CN109890002A - A kind of method and device of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference - Google Patents
A kind of method and device of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference Download PDFInfo
- Publication number
- CN109890002A CN109890002A CN201910007843.7A CN201910007843A CN109890002A CN 109890002 A CN109890002 A CN 109890002A CN 201910007843 A CN201910007843 A CN 201910007843A CN 109890002 A CN109890002 A CN 109890002A
- Authority
- CN
- China
- Prior art keywords
- node
- environmental information
- danger level
- information
- environment
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
- Alarm Systems (AREA)
Abstract
The invention discloses a kind of method and device of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference, method includes: the prediction environmental information of environment where determining present node based on history environment information and predetermined time sequence prediction algorithm；Determine the probability-distribution function of the corresponding each moment environmental information obtained of history environment information；According to current context information and probability-distribution function determination deviation environmental information；The danger level of environment where determining present node according to current context information, prediction environmental information and deviation environmental information.Radio node of the invention has the ability that perception and reasoning are carried out for the appearance of certain anomalous event, treatment process is executed by each node itself, treatment effeciency is higher, calculated result is returned without waiting for master controller, and the mode of node itself processing can substantially reduce the message number being responsible in network, lifting system performance.
Description
Technical field
The present invention relates to communication fields, method more particularly to a kind of prediction Risk Degree of Maneuvering Environment of knowledge-based inference and
Device.
Background technique
Currently, the Internet of Things monitoring environment of wireless sensor network has comprising several for certain anomalous event (such as fire behavior)
The radio node (i.e. sensor) that is perceived of appearance, the knowledge-based inference system based on wireless sensor network of mainstream is mostly
Integrated system, General controller intensively converge and handle the environment of each sensor acquisition and complete knowledge reasoning, be not only difficult to protect
Knowledge reasoning efficiency is demonstrate,proved, and causes the message number loaded in network excessive；In addition, current dominant systems heavy dependence I type
Fuzzy logic model because the model be applied to dynamic environment or when fuzzy logic ordination building include local knowledge
When caused uncertainty, there are obvious flexibilities to lack when based on fuzzy inference rule expression knowledge for the model
It falls into.
Summary of the invention
The present invention provides a kind of method and device of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference, to solve existing skill
The following problem of art: existing knowledge-based inference system intensively converges and handles the environment of each sensor acquisition by General controller and complete
At knowledge reasoning, it is not only difficult to ensure knowledge reasoning efficiency, but also causes the message number loaded in network excessive.
In order to solve the above technical problems, on the one hand, the present invention provides a kind of prediction Risk Degree of Maneuvering Environment of knowledge-based inference
Method, comprising: the prediction of environment where present node is determined based on history environment information and predetermined time sequence prediction algorithm
Environmental information；Determine the probability-distribution function of the corresponding each moment environmental information obtained of the history environment information；Root
According to current context information and the probability-distribution function determination deviation environmental information；According to the current context information, described pre-
It surveys environmental information and the deviation environmental information determines the danger level of present node place environment.
Optionally, it is determined according to the current context information, the environmental information of the prediction and the deviation environmental information
After the danger level of environment where present node, further includes: detect whether the danger level is more than predetermined danger level threshold value；?
In the case that the danger level is more than the predetermined danger level threshold value, the centromere of the predetermined class cluster where Xiang Suoshu present node
Point sends the danger level and/or alarm information.
Optionally, the central node of the predetermined class cluster where Xiang Suoshu present node sends the danger level and/or alarm
After message, further includes: the central node sends the danger level of each node in the predetermined class cluster to General controller.
Optionally, according to current context information and the probability-distribution function determination deviation environmental information, comprising: according to institute
State the expectation environmental information that probability-distribution function determines current point in time；The expectation environmental information and the current environment are believed
The absolute value of breath difference is determined as the deviation environmental information.
Optionally, it is determined according to current context information, the environmental information of the prediction and the deviation environmental information current
After the danger level of environment where node, further includes: the danger level of itself is sent to adjacent node by the present node.
Optionally, the method also includes: the danger of each node in the predetermined class cluster is counted according to predetermined time interval
Degree；Determine that the maximum node of danger level is new central node in all danger levels.
Optionally, the method also includes: each nodes is arranged select probability according to the danger level of itself, wherein described
Select probability indicates that itself becomes the probability of central node；Determine that the maximum node of select probability is new in all select probabilities
Central node.
