CN107665230A - Training method and device for the users' behavior model of Intelligent housing - Google Patents
Training method and device for the users' behavior model of Intelligent housing Download PDFInfo
- Publication number
- CN107665230A CN107665230A CN201710477025.4A CN201710477025A CN107665230A CN 107665230 A CN107665230 A CN 107665230A CN 201710477025 A CN201710477025 A CN 201710477025A CN 107665230 A CN107665230 A CN 107665230A
- Authority
- CN
- China
- Prior art keywords
- node
- user
- vector data
- knowledge mapping
- characteristic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000012549 training Methods 0.000 title claims abstract description 35
- 239000013598 vector Substances 0.000 claims abstract description 87
- 238000013507 mapping Methods 0.000 claims abstract description 58
- 230000003542 behavioural effect Effects 0.000 claims abstract description 45
- 238000003062 neural network model Methods 0.000 claims abstract description 13
- 230000002123 temporal effect Effects 0.000 claims abstract description 11
- 238000006243 chemical reaction Methods 0.000 claims description 17
- 230000007704 transition Effects 0.000 claims description 17
- 238000004378 air conditioning Methods 0.000 description 12
- 230000006870 function Effects 0.000 description 8
- 238000005057 refrigeration Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 5
- 230000000306 recurrent effect Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B15/00—Systems controlled by a computer
- G05B15/02—Systems controlled by a computer electric
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/901—Indexing; Data structures therefor; Storage structures
- G06F16/9024—Graphs; Linked lists
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Automation & Control Theory (AREA)
- Quality & Reliability (AREA)
- Manufacturing & Machinery (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the present invention provides the training method and device of a kind of users' behavior model for Intelligent housing, wherein, this method includes：Historical behavior data based on user's control intelligent home device, build the user behavior knowledge mapping for controlling intelligent home device, wherein, the user behavior knowledge mapping includes at least two behavioral chain of user, the behavioral chain include facility information, control information and temporal information and its between corresponding relation；The behavioral chain in the user behavior knowledge mapping is converted into characteristic vector data；Based on the characteristic vector data, using users' behavior model of the neural network model training generation for intelligent home device control.The method and device provided according to embodiments of the present invention, the users' behavior model trained, it can accurately predict the imminent behavior of user.
Description
Technical field
The present embodiments relate to Smart Home technical field, more particularly to a kind of user for Intelligent housing
The training method and device of behavior prediction model.
Background technology
With the development of smart home, smart home intellectuality be not limited to equipment it will be appreciated that user from
Right language, also require that equipment can make anticipation to user behavior, and the equipment for meeting user's request is made based on anticipation result
Use recommendation.Therefore, how to realize the accurate anticipation of user behavior is current urgent problem to be solved.
The content of the invention
The embodiment of the present invention provides the training method and dress of a kind of users' behavior model for Intelligent housing
Put, to train the higher users' behavior model of accuracy, realize the Accurate Prediction to user behavior.
First aspect of the embodiment of the present invention provides a kind of training of users' behavior model for Intelligent housing
Method, this method include：
Historical behavior data based on user's control intelligent home device, build the use for controlling intelligent home device
Family behavior knowledge collection of illustrative plates, wherein, the user behavior knowledge mapping includes at least two behavioral chain of user, in the behavioral chain
Including facility information, control information and temporal information and its between corresponding relation；
The behavioral chain in the user behavior knowledge mapping is converted into characteristic vector data；
Based on the characteristic vector data, intelligent home device control is used for using neural network model training generation
Users' behavior model.
Second aspect of the embodiment of the present invention provides a kind of training of users' behavior model for Intelligent housing
Mounted cast trainer, the device include：
Module is built, for the historical behavior data based on user's control intelligent home device, is built for controlling intelligence
The user behavior knowledge mapping of home equipment, wherein, the user behavior knowledge mapping includes at least two behavioral chain of user,
The behavioral chain include facility information, control information and temporal information and its between corresponding relation；
Conversion module, for the behavioral chain in the user behavior knowledge mapping to be converted into characteristic vector data；
Training module, for based on the characteristic vector data, being used for intelligent family using neural network model training generation
Occupy the users' behavior model of equipment control.
The embodiment of the present invention, by the historical behavior data based on user's control intelligent home device, build for controlling
The user behavior knowledge mapping of intelligent home device, and the behavioral chain in user behavior knowledge mapping is converted into characteristic vector
Data, realize that the vectorization of the historical behavior data of user's control intelligent home device represents；And then represented based on vectorization
User's history behavioral data, the user's behavior prediction of intelligent home device control to be used for using neural network model training generation
Model.Because the embodiment of the present invention uses the user's history behavioral data of vectorization expression, the Information Meter of data is high, data dimension
Degree is big, comprising user behavior information, facility information and temporal information it is more, what is trained is used for intelligent home device control
Users' behavior model can provide the user corresponding equipment recommendation information according to the imminent behavior of user, improve
The recommendation degree of accuracy of facility information, improves smart home intelligence degree, enhances the usage experience of user.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will to embodiment or
The required accompanying drawing used is briefly described in description of the prior art, it should be apparent that, drawings in the following description are only
Some embodiments of the present invention, for those of ordinary skill in the art, without having to pay creative labor,
Other accompanying drawings can also be obtained according to these accompanying drawings.
Fig. 1 is a kind of instruction for users' behavior model for Intelligent housing that one embodiment of the invention provides
Practice the schematic flow sheet of method；
Fig. 2 is a kind of structural representation for user behavior knowledge mapping that one embodiment of the invention provides；
Fig. 3 is a kind of execution method flow diagram for step 102 that one embodiment of the invention provides；
Fig. 4 is a kind of execution method flow diagram for step 201 that one embodiment of the invention provides；
Fig. 5 is the behavior prediction accuracy rate curve map that experiment is drawn；
Fig. 6 is a kind of instruction for users' behavior model for Intelligent housing that one embodiment of the invention provides
Practice the structural representation of model training apparatus；
Fig. 7 is the structural representation for the conversion module 12 that another embodiment of the present invention provides；
Fig. 8 is the structural representation for the first conversion submodule 121 that one embodiment of the invention provides.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art obtained under the premise of creative work is not made it is all its
His embodiment, belongs to the scope of protection of the invention.
The term " comprising " and " having " of description and claims of this specification and their any deformation, meaning
Figure be to cover it is non-exclusive include, for example, the device for the process or structure for containing series of steps is not necessarily limited to clearly
It is that those structures or step for listing but may include are not listed clearly or for intrinsic other of these processes or device
Step or structure.
