronic home devices is read. At step 504 it is determined if the user inputs comprised in the pattern affects all the operatively connected electronic home devices, i.e. the current state of the operatively coupled electronic devices is consistent with usage pattern. If it is determined at step 504, then at step 512 the pattern is designated as the current state of the operatively coupled electronic home devices. If instead it is determined at step 504 that the current state of the operatively connected electronic devices are not consistent with usage pattern, the method proceeds to step 506. At step 506, usage pattern data is labelled with current state of the operatively connected electronic home devices. There is the possibility that the resulting state of the operatively connected electronic devices will be different for each execution of the usage pattern. It can be understood from the above mentioned example where the resulted state of all the operatively connected devices is different after the execution of same usage pattern. At step 506, the resulting state of the operatively connected devices due to the usage pattern is stored. Step 506 also maintains and updates an association variable which stores information regarding the frequency of each different resulting state of the operatively connected device. The association variable is also grouped with association variable across different usage patterns. For example in the above example we assume that the initial state is all four lights are ON. Now if the user desires the state in which all the lights in the system are OFF, then the user has to switch OFF four different lights. Now assume that three of the four lights are ON, and the user desires the state of all lights in the system to be OFF. In such a scenario the user needs to switch OFF the remaining three lights. Thus, in both examples the desired state of all operatively connected devices is the same - i.e. OFF - but the usage patterns are different. In such situations, a weight association variable associated with the state is increased, even where the same state results due to different usage patterns. Frequency of the occurrence of the state, repetition of the usage pattern resulting to the same state and time of the event affects the association variable. At step 508, it is determined if the association variable is greater than a threshold value related to the current state of all the operatively connected devices. If the association variable is greater than the threshold, then at step 512, it is designated that the current state is consistent with the usage pattern, and the usage pattern is equal to the state of the operatively connected devices. This method designates the current system state as a usage pattern because sometimes usage patterns are intended to set the whole system to the particular state. If instead it is determined at step 508 that the association variable is not greater than the threshold value, then the pattern is not changed at step 510.

09 16

FIG. 6 is a process flow chart illustrating the methodology adopted by module 234 comprising the Frequent pattern Learning 202 and Analysing and labelling frequent patterns 206 modules comprised in the device 102. Steps 602 and 604 occur in frequent pattern learning module 202. Remaining steps occur in Analysing and labelling frequent pattern module 206. The method commences at step 602, where the directed graph is fetched from the database 204. At step 604, frequent itemset mining algorithm is applied on the loaded directed graph. Frequent patterns are classified at steps 606, 608 and 610. At step 606, frequent patterns are classified based on first event in the frequent pattern. Frequent patterns are classified as user triggered and sensor triggered (eg. motion sensor). Sensor triggered frequent patterns can have sensor input at first element in the pattern while user triggered starts with normal user inputs. At step 608, patterns are classified based on the users if device is provided with user identification. At step 610, patterns are classified based on time of their execution. At step 612, the processed frequent patterns and classification information is stored in the database 210.

FIG. 7 is a process flow chart illustrating the methodology adopted by the power usage capping module 220 comprised in the device 102. The method commences at step 702, where refined frequent pattern is loaded from database 210. At step 704, method scans all the refined patterns data and matches that with the associated device. Based on that, it fetches power consumption information of those associated devices from database 218. At step 706, the user inputs in the refined patterns are sorted based on their power consumption. At step 708, user inputs in refined frequent patterns is compared with the corresponding sensor information. For example if a user input in a pattern denotes ramp up the dimmable light to 100 percent, then the function 708 will compare the ambient light sensor information before and after the user input was executed. At step 710, user input values are tuned to optimal values based on sensor data and other source of information like scientific research, geographical location, local weather, etc. The user values are optimized so that the energy consumption goes down. Which user inputs need to optimize first is based on sorted user input. After the optimization of user values, the information is stored back to the database 210.