On the other hand, the present invention also provides a kind of devices of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference, comprising: first
Determining module, the prediction for environment where determining present node based on history environment information and predetermined time sequence prediction algorithm
Environmental information；Second determining module, for determining the corresponding each moment environmental information obtained of the history environment information
Probability-distribution function；Third determining module, for according to current context information and the probability-distribution function determination deviation ring
Border information；Execution module, for true according to the current context information, the prediction environmental information and the deviation environmental information
The danger level of environment where settled front nodal point.
Optionally, further includes: detection module, for detecting whether the danger level is more than predetermined danger level threshold value；It sends
Module, for making a reservation for where Xiang Suoshu present node in the case where the danger level is more than the predetermined danger level threshold value
The central node of class cluster sends the danger level and/or alarm information.
Optionally, the third determining module, is specifically used for: determining current point in time according to the probability-distribution function
It is expected that environmental information；The absolute value of the expectation environmental information and the current context information difference is determined as the offset loop
Border information.
The radio node of the embodiment of the present invention has the ability that perception and reasoning are carried out for the appearance of certain anomalous event,
The danger of environment where determining present node jointly according to current context information, prediction environmental information and deviation environmental information
Degree, treatment process are executed by each node itself, and treatment effeciency is higher, return to calculated result without waiting for total General controller, and save
The mode of point itself processing can substantially reduce the message number being responsible in network, lifting system performance.
Detailed description of the invention
Fig. 1 is the flow chart of the method for the prediction Risk Degree of Maneuvering Environment of knowledge-based inference in first embodiment of the invention；
Fig. 2 is that a kind of structure of the device of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference in second embodiment of the invention is shown
It is intended to；
Fig. 3 is another structure of the device of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference in second embodiment of the invention
Schematic diagram；
Fig. 4 is the system architecture schematic diagram of wireless sensor network in third embodiment of the invention.
Specific embodiment
In order to solve the problems, such as the as follows of the prior art: existing instruction inference system is intensively converged and is handled by General controller
The environment of each sensor acquisition and complete knowledge reasoning, be not only difficult to ensure knowledge reasoning efficiency, but also cause to load in network
Message number it is excessive；The present invention provides a kind of method and devices of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference, below
In conjunction with attached drawing and embodiment, the present invention will be described in further detail.It should be appreciated that specific embodiment described herein
It is only used to explain the present invention, does not limit the present invention.
First embodiment of the invention provides a kind of method of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference, this method
Process is as shown in Figure 1, include step S101 to S104:
S101, the prediction of environment where determining present node based on history environment information and predetermined time sequence prediction algorithm
Environmental information；
S102 determines the probability-distribution function of the corresponding each moment environmental information obtained of history environment information；
S103, according to current context information and probability-distribution function determination deviation environmental information；
S104, environment where determining present node according to current context information, prediction environmental information and deviation environmental information
Danger level.
The radio node of the embodiment of the present invention has the ability that perception and reasoning are carried out for the appearance of certain anomalous event,
The danger of environment where determining present node jointly according to current context information, prediction environmental information and deviation environmental information
Degree, treatment process are executed by each node itself, and treatment effeciency is higher, return to calculated result without waiting for total General controller, and save
The mode of point itself processing can substantially reduce the message number being responsible in network, lifting system performance.
The environment where determining present node according to the environmental information of current context information, prediction and deviation environmental information
Danger level after, it is also necessary to detect whether danger level is more than predetermined danger level threshold value；If danger level is more than predetermined danger level
Threshold value then illustrates that current environment there are biggish dangerous hidden danger, needs the central node to the predetermined class cluster where present node
Send danger level and/or alarm information；If danger level is not above predetermined danger level threshold value, can not be on central node
Danger level and/or alarm information are reported, the danger level of itself is only sent to adjacent node by present node, after adjacent node
The continuous common recognition that can reach dangerous situation with it.
The central node of the predetermined class cluster where present node to present node send danger level and/or alarm information it
Afterwards, central node can periodically send the danger level of each node in predetermined class cluster to General controller.Pass through the center of predetermined class cluster
The mode that node is interacted with total General controller can not only allow total General controller to know the danger level of each node, but also greatly reduce
The loading condition of total General controller, improves system performance.
It, can be with during according to current context information and probability-distribution function determination deviation environmental information when realization
First determine the expectation environmental information of current point in time according to probability-distribution function, then by desired environmental information and current context information
It is poor to make, and the absolute value of the difference is determined as deviation environmental information.