Fig. 1 is a kind of instruction for users' behavior model for Intelligent housing that one embodiment of the invention provides
Practice the schematic flow sheet of method, this method can be performed by terminal or server.As shown in Fig. 1, what the present embodiment provided
Method comprises the following steps：
Step 101, the historical behavior data based on user's control intelligent home device, build for controlling smart home
The user behavior knowledge mapping of equipment, wherein, the user behavior knowledge mapping includes at least two behavioral chain of user, described
Behavioral chain include facility information, control information and temporal information and its between corresponding relation.
In the present embodiment involved historical behavior data include facility information that user operates in historical time section,
Control information and temporal information and its between corresponding relation.
The acquisition methods of historical behavior data include the possible implementation of the following two kinds in the present embodiment：
In a kind of possible implementation, the historical behavior data of user are sent by equipment during user's operation equipment
Stored to terminal or server, when performing the method for the present embodiment, terminal or server obtain user from local
Historical behavior data.
In alternatively possible implementation, what equipment storage user operated on it in different time points goes through
History behavioral data, when performing the method for the present embodiment, terminal or server obtain the historical behavior of user from each equipment
Data.
Each node represents some of an equipment or equipment in user behavior knowledge mapping involved by the present embodiment
Pattern, the corresponding mode node of each device node are connected, and some mould in behavior knowledge collection of illustrative plates is triggered as user
Some equipment of formula node, i.e. user's control starts some pattern, in special time, such as 10 minutes, 20 minutes or 30 minutes
Then trigger another mode node in behavior knowledge collection of illustrative plates again Deng, user, i.e. another equipment of user's control starts certain
Individual pattern, and so on sequence of operations ultimately form the behavioral chain of user.The same in behavior knowledge mapping
Side on behavioral chain is connected on same node, and the connection of different behavioral chains is on different nodes.Under in order to make it easy to understand,
Face combines an example to be illustrated to the behavior knowledge collection of illustrative plates provided in the present embodiment：
Fig. 2 is a kind of structural representation for user behavior knowledge mapping that one embodiment of the invention provides, shown in Fig. 2
User behavior knowledge mapping include behavioral chain i, in behavioral chain i user perform successively enabling, turn on light, close window, open
The operation of TV is cooked and opened to air-conditioning automatic mode, electric cooker.The time that each operation performs be stored in each operation with
On the side of node i connection, according to the time stored in each edge, you can determine the execution sequence respectively operated in behavioral chain i.
Step 102, the behavioral chain in the user behavior knowledge mapping is converted into characteristic vector data.
Fig. 3 is a kind of execution method flow diagram for step 102 that one embodiment of the invention provides, as shown in Fig. 3, step
102 include：
Node in step 201, the user behavior knowledge mapping is characterized vector data.
Fig. 4 is a kind of execution method flow diagram for step 201 that one embodiment of the invention provides, and in the present embodiment, is used
Family behavior knowledge collection of illustrative plates includes start node and other nodes, as shown in figure 4, step 201 includes：
Step 2011, currently processed destination node is selected from other nodes.
The user behavior knowledge mapping obtained for structure, a node of user behavior knowledge mapping is chosen to be just
Beginning node, be considered as starting point, user behavior knowledge mapping can include multiple nodes, in addition to start node, it is also multiple with it is initial
The associated other nodes of node.
, can be by calculating transition probability value, from a number of other nodes as a kind of example of concrete application of the present invention
Selected currently processed destination node, by calculating transition probability value, selected corresponding transition probability value is more than the first predetermined threshold value
Node be destination node.Transition probability value is the size for the weight that can represent the contact between node.Transition probability value is also
Can be the value that other represent weight size, for example, transition probability matrix, adjacency matrix, certainly, above it is merely meant that node
Between relation information weight size example, the size of any weight that can represent the relation information between node
Value can serve as transition probability value, the embodiment of the present invention is not limited specifically this.
Step 2012, the first eigenvector for obtaining the start node, and, obtain the second of the destination node
Characteristic vector.
In the embodiment of the present invention, all nodes assign one initially in the user behavior knowledge mapping that structure obtains
Vector, this initial vector can be the numerical value of multidimensional, and numerical value assigns at random, it is impossible to represents the contact between node and knot
Structure feature, but after updating initial vector after the method for passing through the embodiment of the present invention, it can be very good to represent between different nodes
Contact and architectural feature, initial vector can include the first eigenvector of start node, or destination node second feature to
Amount.The first eigenvector of start node is the vector of dimension more than one being manually set, and the second feature of destination node is vectorial
Equally it is the vector of dimension more than one being manually set.
Using the embodiment of the present invention, the start node in user behavior knowledge mapping can be defined as dimension more than one
First eigenvector, dimension can be 100 dimensions or 50 dimension, for example, definition initial solid " air-conditioning " vector
Change and represent that f (n) is first eigenvector, i.e. f (air-conditioning)=[0.543,0.381,0.328 ... 0.182], wherein, dimension is
The of 100 dimensions (number be 100) in first eigenvector, and assume that destination node is " refrigeration ", then destination node " refrigeration "
Two characteristic vectors are expressed as f (refrigeration)=[0.337,0.169,0.401 ... 0.403], wherein, dimension is 100 dimensions.
Certainly, the setting for dimension can be determined that the present invention is to this by those skilled in the art according to actual conditions
It is not restricted.
Step 2013, according to the first eigenvector and second feature vector, it is corresponding to calculate the destination node
Characteristic value.
, can be special according to the first eigenvector and second in a kind of preferred embodiment of concrete application of the present invention
Sign vector, calculate the bar of the second feature vector of target start node under conditions of the first eigenvector with start node
Part probable value because one in knowledge mapping start node be generally connected with multiple target start nodes, therefore can obtain more
The conditional probability value of individual different target start node, the method for design conditions probable value can use softmax functional expression tables
Show, tired to multiply the conditional probability value, obtain multiple conditional probability value multiplies value, and the multiple conditional probability value is multiplied
Value, which is taken the logarithm and then added up, obtains characteristic value.
Step 2014, determine eigenvalue of maximum in the characteristic value.
Applied to the embodiment of the present invention, record described according to the first eigenvector and second feature vector, calculating institute
The execution number for the step of stating characteristic value corresponding to destination node, it can be understood as obtain the number of characteristic value, held described in judgement
Whether places number is more than the second predetermined threshold value, when the number is more than the second predetermined threshold value, selects in the characteristic value
Eigenvalue of maximum.Wherein, the second predetermined threshold value is arranged to those skilled in the art depending on actual conditions, and the present invention is real
Example is applied not to be restricted this.
Step 2015, for the eigenvalue of maximum, using its corresponding first eigenvector and second feature vector more
Newly obtain the characteristic vector data of the start node and other nodes.