FIG. 8 is a process flow chart illustrating one of the many methodologies adopted by the Director module 216 comprised in the device 102. The method commences at step 802, where refined frequent pattern and all relevant information is loaded from database 210. At step 804, refined patterns which affects the state of all the devices in the system is filtered out and remaining refined patterns are sustained. This methodology is applied for the refined patterns which affect not all, but certain operatively connected devices when executed. For example, if two rooms have 2 lights each, then the pattern which includes the control of all the 4 devices are removed from the data in this method. Step 806 awaits user input. When user input is received, the process proceeds to step 808. At step 808, the user input is analysed to see whether the user input is related to the device connected to the same node from which the user has initiated the user input. For example, if user switches ON the light which belongs to the room 1, from a node within room 1, then the condition at 808 is satisfied. If the light in room 2 was switched ON from the node in room 1, then it wouldn’t satisfy the condition 808. This mechanism is adopted for simplicity of autonomous execution of the system. Any pattern starting from a particular room, then the user should be present in that particular room to execute the associated refined frequent pattern. If 808 is satisfied then 810, checks if any refined pattern related to the user starts with the user input. At step 812, the process checks if the time of user input and time of the usage patterns which satisfied conditions at 808 and 810 are within the time constraints. If the process has the refined pattern satisfying the conditions at 808, 810 and 812, then at step 814, the process requests the user if he/she wants to execute the whole refined pattern autonomously. At step 816, the user response is awaited by the system. If the user accepts the request of executing the complete refined pattern, then at step 818, all the control inputs of the refined pattern are automatically triggered. For example, consider a refined pattern [{room 1 - light 1 ON, room 1 - light 2 ON, room 2 light 1 OFF}, user 1, time t]. If at time t, user 1 switches ON light 1 in room 1 from a node within room 1, then the user will be asked if he/she wants to trigger the refined pattern autonomously. If the user accepts, then the system autonomously switches ON the light 2 in rooml and switches OFF light 1 in room 2. If any of the conditions in 806, 808, 810, 812 and 816 are not satisfied, then the system loops back to step 802 where it would fetch the updated refined pattern information once again and proceed further.

09 16

FIG. 9 is a process flow chart illustrating one of the many methodologies adopted by the Director module 216 comprised in the device 102. The method commences at step 902, where refined frequent pattern and all relevant information is loaded from database 210. At step 904, refined patterns which affect the state of all the devices in the system are obtained, and remaining refined patterns are removed. This methodology is applied for the refined patterns which affects all operatively connected devices when executed. For example, if two rooms have 2 lights each, then the pattern which includes the control of all the 4 devices are used for this method. This method is designed to predict the intended user behaviour when user gives multiple inputs in short time difference. At step 906, variables threshold_state is initialised as current state of the system. Step 908 then awaits user input. Once the user input is received, the timer starts/restarts. The timer is started to get the time between two consecutive user inputs. Step 908 will start the timer if no timer exists or expired or it will reset the timer in case timer is active. The process then proceeds to step 918 where the curr_state variable is updated as current state of the system. This is done because, the user input affects the state of the system and we need to keep curr_state variable up-to date with the current state of the system. At step 916, the process checks if temp_pattern variable is created or not. This variable temporarily stores the usage pattern which will be used in this method later. If temp_pattern variable is not created, then the step 912 will create one and assign the current user input as the first element of the temp_pattern. If the temp_pattern already exists then step 914 will simply append the current user input to the temp_pattern variable. At step 920, method waits for the user input and checks if the timer is expired or not. If timer expired and no user input was received, then at step 910, the threshold_state is set as curr_state and temp_pattern is deleted. Threshold_state is reference state which needs to be updated because user is no longer trying to give multiple inputs to identify the intended usage pattern. Method loops back to 908 where method awaits user input. At step 920, if user input is received before threshold time expires, then the process proceeds to step 922, where the difference between curr_state and threshold_state are checked. If the difference is greater than or equal to the threshold, then at step 924 the process tries to match the temp_pattern variable with the existing frequent usage patterns extracted during step 904. 924 then requests user to execute the frequent usage pattern which has the closest match with the temp_pattern variable. If at step 922, the difference is less than the threshold, then the process proceeds to step 908.