The central node of each predetermined class cluster can be with real-time change in the embodiment of the present invention.For example, according to predetermined
Time interval counts the danger level of each node in predetermined class cluster, determines that the maximum node of danger level is new in all danger levels
Central node.For another example select probability is arranged according to the danger level of itself in each node, wherein select probability indicates itself
As the probability of central node, determine that the maximum node of select probability is new central node in all select probabilities.This hair
Bright embodiment can allow the node for being currently at most unsafe condition and General controller to be handed in such a way that danger level adjusts central node
Mutually, more attention of General controller can be obtained.
Second embodiment of the invention provides a kind of device of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference, which can
To be arranged in each node including central node of predetermined class cluster, the structural representation of the device is as shown in Figure 2, comprising:
First determining module 10, for determining present node based on history environment information and predetermined time sequence prediction algorithm
The prediction environmental information of place environment；Second determining module 20 is coupled with the first determining module 10, for determining that history environment is believed
Cease the probability-distribution function of corresponding environmental information obtained of each moment；Third determining module 30, with the second determining module
20 couplings, for according to current context information and probability-distribution function determination deviation environmental information；Execution module 40, it is true with third
Cover half block 30 couples, for determining present node place according to current context information, prediction environmental information and deviation environmental information
The danger level of environment.
The radio node of the embodiment of the present invention has the ability that perception and reasoning are carried out for the appearance of certain anomalous event,
The danger of environment where determining present node jointly according to current context information, prediction environmental information and deviation environmental information
Degree, treatment process are executed by each node itself, and treatment effeciency is higher, return to calculated result without waiting for total General controller, and save
The mode of point itself processing can substantially reduce the message number being responsible in network, lifting system performance.
Above-mentioned apparatus can also be as shown in Figure 3, comprising: and detection module 50 is coupled with execution module 40 and sending module 60,
For detecting whether danger level is more than predetermined danger level threshold value；Sending module 60, for being more than predetermined danger level threshold in danger level
In the case where value, danger level and/or alarm information are sent to the central node of the predetermined class cluster where present node.
Since the central node of the embodiment of the present invention can convert, the sending module 60 of each node all is also
It can be used for sending the danger level of each node in predetermined class cluster to General controller.Pass through the central node and master control of predetermined class cluster
The mode of device interaction, can not only allow total General controller to know the danger level of each node, but also greatly reduce the negative of total General controller
Situation is carried, system performance is improved.
The danger level of itself can also be sent to adjacent node by sending module 60.Present node sends out the danger level of itself
It send to adjacent node, so as to the subsequent common recognition that can reach dangerous situation with it of adjacent node.
When specific implementation, sending function can also be arranged respectively to the function of different modules, and be not all disposed within
In sending module 60, those skilled in the art can be arranged according to the actual situation, and details are not described herein again.
Above-mentioned third determining module, is specifically used for: determining that the expectation environment of current point in time is believed according to probability-distribution function
Breath；The absolute value of desired environmental information and current context information difference is determined as deviation environmental information.
Above-mentioned apparatus can also include adjustment module, to schedule in interval stats predetermined class cluster each node danger
Dangerous degree determines that the maximum node of danger level is new central node in all danger levels；Alternatively, being used for according to each node certainly
The select probability of the danger level of body setting determines that the maximum node of select probability is new centromere in all select probabilities
Point, wherein select probability indicates that itself becomes the probability of central node.The embodiment of the present invention adjusts central node by danger level
Mode, the node for being currently at most unsafe condition can be allowed to interact with General controller, more attention of General controller can be obtained.
The shortcomings that in view of the prior art, third embodiment of the invention are designed to provide under wireless sensor network environment
Distributed knowledge inference technology (process predicted by data) realizes that the distribution based on local environment knowledge is abnormal
Event (such as fire behavior) reasoning can significantly reduce wireless sensor network under the premise of improving reasoning process instantaneity, accuracy
Middle Message Payload amount.In order to achieve the above objects and other related objects, the present invention provides the first and is based on On Local Fuzzy logic sum
The distributed knowledge inference schemes of Knowledge driving cluster, the i.e. method of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference and application should
The system of method.The system includes several nodes (sensor), and each node observes same anomalous event (by taking fire behavior as an example), is led to
The environment of accepting information data reasoning anomalous event is crossed whether to occur.The anomalous event that the present invention obtains each node reasoning is sent out
A possibility that giving birth to degree, referred to as danger level.