In embodiments of the present invention, can be by constantly adjusting first eigenvector and after characteristic value is calculated
Every one-dimensional value of two characteristic vectors, so as to obtain different characteristic values.For each characteristic value, maximum characteristic value is selected
Corresponding first eigenvector and second feature vector, as start node and the vectorization table of other nodes (destination node)
Show, because after all knot vectorizations represent in user behavior knowledge mapping, if the vectorization expression of all nodes can
Good relation information and architectural feature of the expression node in user behavior knowledge mapping, at this moment, node " air-conditioning " surrounding phase
The conditional probability value that adjacent other node sets occur will be maximum.
Because characteristic value is obtained by conditional probability by certain operations, thus, definable is to all node n in collection of illustrative plates
Maximized characteristic value, characteristic value can include target function value.When characteristic value be not maximum when, represent it is now corresponding
First eigenvector and second feature vector are not optimal solutions.
It is at this time, it may be necessary to one-dimensional or more in continuing adjustment (increase or reduce) first eigenvector and second feature vector
The number of dimension, obtain different characteristic values, choose first eigenvector value and multiple second feature corresponding to the characteristic value of maximum to
Value represents as the vectorization of node, otherwise, updates in first eigenvector and second feature vector, and return to foundation
First eigenvector and second feature vector, corresponding to calculating destination node the step of characteristic value.
Still by taking Fig. 2 as an example, in the behavior knowledge collection of illustrative plates shown in Fig. 2, the relation between equipment and pattern can be by expressing
Formula G=(V, E) is indicated, all nodes for representing equipment and pattern wherein in V representative graphs 3, all in E representative graphs 2
Side.The neural network model of similar term vector, the probability of occurrence of a word are related to the word of its context.In Fig. 2 is calculated
During the character representation of node, the conditional probability definition occurred according to the adjacent point set of a point maximizes object function (i.e. in advance
If knowledge mapping vectorization model)：
Wherein f (n) is node n vector representation, and its dimension is d, and d value is the value of random initializtion, false in this example
If d=100, object function is maximized come adjusting and optimizing f (n) parameter by the training of model.So, knowledge mapping vector
Changing model just has | V | × d parameter needs to estimate.N (n) is node n adjacent node set.P (N (n) | f (n)) it is when row
For when all nodes are quantified expression in knowledge mapping, conditional probability that node n adjacent node set occurs.If behavior
In knowledge mapping vectorization a little represent to can be good at expressing relation and architectural feature of each node in collection of illustrative plates,
Then the conditional probability of the adjacent point set of all nodes is up to maximum in V, i.e., above-mentioned object function reaches maximization.And then
It is the vector representation form that can obtain each node in Fig. 2 according to above-mentioned object function.
Due to be between node n its adjacent node it is separate, then
Node n point n adjacent theretoiConditional probability p (ni| f (n)) their respective characteristic vector dot products can be used
Softmax function representations.
Adjacent node set N (n) the most common graph search algorithm of acquisition includes BFS (BFS) and depth
First search (DFS).The adjacent node set of one node of sampling may be different in each iterative process, can carry out more
Secondary iteration carries out obtaining different N (n), and the node n so obtained vector representation form can more embody node n and know in behavior
Know the characteristic in collection of illustrative plates.
It should be noted that an iterations can be set in the embodiment of the present invention, select in iterations most
Big characteristic value, ensure that the performance of hardware is without prejudice, the setting of iterations can be by those skilled in the art according to actual
Situation determines, the invention is not limited in this regard.
Step 202, the characteristic vector data based on the node, by the row in the user behavior knowledge mapping
Characteristic vector data is converted into for chain.
In the embodiment of the present invention, some pattern corresponding to a certain moment some equipment can be with user behavior knowledge mapping
Represented by the Hadamard products of the two nodes：
Vec (operation)=Vec (equipment) * Vec (pattern)
Wherein, Vec (equipment) is the vector representation of equipment, and Vec (pattern) is the vector representation of equipment associative mode.
So, the behavioral data of user can carry out data expression, and as the training data of users' behavior model,
For example, the behavioral chain i of user：(x1,x2...xT) can be expressed as follows：
x1=Vec (door) * Vec (opening), x2=Vec (lamp) * Vec (opening), x3=Vec (window) * Vec (pass), x4=Vec
(air-conditioning) * Vec (automatic) ... ....
Step 103, based on the characteristic vector data, being set using neural network model training generation for smart home
The users' behavior model of standby control.
Applied to the embodiment of the present invention, the characteristic vector data of behavioral chain in user behavior knowledge mapping is inputted default
Neural network model in, training obtain users' behavior model.
Wherein, default neural network model can be such as any one in drag：Hidden Markov model HMM,
Time series models, Logic Regression Models and recurrent neural networks model etc..Wherein, recurrence is preferably used in the present embodiment
Neural network model.
By taking recurrent neural networks model as an example, it is assumed that the behavioral chain of user is (x1,x2...xT), xtIt is user in t
The operation that moment is carried out to equipment, then the probability of behavior chain be：
The principle of recurrent neural networks model training is continuous more by setting of user's subsequent time to equipment or operation
New hidden layer：
Wherein, σ is chosen for sigmoid functions, and W, U are parameter to be determined.
The training goal of recurrent neural networks model is to make p (x1,x2...xT) maximum, wherein, g (ht)=p (xt|x1,
x2...xt-1), g is chosen for sigmoid or tanh functions.Therefore, the behavioral chain in the behavior knowledge collection of illustrative plates of vectorization is inputted
Recurrent neural networks model, use gradient descent method constantly to adjust W, U value and make it that model is optimal, the present embodiment can be obtained
In required for behavior prediction model.
The training method of behavior prediction model has following advantage compared with prior art in the present embodiment：
(1) in terms of equipment, pattern, the correlation embodiment between the operating time
Prior art represents history controlling behavior of the user to equipment by the form of list daily record.Equipment, pattern, behaviour
Make record of the time integrally as a whole piece, embodiment device, pattern, correlation and difference between the operating time can not be distinguished
Property, for example, operation note " air-conditioning, refrigeration, 2016-08-10 18:00 ", " air-conditioning, automatically, 2016-08-20 18:00 ", " electricity
Depending on opening, 2016-08-10 20:It is separate, mutually incoherent, the different mode of similar same equipment between 00 "
The similitude of control, otherness of distinct device control etc. can not embody, and can influence the effect of behavior prediction model trained
Fruit and accuracy rate.
And the present embodiment is based on user's history behavioral data, user behavior knowledge mapping is built, and by user behavior knowledge
Behavioral chain in collection of illustrative plates carries out vectorization expression, has that the Information Meter of data is high, and data dimension is big, comprising user behavior letter
Breath, the advantages of facility information and temporal information are more, and the contact between user behavior can be embodied.For example, the system of air-conditioning
The automatic mode of chill formula and air-conditioning, knowledge mapping vectorization expression being carried out, Vec (refrigeration), Vec (automatic) similarity is very high,
Vec (air-conditioning, refrigeration) and Vec (air-conditioning, automatic) similitude are so given expression to higher than Vec (air-conditioning, refrigeration) and Vec (electricity
Depending on opening), user behavior is not what is be independent of each other, and the correlation between controlling behavior is represented with otherness by vectorization
User's history behavioral data expressed, improve and improve the effect and accuracy rate of behavior prediction model.