For example, assume that the system has three nodes 102(1), 102(2) and 102(3). Each node is connected with two non-dimmable lights. Let us name the connected lights of node 102(1) as L11 and L12, node 102(2) as L21 and L22 and node 102(3) as L31 and L32. All the lights will have two states,

09 16

ON or OFF. Let us also assume that the current state of all the lights in the system are ON. So the initial state of the system will be L11-ON, L12-ON, L21-ON, L22-ON.L31-ON, L32-ON Let us assume that we have one refined pattern which affects all the connected devices in the system

L11-OFF, L12-OFF, L21-OFF, L22-OFF.L31-OFF, L32-OFF

Now let us also assume the threshold of curr_state and threshold_state difference is 50% and timer count is 5 seconds. User input 1,2 and 3 happens consecutively within 5 seconds time difference.

Initial Condition:

Threshold_state = L11-ON, L12-ON, L21-ON, L22-ON.L31-ON, L32-ON

User input 1:

User switches OFF L11

Threshold_state = L11-ON, L12-ON, L21-ON, L22-ON, L31-ON, L32-ON

Curr_state = L11-OFF, L12-ON, L21-ON, L22-ON, L31-ON, L32-ON

Difference = ¹/₆ (100)% = 16.66%

Temp pattern = L11-OFF

User input 2:

Before the timer expires, User switches OFF L12

Threshold_state = L11-ON, L12-ON, L21-ON, L22-ON, L31-ON, L32-ON

Curr_state = L11-OFF, L12-OFF, L21-ON, L22-ON, L31-ON, L32-ON

Difference = 2/6(100) % = 33.33%

Temp pattern = L1 l-OFF, L12-OFF

User input 3:

Before timer expired, User switches OFF L21

Threshold_state = L11-ON, L12-ON, L21-ON, L22-ON, L31-ON, L32-ON

Curr_state = L11-OFF, L12-OFF, L21-OFF, L22-ON, L31-ON, L32-ON

Difference = 3/6(100) % = 50%

Temp pattern = L11 -OFF, L12-OFF, L21 -OFF

This satisfies the condition at 922, after which the function 924 matches the temp_pattern with refined patterns and asks user if he/she wants to execute the refined pattern L11-OFF, L12-OFF, L21-OFF, L22-OFF,L31-OFF, L32-OFF If the user accepts, the system will turn OFF all the remaining lights autonomously.

Consider the following example where timer expires before the new user input is received:

Initial Condition:

Threshold_state = L11-ON, L12-ON, L21-ON, L22-ON.L31-ON, L32-ON

User input 1:

User switches OFF L11

Threshold_state = L11-ON, L12-ON, L21-ON, L22-ON, L31-ON, L32-ON

Curr_state = L11-OFF, L12-ON, L21-ON, L22-ON, L31-ON, L32-ON

Difference = %(100)% = 16.66%

Temp pattern = L11-OFF

User input 2:

After the timer expires, User switches OFF L12

Threshold_state = L11-OFF, L12-ON, L21-ON, L22-ON, L31-ON, L32-ON

Curr_state = L11-OFF, L12-OFF, L21-ON, L22-ON, L31-ON, L32-ON

Difference = ¹/₆ (100) % = 16.66%

Temp pattern = L12-OFF

User input 3:

Before timer expired, User switches OFF L21

Threshold_state = L11-OFF, L12-ON, L21-ON, L22-ON, L31-ON, L32-ON

Curr_state = L11-OFF, L12-OFF, L21-OFF, L22-ON, L31-ON, L32-ON

Difference = 2/6(100) % = 33.33%

Temp pattern = L12-OFF, L21-OFF

09 16

User input 4:

Before timer expired, User switches OFF L22

Threshold_state = L11-OFF, L12-ON, L21-ON, L22-ON, L31-ON, L32-ON

Curr_state = L11-OFF, L12-OFF, L21-OFF, L22-OFF, L31-ON, L32-ON

Difference = 3/6(100) % = 50%

Temp pattern = L12-OFF, L21-OFF,L22-OFF

FIG. 10(A) and FIG. 10(B) illustrates the methodology of generating directed graph from usage patterns. Let us assume user has 4 lights which he/she can control namely L1, L2, L3 and L4. Each light can be either switched ON or OFF. 1002 in FIG. 10(A) illustrates the refined pattern where user has switched ON lights L1, L2 and L3 in the given order and switched OFF L4 after that. User control inputs which happened first will have direct links with all the user control inputs which happened after. Hence in 1002 we have links between L1-ON user control input to all other user control inputs in the usage pattern. Similarly user control input L2-ON have links with all other user control inputs but not with L1-ON as it happened before L2-ON. The sequence is maintained to simplify the association of consecutive user inputs. 1004 in FIG. 10(A) demonstrates the directed graph. In this directed graph, all the user control inputs are assigned as vertex of the graph, and all the links between the user control inputs are assigned as the edge of the directed graph. Each edge contains important information like count, time, user, and information required for frequent itemset mining algorithm. 1006 in FIG. 10(A), 1010 in FIG. 10(B) and 1014 in FIG. 10(B) are usage patterns. We assume that these usage patterns are executed in sequence.

Similarly 1008 in FIG. 10(A), 1012 in FIG. 10(B) and 1016 in FIG. 10(B) are directed graph. They demonstrate the generation of the directed graph after execution of each usage patterns 1006, 1010 and 1014.

09 16

Claims

1. A method of determining a usage pattern of one or more electronic devices by a user, using a control device suitable for coupling to the one or more electronic devices, the method comprising:

receiving a first input for operation of the one or more electronic devices; initiating a timer upon receipt of the first input;

receiving a second input for operation of the one or more electronic devices;

determining if the second input was received within a threshold time period; and generating a usage pattern associated with the received inputs, the usage pattern comprising all inputs received within the threshold time period.

2. The method of Claim 1, comprising:

automating operation of the one or more electronic devices in accordance with the generated usage pattern.

3. The method of any preceding Claim, comprising: analysing the generated usage pattern;

determining a power consumption associated with the usage pattern; and optimizing the usage pattern to reduce the power consumption.

4. The method of Claim 3, wherein the control device or the one or more electronic devices comprise one or more ambient sensors, wherein the optimizing comprises:

receiving ambient data from the one or more ambient sensors;

identifying one or more correlations between the usage pattern, the power consumption and the received ambient data; and amending the usage pattern on the basis of the one or more identified correlations to reduce the power consumption associated with the usage pattern.

5. A control device, suitable for coupling to one or more electronic devices, the control device comprising:

a receiver for receiving a first and second input for operation of the one or more electronic devices;

a timing module arranged to measure an amount of time lapsed between the first and second received inputs;

a determining module arranged to determine if the second input was received within a threshold time period from the first input;

a usage pattern generator module arranged to generate a usage pattern associated with the received inputs, the usage pattern comprising all inputs received within the threshold time period; and a control module arranged to control subsequent operation of the one or more electronic devices in accordance with the generated usage pattern.

6. The control device of Claim 5, comprising:

a power consumption usage module arranged to determine a power consumption associated with the usage pattern;

one or more ambient sensors configured to measure ambient data associated with one or more characteristics of an environment encompassing the control device;

a correlation identification module arranged to identify one or more correlations between the measured ambient data, the power consumption and the usage pattern; and an optimization module arranged to amend the usage pattern on the basis of the one or more identified correlations to reduce the power consumption associated with the usage pattern.

Intellectual

Property

Office

Application No: GB1613965.1 Examiner: Dr Maurice Blount