The system globe area current context information of certain node, the environmental information and deviation environmental information being predicted, therefore,
Each node is required to comprehensive consideration: (1) current context information and being predicted between environmental information with the presence or absence of obvious deviation；
(2) current context information and the departure degree for arriving the current position node statistics distribution pattern.The former using short-term knowledge come into
Row reasoning, the latter then make inferences the statistical law of node using long-term knowledge.The fusion of above-mentioned short-term knowledge and long-term knowledge,
The reasoning results more complicated and fine for certain event can be generated.
Framework locating for the system is illustrated as shown in figure 4, the present embodiment is with an one of class of wireless sensor network
It is illustrated for cluster and the interaction of total General controller, realizes that the process of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference includes following
Key step:
Step 1, localize environment information prediction.
Based on history environment information, short-term upcoming environmental information is predicted using time series forecasting algorithm (i.e.
The environmental information being predicted).
Step 2, probability-distribution function incremental learning.
Based on history environment information, incrementally learn the probability distribution letter of passing environmental information obtained of each moment
Number.The probability-distribution function of certain node environmental information can measure the deviation of current context information and passing environmental information, in turn
Generate deviation environmental information.
Step 3, danger level reasoning.
Based on II type fuzzy logic model, the output of above-mentioned localization environment information prediction process (step 1) and general is merged
The output of rate distribution function incremental learning process (step 2), target are to generate danger level for each node.The danger of certain node
Deviation, pre- of the degree based on the node current context information, current context information and desired value (being drawn from probability-distribution function)
The environmental information of survey provides the local reasoning for monitored anomalous event.When danger level is more than some preset threshold
It waits, which will confirm the generation of certain anomalous event, then alert to its class cluster central node.
Step 4, Knowledge driving node automatic cluster.
Knowledge driving node clustering process is completed based on neighbor node, the process is for enhancing local reasoning effect and reduction
Network environment load.The danger level numerical value that certain node generates can only be delivered to the neighbor node of the node, to further enhance
The local knowledge of its neighbor node, the automatic cluster process of this danger level transmittance process excitation node.Derived from the process
Class cluster includes the node for holding similar " viewpoint (i.e. local environmental information the reasoning results) " for certain anomalous event.The present invention executes
More wheel Knowledge driving node automatic cluster processes, after every wheel cluster process terminates, each class cluster central node will be after polymerization
Local knowledge (including the knowledge reasoning result of itself) is sent to General controller.
The above method is described in detail below with reference to two specific examples.
In the first embodiment of the invention, in step 1, used linear session prediction algorithm is
LevinsonDurbin algorithm.To Mr. Yu's node, which generates the history environment value of information weighted sum of the node pre-
The environmental information of survey.Since Internet of Things monitors environment for the demand of signal processing instantaneity, so needing in the shortest possible time
Prediction result is generated, and linear session sequence prediction algorithm selected by the present invention can control time overhead linear complicated
Degree.
In the first embodiment of the invention, in step 2, used Incremental Learning Algorithm is Kernel Density
Estimator algorithm.The algorithm can generate the implicit distribution for describing passing each moment environmental information of given node.Pass through
To the incremental learning of probability-distribution function, the node can differentiate current context information whether with up to the present environmental information
There are significant deviations for desired value, that is, differentiate whether instantaneous current context information has not met the intrinsic statistical distribution of the node
Mode (i.e. probability-distribution function) --- the difference between this current context information and environmental information desired value is referred to as deviation
Environmental information.
In the first embodiment of the invention, in step 3, believed by fusion current context information, the environment being predicted
Breath and deviation environmental information complete the reasoning to certain anomalous event, and generating danger level, (danger level numerical value is higher, indicates danger etc.
Grade is higher).The fusion process is based on fuzzy inference rule, and the present invention, which designs, realizes II type fuzzy logic model.For fire behavior report
Alert application scenarios, under the guidance of domain expertise, the fuzzy inference rule that the present invention designs is (as shown in table 1) to use three kinds
Logical term: low (numerical value close to 0), in (numerical value close to 0.5), high (numerical value close to 1).Wherein, each term is corresponding
One numerical intervals；Current context information, the environmental information being predicted and deviation environmental information are by mapping function, by its numerical value
It is mapped to respective logic term.