(2) in terms of data dimension
Prior art represents user to the operation behavior of equipment by list daily record, increase for number of devices and sets
The increase of standby pattern, the data for increasing list daily record corresponding to a pattern represent to be increased by one-dimensional, and it is dilute to be easily caused training data
Dredge and dimension is huge, on the other hand, the difficulty of model training can be increased, increase the training time of model.
And by the way that the behavioral chain in user behavior knowledge mapping is converted into characteristic vector data in the present embodiment so that
The dimension of data is fixed, can't be increased with number of devices, equipment has the increase of pattern quantity and increases, and is made
Reached with distributed vector table and Sparse Problems are not present, and the burden of model training will not increase therewith, the training time also will not
It is multiplied.
Particularly, table one is the recognition accuracy contrast table by testing obtained user behavior, and table two is to pass through experiment
Obtained model training time contrast table, Fig. 5 are the behavior prediction accuracy rate curve map that experiment is drawn.From experiment, knowledge
The mode that the accuracy rate of collection of illustrative plates vectorization expression reaches apparently higher than list log sheet.And with increasing for equipment, equipment mode
Increase, declined based on list expression accuracy rate obvious, but the accuracy rate of knowledge mapping vectorization expression had no significant effect,
RNN models accuracy rate is higher than HMM, LR model.RNN is more suitable for solving the problems, such as time series input and output, its neutral net meeting
The historical behavior data of user are remembered and are applied in the calculating currently exported.
In summary, the present embodiment, by the historical behavior data based on user's control intelligent home device, structure is used
In the user behavior knowledge mapping of control intelligent home device, and the behavioral chain in user behavior knowledge mapping is converted into spy
Vector data is levied, realizes that the vectorization of the historical behavior data of user's control intelligent home device represents；And then it is based on vectorization
The user's history behavioral data of expression, user's row of intelligent home device control to be used for using neural network model training generation
For forecast model.Because the embodiment of the present invention uses the user's history behavioral data of vectorization expression, the Information Meter of data is high,
Data dimension is big, comprising user behavior information, facility information and temporal information it is more, what is trained is used for intelligent home device
The users' behavior model of control can provide the user corresponding equipment recommendation letter according to the imminent behavior of user
Breath, improves the recommendation degree of accuracy of facility information, improves smart home intelligence degree, enhance the usage experience of user.
Table one
Table two
Fig. 6 is a kind of instruction for users' behavior model for Intelligent housing that one embodiment of the invention provides
Practice the structural representation of model training apparatus, as shown in fig. 6, the device includes：
Module 11 is built, for the historical behavior data based on user's control intelligent home device, is built for controlling intelligence
The user behavior knowledge mapping of energy home equipment, wherein, the user behavior knowledge mapping includes at least two behavior of user
Chain, the behavioral chain include facility information, control information and temporal information and its between corresponding relation；
Conversion module 12, for the behavioral chain in the user behavior knowledge mapping to be converted into characteristic vector number
According to；
Training module 13, for based on the characteristic vector data, being used for intelligence using neural network model training generation
The users' behavior model of home equipment control.
The device that the present embodiment provides can be used in the method for performing Fig. 1 embodiments, its executive mode and beneficial effect class
Seemingly, repeat no more herein.
Fig. 7 is the structural representation for the conversion module 12 that another embodiment of the present invention provides, as shown in fig. 7, in Fig. 6 institutes
On the basis of showing embodiment, conversion module 12 also includes：
First conversion submodule 121, for the Node in the user behavior knowledge mapping to be characterized into vectorial number
According to；
Second conversion submodule 122, for the characteristic vector data based on the node, by the user behavior knowledge
The behavioral chain in collection of illustrative plates is converted into characteristic vector data.
The device that the present embodiment provides can be used in the method for performing embodiment illustrated in fig. 3, its executive mode and beneficial effect
Fruit seemingly repeats no more herein.
Fig. 8 is the structural representation for the first conversion submodule 121 that one embodiment of the invention provides, as shown in figure 8,
In device shown in Fig. 7, the first conversion submodule 121 includes：
Subelement 1211 is selected, for selecting currently processed destination node from other nodes；
Subelement 1212 is obtained, for obtaining the first eigenvector of the start node, and, obtain the target
The second feature vector of node；
Computation subunit 1213, for according to the first eigenvector and second feature vector, calculating the mesh
Mark characteristic value corresponding to node；
Determination subelement 1214, for determining the eigenvalue of maximum in the characteristic value；
First renewal subelement 1215, for for the eigenvalue of maximum, using its corresponding first eigenvector and
Second feature vector renewal obtains the characteristic vector data of the start node and other nodes.
Optionally, the start node and other nodes have corresponding word knot vector data, selection respectively
Unit；
The selection subelement 1211, is specifically used for：
Using the knot vector data of the start node, and, the knot vector data of other nodes, calculate
Transition probability value；
Judge whether the transition probability value is more than the first predetermined threshold value；
When the transition probability value is more than the first predetermined threshold value, other nodes corresponding to the transition probability value are determined
For destination node.
Optionally, the computation subunit 1213, is specifically used for：
According to the first eigenvector and second feature vector, conditional probability value corresponding to the destination node is calculated；
Tired to multiply the conditional probability value, acquisition is tired to multiply conditional probability value；
Operation of taking the logarithm is carried out for the tired conditional probability value that multiplies, obtains logarithm conditional probability value；
Add up the logarithm conditional probability value, obtains characteristic value.
Optionally, the first conversion submodule 121, in addition to：
Second renewal subelement 1216, for for off-peak characteristic value, update its corresponding first eigenvector and
Second feature vector.
Optionally, the determination subelement 1214, is specifically used for：
Record calculates characteristic value corresponding to the destination node according to the first eigenvector and second feature vector
The step of execution number；
Judge whether the execution number is more than the second predetermined threshold value；
When the number is more than the second predetermined threshold value, the eigenvalue of maximum in the characteristic value is selected.
The device that the present embodiment provides can be used in the method for performing embodiment illustrated in fig. 4, its executive mode and beneficial effect
Fruit seemingly repeats no more herein.
Finally it should be noted that：Various embodiments above is merely illustrative of the technical solution of the present invention, rather than its limitations；
Although the present invention is described in detail with reference to foregoing embodiments, it will be understood by those within the art that：Its
The technical scheme described in foregoing embodiments can still be modified, it is either special to which part or whole technologies
Sign carries out equivalent substitution；And these modifications or replacement, the essence of appropriate technical solution is departed from various embodiments of the present invention
The scope of technical scheme.
Claims (10)
- A kind of 1. training method of users' behavior model for Intelligent housing, it is characterised in that methods described bag Include：Historical behavior data based on user's control intelligent home device, build the user behavior for controlling intelligent home device Knowledge mapping, wherein, the user behavior knowledge mapping includes at least two behavioral chain of user, and the behavioral chain includes setting Standby information, control information and temporal information and its between corresponding relation；The behavioral chain in the user behavior knowledge mapping is converted into characteristic vector data；Based on the characteristic vector data, using user row of the neural network model training generation for intelligent home device control For forecast model.
- 2. according to the method for claim 1, it is characterised in that the row by the user behavior knowledge mapping Characteristic vector data is converted into for chain, including：Node in the user behavior knowledge mapping is characterized vector data；Based on the characteristic vector data of the node, the behavioral chain conversion in the user behavior knowledge mapping is characterized Vector data.
- 3. according to the method for claim 2, it is characterised in that the user behavior knowledge mapping include start node with Other nodes；The Node by the user behavior knowledge mapping is characterized vector data, including：Currently processed destination node is selected from other nodes；The first eigenvector of the start node is obtained, and, the second feature for obtaining the destination node is vectorial；According to the first eigenvector and second feature vector, characteristic value corresponding to the destination node is calculated；Determine the eigenvalue of maximum in the characteristic value；For the eigenvalue of maximum, obtained using its corresponding first eigenvector and second feature vector renewal described initial The characteristic vector data of node and other nodes.
- 4. according to the method for claim 3, it is characterised in that the start node and other nodes are respectively with corresponding Word knot vector data, it is described to select currently processed destination node from other nodes, including：Using the knot vector data of the start node, and, the knot vector data of other nodes, it is general to calculate transfer Rate value；Judge whether the transition probability value is more than the first predetermined threshold value；When the transition probability value is more than the first predetermined threshold value, it is target to determine other nodes corresponding to the transition probability value Node.
- 5. the method according to claim 3 or 4, it is characterised in that described according to the first eigenvector and described Two characteristic vectors, characteristic value corresponding to the destination node is calculated, including：According to the first eigenvector and second feature vector, conditional probability value corresponding to the destination node is calculated；Tired to multiply the conditional probability value, acquisition is tired to multiply conditional probability value；Operation of taking the logarithm is carried out for the tired conditional probability value that multiplies, obtains logarithm conditional probability value；Add up the logarithm conditional probability value, obtains characteristic value.
- 6. according to the method for claim 3, it is characterised in that the eigenvalue of maximum determined in the characteristic value, bag Include：Record is according to the first eigenvector and second feature vector, the step of calculating characteristic value corresponding to the destination node Execution number；Judge whether the execution number is more than the second predetermined threshold value；When the number is more than the second predetermined threshold value, the eigenvalue of maximum in the characteristic value is selected.
- A kind of 7. training pattern trainer of users' behavior model for Intelligent housing, it is characterised in that bag Include：Module is built, for the historical behavior data based on user's control intelligent home device, is built for controlling smart home The user behavior knowledge mapping of equipment, wherein, the user behavior knowledge mapping includes at least two behavioral chain of user, described Behavioral chain include facility information, control information and temporal information and its between corresponding relation；Conversion module, for the behavioral chain in the user behavior knowledge mapping to be converted into characteristic vector data；Training module, for based on the characteristic vector data, being set using neural network model training generation for smart home The users' behavior model of standby control.
- 8. device according to claim 7, it is characterised in that the conversion module includes：First conversion submodule, for the Node in the user behavior knowledge mapping to be characterized into vector data；Second conversion submodule, for the characteristic vector data based on the node, by the user behavior knowledge mapping The behavioral chain is converted into characteristic vector data.
- 9. device according to claim 8, it is characterised in that the user behavior knowledge mapping include start node and Other nodes；The first conversion submodule, including：Subelement is selected, for selecting currently processed destination node from other nodes；Subelement is obtained, for obtaining the first eigenvector of the start node, and, obtain the second of the destination node Characteristic vector；Computation subunit, for according to the first eigenvector and second feature vector, calculating the destination node pair The characteristic value answered；Determination subelement, for determining the eigenvalue of maximum in the characteristic value；First renewal subelement, for for the eigenvalue of maximum, using its corresponding first eigenvector and second feature Vector renewal obtains the characteristic vector data of the start node and other nodes.
- 10. device according to claim 9, it is characterised in that the start node and other nodes are respectively with corresponding Word knot vector data, the selection subelement, be specifically used for：Using the knot vector data of the start node, and, the knot vector data of other nodes, it is general to calculate transfer Rate value；Judge whether the transition probability value is more than the first predetermined threshold value；When the transition probability value is more than the first predetermined threshold value, it is target to determine other nodes corresponding to the transition probability value Node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710477025.4A CN107665230B (en) | 2017-06-21 | 2017-06-21 | Training method and device of user behavior prediction model for intelligent home control |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710477025.4A CN107665230B (en) | 2017-06-21 | 2017-06-21 | Training method and device of user behavior prediction model for intelligent home control |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107665230A true CN107665230A (en) | 2018-02-06 |