Table 1
In the first embodiment of the invention, in step 4, Knowledge driving cluster process refers to, the danger based on node
Degree generates the process of inhomogeneity cluster to all nodes.Whether the member node in class cluster holds similar sight to anomalous event
Point.The present invention pre-defines multiple cluster periods, carries out a Knowledge driving cluster process in each cluster period.Each
In class cluster, a node is selected to class cluster central node, other nodes then become non-class cluster central node (i.e. class cluster member
Node).Class cluster central node is responsible for the danger level message of polymeric type bunch member node, is then communicated with General controller.Therefore,
Under this communication strategy of the invention, the message number to circulate in the entire network is substantially reduced, because being not necessarily to each node
All communicated with General controller.The basic thought of the selection course of class cluster central node is, if some node danger generated
Dangerous degree is higher than its neighbor node, then the node can be considered to become class cluster central node by emphasis.Once working as some node quilt
It is determined as class cluster central node, class bunch member node sends respective danger level message and gives such cluster central node.
Knowledge driving node automatic cluster process, is described as follows:
4-1. is initial random to select certain nodes as class cluster central node；
In the every wheel iteration of 4-2., dynamic changes class cluster central node；
4-3. is when iterative process quantity (or message number of exchange) reaches preset threshold, class cluster center selection course
It terminates.
Wherein, dynamic change class cluster central node process (step 4-2), is described as follows: for each node, selection course
Need series of iterations；In every wheel iteration, node sends danger level message and receives danger level to its neighbor node, from neighbor node
Message.It is first the select probability of itself configuration one " becoming class cluster central node ", the choosing before certain node starts selection course
Probability selection maximum value from global minima select probability and itself danger level numerical value is selected to obtain.For a certain node,
By comparing the select probability of its own select probability and other nodes, the node for carrying out select probability higher than itself select probability is made
For class cluster central node: there is the node of opposite high selection probability to send select probability message to its neighbour, and become in class cluster
Heart node；Node with smaller select probability equally sends select probability message to neighbours, if received from other sections
The message of point shows that the select probability of other nodes is higher than the select probability of this node, then assert itself for class bunch member node.
To sum up, distributed knowledge inference method of the invention and system are clustered by On Local Fuzzy logic and Knowledge driving,
Effective integration current context information, the environmental information being predicted and deviation environmental information can generate more multiple for certain event
Miscellaneous and fine knowledge reasoning as a result, and significantly reduce the message number of wireless sense network environment carrying, improve knowledge
The instantaneity and accuracy of reasoning.So the present invention effectively overcomes various shortcoming in the prior art and has high industrial benefit
With value.
As above, distributed knowledge inference method of the invention and system have the advantages that short-term knowledge and long-term
The fusion of knowledge can generate the reasoning results more complicated and fine for certain anomalous event.It is passed The present invention reduces wireless
The message number for feeling the carrying of net environment, solves the problems, such as that mainstream central knowledge inference system treatment effeciency is low, improves and know
Know the instantaneity and accuracy of reasoning；Class cluster central node is changed by dynamic, is able to extend the service life of wireless sensor network, into
And balance the energy consumption of knowledge reasoning process and message transmitting procedure.
In second of embodiment of the invention, in step 1, used time series forecasting algorithm is time prediction mould
Type (i.e. LSTM model), key step include: (1) setting LSTM model parameter (activation primitive, the full connection for receiving LSTM output
The activation primitive of artificial neural network, the rejection rate of every layer network node, error calculation mode, weight parameter iteration update side
Formula etc.)；(2) it using training set data training LSTM model, is predicted using trained model.
In second of embodiment of the invention, in step 2, used Incremental Learning Algorithm is Kernel Density
Estimator algorithm.The algorithm can generate the implicit distribution for describing passing each moment environmental information of given node.Pass through
To the incremental learning of probability-distribution function, the node can differentiate current context information whether with up to the present environmental information
There are significant deviations for desired value, that is, differentiate whether instantaneous current context information has not met the intrinsic statistical distribution of the node
Mode (i.e. probability-distribution function), wherein the difference between this current context information and environmental information desired value is referred to as inclined
Poor environmental information.
In second of embodiment of the invention, in step 3, believed by fusion current context information, the environment being predicted
Breath and deviation environmental information complete the reasoning to certain anomalous event, and generating danger level, (danger level numerical value is higher, indicates danger etc.
Grade is higher).The fusion process is based on fuzzy inference rule, and the present invention, which designs, realizes II type fuzzy logic model.For fire behavior report
Alert application scenarios, under the guidance of domain expertise, the fuzzy inference rule that the present invention designs is (as shown in table 1) to use three kinds
Logical term: low (numerical value close to 0), in (numerical value close to 0.5), high (numerical value close to 1).Wherein, each term is corresponding
One numerical intervals；Current context information, the environmental information being predicted and deviation environmental information are by mapping function, by its numerical value
It is mapped to respective logic term.