| CN107665230B CN107665230B (en) | 2021-06-01 |
Family
ID=61121883
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710477025.4A Active CN107665230B (en) | 2017-06-21 | 2017-06-21 | Training method and device of user behavior prediction model for intelligent home control |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107665230B (en) |
Cited By (44)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108320061A (en) * | 2018-03-20 | 2018-07-24 | 北京工业大学 | A kind of window trend prediction method and system based on neural network |
| CN108681490A (en) * | 2018-03-15 | 2018-10-19 | 阿里巴巴集团控股有限公司 | For the vector processing method, device and equipment of RPC information |
| CN108710947A (en) * | 2018-04-10 | 2018-10-26 | 杭州善居科技有限公司 | A kind of smart home machine learning system design method based on LSTM |
| CN108919669A (en) * | 2018-09-11 | 2018-11-30 | 深圳和而泰数据资源与云技术有限公司 | A kind of smart home dynamic decision method, apparatus and service terminal |
| CN109212988A (en) * | 2018-09-21 | 2019-01-15 | 中国联合网络通信集团有限公司 | Intelligent home furnishing control method and system |
| CN109299724A (en) * | 2018-07-17 | 2019-02-01 | 广东工业大学 | Smart home user based on deep learning manipulates habit excavation and recommended method |
| CN109330503A (en) * | 2018-12-20 | 2019-02-15 | 江苏美的清洁电器股份有限公司 | Cleaning household electrical appliance and its control method and system |
| CN109361766A (en) * | 2018-12-04 | 2019-02-19 | 安徽信息工程学院 | A kind of internet of things net controller based on SDN |
| CN109684548A (en) * | 2018-11-30 | 2019-04-26 | 内江亿橙网络科技有限公司 | A kind of data recommendation method based on user's map |
| CN109709813A (en) * | 2018-12-20 | 2019-05-03 | 无锡小天鹅股份有限公司 | Application model display methods, device and household electrical appliance |
| CN110377829A (en) * | 2019-07-24 | 2019-10-25 | 中国工商银行股份有限公司 | Function recommended method and device applied to electronic equipment |
| CN110456647A (en) * | 2019-07-02 | 2019-11-15 | 珠海格力电器股份有限公司 | Intelligent household control method and intelligent household control device |
| CN110471301A (en) * | 2019-08-23 | 2019-11-19 | 宁波智轩物联网科技有限公司 | A kind of smart home service recommendation system and method based on user behavior |
| CN110543102A (en) * | 2018-05-29 | 2019-12-06 | 珠海格力电器股份有限公司 | method and device for controlling intelligent household equipment and computer storage medium |
| CN110598916A (en) * | 2019-08-23 | 2019-12-20 | 宁波智轩物联网科技有限公司 | Method and system for constructing user behavior model |
| CN110750653A (en) * | 2019-10-22 | 2020-02-04 | 中国工商银行股份有限公司 | Information processing method, information processing apparatus, electronic device, and medium |
| CN110895721A (en) * | 2018-09-12 | 2020-03-20 | 珠海格力电器股份有限公司 | Method and device for predicting electric appliance function |
| CN110989370A (en) * | 2019-11-06 | 2020-04-10 | 珠海格力电器股份有限公司 | Intelligent household interaction control method and system |
| CN111000492A (en) * | 2019-11-08 | 2020-04-14 | 尚科宁家(中国)科技有限公司 | Intelligent sweeper behavior decision method based on knowledge graph and intelligent sweeper |
| CN111103805A (en) * | 2018-10-25 | 2020-05-05 | 珠海格力电器股份有限公司 | Method, system and device for controlling household appliance and household appliance |
| CN111306803A (en) * | 2020-03-01 | 2020-06-19 | 苏州淘喜网络科技有限公司 | Water source supply control system and method based on big data |
| CN111736481A (en) * | 2020-07-14 | 2020-10-02 | 北京自如信息科技有限公司 | Model training and intelligent home control method and device based on user behavior characteristics |
| CN111798260A (en) * | 2019-04-09 | 2020-10-20 | Oppo广东移动通信有限公司 | User behavior prediction model construction method and device, storage medium and electronic equipment |
| CN111913921A (en) * | 2019-05-08 | 2020-11-10 | 中国移动通信集团福建有限公司 | User behavior analysis method and device, equipment and storage medium |
| WO2020238385A1 (en) * | 2019-05-31 | 2020-12-03 | 华为技术有限公司 | Apparatus service control method, cloud server, and smart home system |
| CN112084376A (en) * | 2020-09-04 | 2020-12-15 | 北京明略昭辉科技有限公司 | Map knowledge based recommendation method and system and electronic device |
| CN112286150A (en) * | 2020-10-15 | 2021-01-29 | 珠海格力电器股份有限公司 | Intelligent household equipment management method, device and system and storage medium |
| CN112380334A (en) * | 2020-12-04 | 2021-02-19 | 三星电子（中国）研发中心 | Intelligent interaction method and device and intelligent equipment |
| CN112587932A (en) * | 2020-12-30 | 2021-04-02 | 网易（杭州）网络有限公司 | Game plug-in detection method and device, electronic equipment and storage medium |
| CN112668384A (en) * | 2020-08-07 | 2021-04-16 | 深圳市唯特视科技有限公司 | Knowledge graph construction method and system, electronic equipment and storage medium |
| CN112734132A (en) * | 2021-01-22 | 2021-04-30 | 珠海格力电器股份有限公司 | Equipment recommendation method and device, electronic equipment and storage medium |
| CN112749625A (en) * | 2020-12-10 | 2021-05-04 | 深圳市优必选科技股份有限公司 | Time sequence behavior detection method, time sequence behavior detection device and terminal equipment |
| CN112782988A (en) * | 2020-12-30 | 2021-05-11 | 深圳市微网力合信息技术有限公司 | Control method of intelligent household curtain based on Internet of things |
| CN113485144A (en) * | 2021-07-27 | 2021-10-08 | 广州市威士丹利智能科技有限公司 | Intelligent household control method and system based on Internet of things |
| CN113848738A (en) * | 2021-09-22 | 2021-12-28 | 深圳市欧瑞博科技股份有限公司 | Control method and device of intelligent equipment |
| CN113885706A (en) * | 2021-10-19 | 2022-01-04 | 清华大学 | Interaction control method, device and system |
| CN113934930A (en) * | 2021-10-08 | 2022-01-14 | 宜兴市旭航电子有限公司 | User preference prediction system and method based on probability and hidden Markov model |
| CN114114949A (en) * | 2022-01-12 | 2022-03-01 | 慕思健康睡眠股份有限公司 | Intelligent home control system and method based on big data analysis |
| EP4030349A1 (en) * | 2021-01-18 | 2022-07-20 | Siemens Aktiengesellschaft | Neuromorphic hardware for processing a knowledge graph represented by observed triple statements and method for training a learning component |
| EP4030350A1 (en) * | 2021-01-18 | 2022-07-20 | Siemens Aktiengesellschaft | Neuromorphic hardware and method for storing and/or processing a knowledge graph |
| CN115268282A (en) * | 2022-06-29 | 2022-11-01 | 青岛海尔科技有限公司 | Control method and device of household appliance, storage medium and electronic device |
| CN115542755A (en) * | 2022-05-31 | 2022-12-30 | 青岛海尔智能家电科技有限公司 | Method and device for predicting equipment control command, electronic equipment and storage medium |
| WO2023159821A1 (en) * | 2022-02-23 | 2023-08-31 | 青岛海尔科技有限公司 | Method and device for determining operational behavior, storage medium, and electronic device |
| CN118134539A (en) * | 2024-05-06 | 2024-06-04 | 山东传奇新力科技有限公司 | User behavior prediction method based on intelligent kitchen multi-source data fusion |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011061329A1 (en) * | 2009-11-20 | 2011-05-26 | Zerogroup Holding Oü | A method and system for controlling environmental conditions of entity |
| CN105652677A (en) * | 2016-02-24 | 2016-06-08 | 深圳众乐智府科技有限公司 | Intelligent home control method, device and system based on user behavior analysis |
| CN106383450A (en) * | 2016-11-10 | 2017-02-08 | 北京工商大学 | Smart home user behavior analyzing system and smart home user behavior analyzing method based on big data |
| CN106503035A (en) * | 2016-09-14 | 2017-03-15 | 海信集团有限公司 | A kind of data processing method of knowledge mapping and device |
| CN106649774A (en) * | 2016-12-27 | 2017-05-10 | 北京百度网讯科技有限公司 | Artificial intelligence-based object pushing method and apparatus |
-
2017
- 2017-06-21 CN CN201710477025.4A patent/CN107665230B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2011061329A1 (en) * | 2009-11-20 | 2011-05-26 | Zerogroup Holding Oü | A method and system for controlling environmental conditions of entity |
| CN105652677A (en) * | 2016-02-24 | 2016-06-08 | 深圳众乐智府科技有限公司 | Intelligent home control method, device and system based on user behavior analysis |
| CN106503035A (en) * | 2016-09-14 | 2017-03-15 | 海信集团有限公司 | A kind of data processing method of knowledge mapping and device |
| CN106383450A (en) * | 2016-11-10 | 2017-02-08 | 北京工商大学 | Smart home user behavior analyzing system and smart home user behavior analyzing method based on big data |
| CN106649774A (en) * | 2016-12-27 | 2017-05-10 | 北京百度网讯科技有限公司 | Artificial intelligence-based object pushing method and apparatus |
Cited By (58)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108681490B (en) * | 2018-03-15 | 2020-04-28 | 阿里巴巴集团控股有限公司 | Vector processing method, device and equipment for RPC information |
| CN108681490A (en) * | 2018-03-15 | 2018-10-19 | 阿里巴巴集团控股有限公司 | For the vector processing method, device and equipment of RPC information |
| CN108320061A (en) * | 2018-03-20 | 2018-07-24 | 北京工业大学 | A kind of window trend prediction method and system based on neural network |
| CN108710947A (en) * | 2018-04-10 | 2018-10-26 | 杭州善居科技有限公司 | A kind of smart home machine learning system design method based on LSTM |
| CN110543102A (en) * | 2018-05-29 | 2019-12-06 | 珠海格力电器股份有限公司 | method and device for controlling intelligent household equipment and computer storage medium |
| CN109299724A (en) * | 2018-07-17 | 2019-02-01 | 广东工业大学 | Smart home user based on deep learning manipulates habit excavation and recommended method |
| CN109299724B (en) * | 2018-07-17 | 2022-01-28 | 广东工业大学 | Intelligent household user control habit mining and recommending method based on deep learning |
| CN108919669A (en) * | 2018-09-11 | 2018-11-30 | 深圳和而泰数据资源与云技术有限公司 | A kind of smart home dynamic decision method, apparatus and service terminal |
| CN108919669B (en) * | 2018-09-11 | 2022-04-29 | 深圳和而泰数据资源与云技术有限公司 | Intelligent home dynamic decision method and device and service terminal |
| CN110895721A (en) * | 2018-09-12 | 2020-03-20 | 珠海格力电器股份有限公司 | Method and device for predicting electric appliance function |
| CN109212988B (en) * | 2018-09-21 | 2021-07-06 | 中国联合网络通信集团有限公司 | Intelligent household control method and system |
| CN109212988A (en) * | 2018-09-21 | 2019-01-15 | 中国联合网络通信集团有限公司 | Intelligent home furnishing control method and system |
| CN111103805A (en) * | 2018-10-25 | 2020-05-05 | 珠海格力电器股份有限公司 | Method, system and device for controlling household appliance and household appliance |
| CN109684548B (en) * | 2018-11-30 | 2024-02-09 | 索信达(深圳)软件技术有限公司 | Data recommendation method based on user map |
| CN109684548A (en) * | 2018-11-30 | 2019-04-26 | 内江亿橙网络科技有限公司 | A kind of data recommendation method based on user's map |
| CN109361766A (en) * | 2018-12-04 | 2019-02-19 | 安徽信息工程学院 | A kind of internet of things net controller based on SDN |
| CN109709813B (en) * | 2018-12-20 | 2022-08-12 | 合肥美的洗衣机有限公司 | Application mode display method and device and household appliance |
| CN109709813A (en) * | 2018-12-20 | 2019-05-03 | 无锡小天鹅股份有限公司 | Application model display methods, device and household electrical appliance |
| CN109330503A (en) * | 2018-12-20 | 2019-02-15 | 江苏美的清洁电器股份有限公司 | Cleaning household electrical appliance and its control method and system |
| CN111798260A (en) * | 2019-04-09 | 2020-10-20 | Oppo广东移动通信有限公司 | User behavior prediction model construction method and device, storage medium and electronic equipment |
| CN111913921A (en) * | 2019-05-08 | 2020-11-10 | 中国移动通信集团福建有限公司 | User behavior analysis method and device, equipment and storage medium |
| WO2020238385A1 (en) * | 2019-05-31 | 2020-12-03 | 华为技术有限公司 | Apparatus service control method, cloud server, and smart home system |
| CN110456647B (en) * | 2019-07-02 | 2020-11-27 | 珠海格力电器股份有限公司 | Intelligent household control method and intelligent household control device |
| CN110456647A (en) * | 2019-07-02 | 2019-11-15 | 珠海格力电器股份有限公司 | Intelligent household control method and intelligent household control device |
| CN110377829A (en) * | 2019-07-24 | 2019-10-25 | 中国工商银行股份有限公司 | Function recommended method and device applied to electronic equipment |
| CN110598916A (en) * | 2019-08-23 | 2019-12-20 | 宁波智轩物联网科技有限公司 | Method and system for constructing user behavior model |
| CN110471301A (en) * | 2019-08-23 | 2019-11-19 | 宁波智轩物联网科技有限公司 | A kind of smart home service recommendation system and method based on user behavior |
| CN110750653A (en) * | 2019-10-22 | 2020-02-04 | 中国工商银行股份有限公司 | Information processing method, information processing apparatus, electronic device, and medium |
| CN110989370A (en) * | 2019-11-06 | 2020-04-10 | 珠海格力电器股份有限公司 | Intelligent household interaction control method and system |
| CN111000492A (en) * | 2019-11-08 | 2020-04-14 | 尚科宁家(中国)科技有限公司 | Intelligent sweeper behavior decision method based on knowledge graph and intelligent sweeper |
| CN111306803A (en) * | 2020-03-01 | 2020-06-19 | 苏州淘喜网络科技有限公司 | Water source supply control system and method based on big data |
| CN111736481A (en) * | 2020-07-14 | 2020-10-02 | 北京自如信息科技有限公司 | Model training and intelligent home control method and device based on user behavior characteristics |
| CN112668384A (en) * | 2020-08-07 | 2021-04-16 | 深圳市唯特视科技有限公司 | Knowledge graph construction method and system, electronic equipment and storage medium |
| CN112668384B (en) * | 2020-08-07 | 2024-05-31 | 深圳市唯特视科技有限公司 | Knowledge graph construction method, system, electronic equipment and storage medium |