In second of embodiment of the invention, in step 4, Knowledge driving cluster process refers to, the danger based on node
Degree generates the process of inhomogeneity cluster to all nodes.Whether the member node in class cluster holds similar sight to anomalous event
Point.The present invention pre-defines multiple cluster periods, carries out a Knowledge driving cluster process in each cluster period.Each
In class cluster, a node is selected to class cluster central node, other nodes then become non-class cluster central node (i.e. class cluster member
Node).Class cluster central node is responsible for the danger level message of polymeric type bunch member node, is then communicated with General controller.Therefore,
Under this communication strategy of the invention, the message number to circulate in the entire network is substantially reduced, because being not necessarily to each node
All communicated with General controller.The basic thought of the selection course of class cluster central node is, if some node danger generated
Dangerous degree is higher than its neighbor node, then the node can be considered to become class cluster central node by emphasis.Once working as some node quilt
It is determined as class cluster central node, class bunch member node sends respective danger level message and gives such cluster central node.
Knowledge driving node automatic cluster process, is described as follows:
4-1. is initial random to select certain nodes as class cluster central node；
In the every wheel iteration of 4-2., dynamic changes class cluster central node；
4-3. is when iterative process quantity (or message number of exchange) reaches preset threshold, class cluster center selection course
It terminates.
Wherein, dynamic change class cluster central node process (step 4-2), is described as follows: for each node, selection course
Need series of iterations；In every wheel iteration, node sends danger level message and receives danger level to its neighbor node, from neighbor node
Message.For a certain node, by comparing the danger level of its own danger level and other nodes, to select danger level to be higher than
For the node of itself danger level as class cluster central node, the node with highest danger level becomes class cluster central node: having phase
Danger level message is sent to its neighbour to the node of high-risk, and becomes class cluster central node；Section with minor danger degree
Point equally sends danger level message to neighbours, if receiving the message from other nodes shows that the danger level of other nodes is high
In the danger level of this node, then assert itself for class bunch member node.In certain wheel iteration, certain node determines to be with danger level
It is no to become class cluster central node.
The radio node of the Internet of Things monitoring environment of the embodiment of the present invention can be (such as fiery for certain anomalous event to having
Feelings) appearance carry out perception and reasoning, each node be provided with reasoning localization knowledge (such as whether there is fire behavior and confidence level)
Perception and computing capability, in the case where no manual intervention to network environment transmit localization the reasoning results, i.e., will be local
Change the reasoning results to be transferred to neighbor node or be transferred to centralization information processing system (i.e. General controller).
The environmental information and deviation environmental information that the embodiment of the present invention merges current context information, is predicted, short-term knowledge
With the fusion of long-term knowledge, the reasoning results more complicated and fine for certain event are produced；Using II pattern fuzzy logic mould
Type avoids flexibility and accuracy defect that I type fuzzy logic model is generated when modeling uncertainty local knowledge；Using
Knowledge driving node automatic cluster strategy, and introduce select probability mechanism, network load can be reduced and promoted to uncertainty
Modeling ability；The range that can be applicable in is wide, can be generally applicable in smart machine Interconnection Environment.
Although for illustrative purposes, the preferred embodiment of the present invention has been disclosed, those skilled in the art will recognize
It is various improve, increase and replace be also it is possible, therefore, the scope of the present invention should be not limited to the above embodiments.
Claims (10)
1. a kind of method of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference characterized by comprising
The prediction environmental information of environment where determining present node based on history environment information and predetermined time sequence prediction algorithm；
Determine the probability-distribution function of the corresponding each moment environmental information obtained of the history environment information；
According to current context information and the probability-distribution function determination deviation environmental information；
Ring where determining present node according to the current context information, the prediction environmental information and the deviation environmental information
The danger level in border.
2. the method as described in claim 1, which is characterized in that believed according to the environment of the current context information, the prediction
After the danger level of environment where breath and the deviation environmental information determine present node, further includes:
Detect whether the danger level is more than predetermined danger level threshold value；
In the case where the danger level is more than the predetermined danger level threshold value, predetermined class cluster where Xiang Suoshu present node
Central node sends the danger level and/or alarm information.
3. method according to claim 2, which is characterized in that the central node of the predetermined class cluster where Xiang Suoshu present node
After sending the danger level and/or alarm information, further includes:
The central node sends the danger level of each node in the predetermined class cluster to General controller.
4. the method as described in claim 1, which is characterized in that determined according to current context information and the probability-distribution function
Deviation environmental information, comprising:
The expectation environmental information of current point in time is determined according to the probability-distribution function；
The absolute value of the expectation environmental information and the current context information difference is determined as the deviation environmental information.
5. the method as described in claim to 1, which is characterized in that according to current context information, the environmental information of the prediction
After the danger level of environment where determining present node with the deviation environmental information, further includes:
The danger level of itself is sent to adjacent node by the present node.
6. the method as described in any one of claims 1 to 5, which is characterized in that the method also includes:
To schedule in predetermined class cluster described in interval stats each node danger level；
Determine that the maximum node of danger level is new central node in all danger levels.
7. any one of claims 1 to 5 as described in method, which is characterized in that the method also includes:
Select probability is arranged according to the danger level of itself in each node, wherein the select probability indicates that itself becomes centromere
The probability of point；
Determine that the maximum node of select probability is new central node in all select probabilities.
8. a kind of device of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference characterized by comprising
First determining module, for ring where determining present node based on history environment information and predetermined time sequence prediction algorithm
The prediction environmental information in border；
Second determining module, for determining the probability of the corresponding each moment environmental information obtained of the history environment information
Distribution function；
Third determining module, for according to current context information and the probability-distribution function determination deviation environmental information；
Execution module, for being determined according to the current context information, the prediction environmental information and the deviation environmental information
The danger level of environment where present node.
9. device as claimed in claim 8, which is characterized in that further include:
Detection module, for detecting whether the danger level is more than predetermined danger level threshold value；
Sending module is used in the case where the danger level is more than the predetermined danger level threshold value, Xiang Suoshu present node institute
The central node of predetermined class cluster send the danger level and/or alarm information.
10. device as claimed in claim 8 or 9, which is characterized in that
The third determining module, is specifically used for: determining that the expectation environment of current point in time is believed according to the probability-distribution function
Breath；The absolute value of the expectation environmental information and the current context information difference is determined as the deviation environmental information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910007843.7A CN109890002B (en) | 2019-01-04 | 2019-01-04 | Method and device for predicting environmental risk based on knowledge reasoning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910007843.7A CN109890002B (en) | 2019-01-04 | 2019-01-04 | Method and device for predicting environmental risk based on knowledge reasoning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109890002A true CN109890002A (en) | 2019-06-14 |
| CN109890002B CN109890002B (en) | 2022-08-16 |
Family
ID=66925571
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910007843.7A Active CN109890002B (en) | 2019-01-04 | 2019-01-04 | Method and device for predicting environmental risk based on knowledge reasoning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109890002B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110766140A (en) * | 2019-10-22 | 2020-02-07 | 电子科技大学 | XGboost and LSTM-based multi-sensor real-time threat level classification method |
| CN111325347A (en) * | 2020-02-19 | 2020-06-23 | 山东大学 | Automatic danger early warning description generation method based on interpretable visual reasoning model |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104469875A (en) * | 2014-11-26 | 2015-03-25 | 北京邮电大学 | Prediction-based target tracking method and system in wireless sensor network |
| CN106302487A (en) * | 2016-08-22 | 2017-01-04 | 中国农业大学 | Agricultural Internet of Things data flow anomaly detects processing method and processing device in real time |
| CN106872657A (en) * | 2017-01-05 | 2017-06-20 | 河海大学 | A kind of multivariable water quality parameter time series data accident detection method |
| CN107071848A (en) * | 2017-05-27 | 2017-08-18 | 河南交通职业技术学院 | A kind of target tracking algorism based on cluster structured reduction energy consumption |
| CN107092582A (en) * | 2017-03-31 | 2017-08-25 | 江苏方天电力技术有限公司 | One kind is based on the posterior exceptional value on-line checking of residual error and method for evaluating confidence |
| CN108052528A (en) * | 2017-11-09 | 2018-05-18 | 华中科技大学 | A kind of storage device sequential classification method for early warning |
| CN108089962A (en) * | 2017-11-13 | 2018-05-29 | 北京奇艺世纪科技有限公司 | A kind of method for detecting abnormality, device and electronic equipment |
| WO2018098754A1 (en) * | 2016-11-30 | 2018-06-07 | 深圳天珑无线科技有限公司 | Cluster head selection method for distributed network, node and system |
| US20180189664A1 (en) * | 2015-06-26 | 2018-07-05 | National University Of Ireland, Galway | Data analysis and event detection method and system |
-
2019
- 2019-01-04 CN CN201910007843.7A patent/CN109890002B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104469875A (en) * | 2014-11-26 | 2015-03-25 | 北京邮电大学 | Prediction-based target tracking method and system in wireless sensor network |
| US20180189664A1 (en) * | 2015-06-26 | 2018-07-05 | National University Of Ireland, Galway | Data analysis and event detection method and system |
| CN106302487A (en) * | 2016-08-22 | 2017-01-04 | 中国农业大学 | Agricultural Internet of Things data flow anomaly detects processing method and processing device in real time |
| WO2018098754A1 (en) * | 2016-11-30 | 2018-06-07 | 深圳天珑无线科技有限公司 | Cluster head selection method for distributed network, node and system |
| CN106872657A (en) * | 2017-01-05 | 2017-06-20 | 河海大学 | A kind of multivariable water quality parameter time series data accident detection method |
| CN107092582A (en) * | 2017-03-31 | 2017-08-25 | 江苏方天电力技术有限公司 | One kind is based on the posterior exceptional value on-line checking of residual error and method for evaluating confidence |
| CN107071848A (en) * | 2017-05-27 | 2017-08-18 | 河南交通职业技术学院 | A kind of target tracking algorism based on cluster structured reduction energy consumption |
| CN108052528A (en) * | 2017-11-09 | 2018-05-18 | 华中科技大学 | A kind of storage device sequential classification method for early warning |
| CN108089962A (en) * | 2017-11-13 | 2018-05-29 | 北京奇艺世纪科技有限公司 | A kind of method for detecting abnormality, device and electronic equipment |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110766140A (en) * | 2019-10-22 | 2020-02-07 | 电子科技大学 | XGboost and LSTM-based multi-sensor real-time threat level classification method |
| CN111325347A (en) * | 2020-02-19 | 2020-06-23 | 山东大学 | Automatic danger early warning description generation method based on interpretable visual reasoning model |
| CN111325347B (en) * | 2020-02-19 | 2023-04-11 | 山东大学 | Automatic danger early warning description generation method based on interpretable visual reasoning model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109890002B (en) | 2022-08-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liang et al. | A deep reinforcement learning network for traffic light cycle control | |
| CN101807159B (en) | Self-adapting task scheduling method | |
| Balaji et al. | Multi-agent system in urban traffic signal control | |
| Chen et al. | A reinforcement learning-empowered feedback control system for industrial internet of things | |
| CN108197845A (en) | A kind of monitoring method of the transaction Indexes Abnormality based on deep learning model LSTM | |
| Zhang et al. | A novel job-shop scheduling strategy based on particle swarm optimization and neural network | |
| CN101448267A (en) | Wireless sensor network node coverage optimization method based on particle swarm algorithm | |
| CN109890002A (en) | A kind of method and device of the prediction Risk Degree of Maneuvering Environment of knowledge-based inference | |
| Vanderhaegen et al. | A multi-viewpoint system to support abductive reasoning | |
| CN113141012B (en) | Power grid power flow regulation and control decision reasoning method | |
| Liang et al. | A novel task optimal allocation approach based on contract net protocol for agent-oriented UUV swarm system modeling | |
| Zhou et al. | Solving large-scale 0-1 knapsack problem by the social-spider optimisation algorithm | |
| CN106817256A (en) | A kind of distributed system network resource operation management reliability method for improving | |
| CN116914751B (en) | Intelligent power distribution control system | |
| Zeng et al. | Trust-based multi-agent imitation learning for green edge computing in smart cities | |
| CN110443399A (en) | A kind of car accident air rescue intelligent dispatching method | |
| CN103686923A (en) | Target-tracking-oriented wireless sensor cluster energy managing method | |
| Hu et al. | Multi-objective deep reinforcement learning for emergency scheduling in a water distribution network | |
| Huong et al. | Effective partial charging scheme for minimizing the energy depletion and charging cost in wireless rechargeable sensor networks | |
| Udofia et al. | Dynamic traffic signal phase sequencing for an isolated intersection using ANFIS | |
| Ou et al. | Modeling wireless sensor network coverage based on random radius | |
| CN110245019A (en) | A kind of the thread concurrent method and device of Adaptable System resource | |
| Ng | Routing in wireless sensor network based on soft computing technique | |
| Wang et al. | A cooperative outage detection approach using an improved RBF neural network with genetic ABC algorithm | |
| Fang et al. | A new approach to time-cost-quality tradeoff problem for construction project |
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 |