| CN112084376A (en) * | 2020-09-04 | 2020-12-15 | 北京明略昭辉科技有限公司 | Map knowledge based recommendation method and system and electronic device |
| CN112286150A (en) * | 2020-10-15 | 2021-01-29 | 珠海格力电器股份有限公司 | Intelligent household equipment management method, device and system and storage medium |
| CN112286150B (en) * | 2020-10-15 | 2021-09-14 | 珠海格力电器股份有限公司 | Intelligent household equipment management method, device and system and storage medium |
| CN112380334A (en) * | 2020-12-04 | 2021-02-19 | 三星电子（中国）研发中心 | Intelligent interaction method and device and intelligent equipment |
| CN112380334B (en) * | 2020-12-04 | 2023-03-24 | 三星电子（中国）研发中心 | Intelligent interaction method and device and intelligent equipment |
| CN112749625A (en) * | 2020-12-10 | 2021-05-04 | 深圳市优必选科技股份有限公司 | Time sequence behavior detection method, time sequence behavior detection device and terminal equipment |
| CN112749625B (en) * | 2020-12-10 | 2023-12-15 | 深圳市优必选科技股份有限公司 | Time sequence behavior detection method, time sequence behavior detection device and terminal equipment |
| CN112587932A (en) * | 2020-12-30 | 2021-04-02 | 网易（杭州）网络有限公司 | Game plug-in detection method and device, electronic equipment and storage medium |
| CN112782988A (en) * | 2020-12-30 | 2021-05-11 | 深圳市微网力合信息技术有限公司 | Control method of intelligent household curtain based on Internet of things |
| EP4030350A1 (en) * | 2021-01-18 | 2022-07-20 | Siemens Aktiengesellschaft | Neuromorphic hardware and method for storing and/or processing a knowledge graph |
| EP4030349A1 (en) * | 2021-01-18 | 2022-07-20 | Siemens Aktiengesellschaft | Neuromorphic hardware for processing a knowledge graph represented by observed triple statements and method for training a learning component |
| CN112734132A (en) * | 2021-01-22 | 2021-04-30 | 珠海格力电器股份有限公司 | Equipment recommendation method and device, electronic equipment and storage medium |
| CN113485144A (en) * | 2021-07-27 | 2021-10-08 | 广州市威士丹利智能科技有限公司 | Intelligent household control method and system based on Internet of things |
| CN113485144B (en) * | 2021-07-27 | 2023-11-03 | 广州市威士丹利智能科技有限公司 | Intelligent home control method and system based on Internet of things |
| CN113848738B (en) * | 2021-09-22 | 2024-05-03 | 深圳市欧瑞博科技股份有限公司 | Control method and device of intelligent equipment |
| CN113848738A (en) * | 2021-09-22 | 2021-12-28 | 深圳市欧瑞博科技股份有限公司 | Control method and device of intelligent equipment |
| CN113934930A (en) * | 2021-10-08 | 2022-01-14 | 宜兴市旭航电子有限公司 | User preference prediction system and method based on probability and hidden Markov model |
| CN113885706A (en) * | 2021-10-19 | 2022-01-04 | 清华大学 | Interaction control method, device and system |
| CN114114949A (en) * | 2022-01-12 | 2022-03-01 | 慕思健康睡眠股份有限公司 | Intelligent home control system and method based on big data analysis |
| WO2023159821A1 (en) * | 2022-02-23 | 2023-08-31 | 青岛海尔科技有限公司 | Method and device for determining operational behavior, storage medium, and electronic device |
| CN115542755A (en) * | 2022-05-31 | 2022-12-30 | 青岛海尔智能家电科技有限公司 | Method and device for predicting equipment control command, electronic equipment and storage medium |
| CN115268282A (en) * | 2022-06-29 | 2022-11-01 | 青岛海尔科技有限公司 | Control method and device of household appliance, storage medium and electronic device |
| CN118134539A (en) * | 2024-05-06 | 2024-06-04 | 山东传奇新力科技有限公司 | User behavior prediction method based on intelligent kitchen multi-source data fusion |
| CN118134539B (en) * | 2024-05-06 | 2024-07-19 | 山东传奇新力科技有限公司 | User behavior prediction method based on intelligent kitchen multi-source data fusion |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107665230B (en) | 2021-06-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107665230A (en) | Training method and device for the users' behavior model of Intelligent housing | |
| Wu et al. | Evolving RBF neural networks for rainfall prediction using hybrid particle swarm optimization and genetic algorithm | |
| CN111625361B (en) | Joint learning framework based on cooperation of cloud server and IoT (Internet of things) equipment | |
| CN112382082A (en) | Method and system for predicting traffic running state in congested area | |
| CN110851662B (en) | Heterogeneous information network link prediction method based on meta-path | |
| CN105447510B (en) | Fluctuating wind speed prediction technique based on artificial bee colony optimization LSSVM | |
| CN110866528A (en) | Model training method, energy consumption use efficiency prediction method, device and medium | |
| Goh et al. | A multimodal approach to chaotic renewable energy prediction using meteorological and historical information | |
| Xing et al. | Research of a novel short-term wind forecasting system based on multi-objective Aquila optimizer for point and interval forecast | |
| CN111414681B (en) | Building evacuation simulation method and system based on shared deep reinforcement learning | |
| CN109215344B (en) | Method and system for urban road short-time traffic flow prediction | |
| CN110826784B (en) | Method and device for predicting energy use efficiency, storage medium and terminal equipment | |
| CN113780002A (en) | Knowledge reasoning method and device based on graph representation learning and deep reinforcement learning | |
| CN111416797A (en) | Intrusion detection method for optimizing regularization extreme learning machine by improving longicorn herd algorithm | |
| CN110781595A (en) | Energy use efficiency PUE prediction method, device, terminal and medium | |
| CN114572229B (en) | Vehicle speed prediction method, device, medium and equipment based on graph neural network | |
| CN110414718A (en) | A kind of distribution network reliability index optimization method under deep learning | |
| CN113784410A (en) | Heterogeneous wireless network vertical switching method based on reinforcement learning TD3 algorithm | |
| CN116722545B (en) | Photovoltaic power generation prediction method based on multi-source data and related equipment | |
| CN107453921A (en) | Smart city system artificial intelligence evaluation method based on nonlinear neural network | |
| CN114861917B (en) | Knowledge graph reasoning model, system and reasoning method for Bayesian small sample learning | |
| CN108733921A (en) | Coiling hot point of transformer temperature fluctuation range prediction technique based on Fuzzy Information Granulation | |
| CN114065646A (en) | Energy consumption prediction method based on hybrid optimization algorithm, cloud computing platform and system | |
| CN114841199A (en) | Power distribution network fault diagnosis method, device, equipment and readable storage medium | |
| CN115423162A (en) | Traffic flow prediction method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |