WO2018120962A1 - 一种基于可靠性管理的不确定性消除情景感知系统及其工作方法 - Google Patents

一种基于可靠性管理的不确定性消除情景感知系统及其工作方法 Download PDF

Info

Publication number
WO2018120962A1
WO2018120962A1 PCT/CN2017/104792 CN2017104792W WO2018120962A1 WO 2018120962 A1 WO2018120962 A1 WO 2018120962A1 CN 2017104792 W CN2017104792 W CN 2017104792W WO 2018120962 A1 WO2018120962 A1 WO 2018120962A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
module
context
scenario
reliability
Prior art date
Application number
PCT/CN2017/104792
Other languages
English (en)
French (fr)
Inventor
许宏吉
潘玲玲
季名扬
孙君凤
周英明
房海腾
陈敏
Original Assignee
山东大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 山东大学 filed Critical 山东大学
Priority to AU2017386444A priority Critical patent/AU2017386444B2/en
Publication of WO2018120962A1 publication Critical patent/WO2018120962A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/02—Knowledge representation; Symbolic representation
    • G06N5/022—Knowledge engineering; Knowledge acquisition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/02—Knowledge representation; Symbolic representation
    • G06N5/022—Knowledge engineering; Knowledge acquisition
    • G06N5/025—Extracting rules from data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H04W4/025—Services making use of location information using location based information parameters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/80—Services using short range communication, e.g. near-field communication [NFC], radio-frequency identification [RFID] or low energy communication

Definitions

  • the invention proposes an uncertainty-based situational awareness system based on reliability management and a working method thereof, and belongs to the technical field of situational awareness.
  • the context-aware system is a human-centered computing system in which sensing devices can automatically sense scenarios and changes in contexts, providing users with services related to the current context.
  • the ideal situation information should be accurate, complete and consistent deterministic information, but in the actual situational awareness application, the same situation information can be obtained from different information sources in different ways. Due to the jitter of the acquisition time interval, the sensor itself exists. The accuracy and reliability issues, as well as a series of problems such as packet loss and delay during network transmission, may result in inaccuracies in the collected original context information (very different from real situation information) and incompleteness ( Uncertainty issues such as missing context information and inconsistency (the conflicting information collected by each sensor).
  • the original context information usually needs to be merged into high-level context information to be used by applications and devices.
  • the quality of the original context information plays a crucial role in the result of context information fusion reasoning. Therefore, the basis for effectively utilizing the original context information is to eliminate The uncertainty of the original situation information, and thus the accuracy and reliability of the context information fusion reasoning.
  • the present invention provides an uncertainty-based situational awareness system based on reliability management
  • the invention also provides a working method of the above system
  • the system treats the inaccurate original situation information as incomplete situation information, and then transforms it into the processing of incomplete situation information.
  • the system is in the horizontal aspect of incomplete original scene information processing.
  • vertical horizontal refers to different moments of the same sensor, vertical refers to the same moment of different sensors
  • algorithms neural network, evidence theory, maximum expectation algorithm, voting election, fuzzy set theory, etc.
  • the system uses the evidence theory method based on reliability management to effectively eliminate the inconsistency of the scene information in the processing of the inconsistent original scene information, and then obtain the high reliability scene information.
  • the method can combine the accuracy of each sensor and the situation information. Reliability two parameters are calculated to obtain reliability information;
  • the system can effectively eliminate the uncertainty of inaccuracy, incompleteness and inconsistency of the original situation information.
  • a uncertainty-based situational awareness system based on reliability management including scene information collection module, scene information processing module, knowledge base module, context information response module, context information application module, context information retrieval/subscription module, and context information correction Module and user feedback module;
  • the context information collection module, the context information processing module, and the knowledge base module are sequentially connected, the knowledge base module, the context information application module, the context information retrieval/subscription module, and the context information correction module
  • the knowledge base module, the context information response module, and the context information application module are sequentially connected, and the user feedback module is respectively connected to the knowledge base module and the context information application module;
  • the scenario information collection module is configured to periodically collect original scenario information by using multiple physical sensors, virtual sensors, and logic sensors, and send the collected original scenario information to the context information processing module, where the original scenario information is
  • the multi-source scenario information is a scenario information collected by multiple sensors; for example, the location information of the user may be collected by using multiple sensors such as Bluetooth, WIFI, infrared, and Zigbee;
  • the scenario information processing module is configured to process the original scenario information from the scenario information collection module;
  • the knowledge base module is configured to store user feedback information, context information fusion reasoning information, context information retrieval/subscription correction information, and scenario application information, and provide various scenario application information for the context information response module, and the context information is
  • the processing module provides various information required;
  • the scenario information application module is configured to display the comprehensive information from the context information response module and the user feedback module on the context aware system interface, and send the context information to the context information retrieval/subscription module;
  • the context information retrieval/subscription module according to the retrieval requirement of the scenario information application module for the context information, searching corresponding knowledge information in the knowledge base module, according to the subscription requirement of the context information application module for the context information, Sending relevant subscription information to the scenario information correction module;
  • the context information response module according to the scenario information application module needs for the context information, retrieves corresponding context information in the knowledge base module, and sends the scenario information required by the context information application module to the scenario Information application module;
  • the scenario information correction module is configured to correct the scenario information sent by the context information retrieval/subscription module, and send the corrected scenario information to the knowledge base module;
  • the user feedback module is configured to store the scenario information of the user in certain environments into the knowledge base module, and provide the scenario information application module with the required application scenario information.
  • the scenario information processing module includes a scenario information modeling unit, a multi-algorithm incompleteness elimination unit, an algorithm result inconsistency elimination unit, each source inconsistency elimination unit, a credibility management unit, and a reliable Sex management unit, context information fusion reasoning unit and adaptive management unit;
  • Each of the source inconsistency eliminating unit and the context information fusion inference unit are sequentially connected, and the adaptive management unit is respectively connected to the multi-algorithm incompleteness eliminating unit, the credibility management unit, and the respective letters.
  • a source inconsistency eliminating unit, the context information fusion inference unit, the source inconsistency eliminating unit is connected to the credibility management unit; and the scenario information fusion inference unit is connected to the knowledge base module, the knowledge base a module is connected to the adaptive management unit;
  • the scenario information modeling unit is configured to model the multi-source scenario information collected by the scenario information collection module according to the scenario information modeling manner in the knowledge base module, and the modeling mode is “scenario perception type + context awareness information”.
  • the context-aware type is the type of the context-aware information, such as the context-aware information "bedroom”, the sensing type is "location”, and the context-aware information is the original scene information collected by each sensor.
  • the context-awareness accuracy is a sensor-specific perceptual accuracy, such as “perceive type-user position”+“perception information-bedroom”+“perceptual accuracy-90%”, and the modeled situation information is sent to the multi-algorithm incomplete Elimination unit;
  • the multi-algorithm incompleteness elimination unit is responsible for simultaneously eliminating the incompleteness of the scene information by using multiple algorithms, obtaining the result of the incompleteness elimination of the multiple algorithms, and transmitting the result of the incompleteness elimination of the multiple algorithms to the inconsistency of the results of the algorithms. Eliminating the unit; the incompleteness means that the original scene information collected by a certain sensor is missing; the plurality of algorithms include a neural network, an evidence theory, a maximum expectation algorithm, a voting election, and a fuzzy set theory algorithm;
  • the algorithm inconsistency elimination unit is responsible for eliminating the inconsistency in the result of the incompleteness elimination of the multiple algorithms by using the voting election algorithm, obtaining complete situation information with high accuracy, and transmitting the complete situation information with higher accuracy to the institute.
  • the credibility management unit the inconsistency is inconsistent between the results of using various algorithms to eliminate the incompleteness of the context information.
  • the credibility management unit uses the received user feedback information to determine the correctness of the complete and accurate context information, and then calculates and stores the credibility of the corresponding source information of the complete and accurate information:
  • the user feedback information refers to scenario information that the user actively returns according to the environment; the correctness of the context information is correct scenario information determined based on the user feedback information;
  • the reliability management unit combines the accuracy of each sensor and the credibility of the source to perform reliability management on each source:
  • the sensor is a sensor used when collecting the source; the sensor accuracy is a static nominal value, and the reliability is a dynamic evaluation value; and the reliability of each source is sent to the source inconsistency eliminating unit;
  • the source inconsistency eliminating unit uses a reliability-based evidence theory algorithm to eliminate inconsistencies between the sources, and the inconsistency refers to inconsistencies between scene information collected by different sensors at the same time.
  • the infrared sensor collects the current location information of the user as “bedroom”, and the Zigbee sensor collects the location information of the user as “living room”, and sends the source inconsistency elimination result to the credibility management unit.
  • each source inconsistency elimination result is sent to the scenario information fusion inference unit;
  • the scenario information fusion inference unit using ontology reasoning, rule-based reasoning, evidence theory or Bayesian network inference method, inferring according to the collected source inconsistency elimination result and the historical information in the adaptive management unit
  • the advanced context information is used to store the advanced context information after the fusion inference into the knowledge base module;
  • the historical information refers to a rule system based rule engine and a rule set in the adaptive management unit;
  • the advanced context information refers to the inference fusion Scenario information obtained by the user or various devices;
  • the adaptive management unit is configured to provide user feedback information and context information retrieval/subscription correction information for the multi-algorithm incompleteness elimination unit, and provide user feedback information for the credibility management unit, which is the scenario
  • the information fusion inference unit provides a rule system based rule system and a rule set, and appropriately adjusts various parameters according to the current situation information, so that the adaptive ability of the context aware system is better.
  • the context information collection module comprises a plurality of physical sensors, virtual sensors and logic sensors.
  • a method for eliminating uncertainty based on reliability management using the above system includes the following steps:
  • the modeling mode is “scenario perception type + context awareness information + context perception accuracy ";
  • step S02' Detecting whether the scene information collected by each sensor has been completed, that is, whether the scene information is missing, if it is complete, proceeds to step S02', otherwise, proceeds to step S03;
  • step S05 Determining whether the incompleteness of the context information is within the replenishable range, if yes, proceeding to step S05', otherwise, proceeding to step S05; for example, the controllable range of the system setting incompleteness is 15%, when the imperfection rate is 10%, less than 15% of the controllable range of incompleteness, that is, within the controllable range;
  • the scene information from the information storage is simultaneously eliminated in the horizontal and vertical directions by a plurality of algorithms for the incompleteness of the scene information.
  • the horizontal refers to the different moments of the same sensor, and the vertical refers to the same moment of different sensors.
  • the various algorithms include: Neural network, maximum expectation algorithm, voting election, evidence theory, fuzzy set theory algorithm
  • step S08 Determining whether there is an inconsistency in the result of the incompleteness of the scenario information by a plurality of algorithms, if the results of the algorithms are consistent, then returning to step S02', otherwise, proceeding to step S08;
  • step S10 Determining whether there is an inconsistency in the scene information collected by each sensor, if yes, proceeding to step S10', otherwise, proceeding to step S10;
  • step S12 the correctness of the context information is determined based on the user feedback information, and the reliability of the source is 1 before the situation information inconsistency occurs;
  • the defined function Bel is a trust function
  • m(A) reflects the degree of joint support of proposition A for the two evidences corresponding to m 1 and m 2 , Indicates the degree of conflict of evidence.
  • 0 ⁇ K ⁇ 1 it is called incomplete conflict
  • the evidence combination rule satisfies the exchange law and combines Law, for the combination of multiple evidences, reusable (2) for multi-evidence reliability synthesis
  • step S11 Determining whether there is user feedback information at the current time, and if so, returning to step S10, otherwise, proceeding to step S11;
  • the reliability calculation is carried out in combination with the accuracy of the sensor itself: And using the reliability result of the source for the sensor inconsistency elimination in the step S11: making the judgment result true and reliable;
  • m is called the basic reliability distribution function on the recognition frame ;
  • m(A) is called the basic reliability value of subset A, and m(A) reflects the reliability of A itself, that is, the reliability of A.
  • the system has more complete functions, which can effectively eliminate the uncertainty of inaccuracy, incompleteness and inconsistency of the original situation information, and has strong practicability;
  • the method of the present invention obtains incomplete information values in horizontal and vertical directions by using various algorithms for the scenario information that has been modeled, and adopts a method of voting election for simple inconsistency elimination. , to obtain complete situational information with higher precision;
  • High reliability The reliability management method combined with the reliability of the source and the inherent accuracy of the sensor eliminates the inconsistency between the sensors and obtains more reliable situation information. At the same time, the user can also according to their own needs and environmental conditions. The change of the active feedback information makes the accuracy, reliability and adaptability of the context aware system significantly improved.
  • FIG. 1 is a block diagram of an uncertainty-based situational awareness system based on reliability management according to the present invention
  • FIG. 2 is a process flow diagram of a method for eliminating uncertainty based on reliability management according to the present invention
  • FIG. 3 is a simulation effect diagram of a prior scene-aware system
  • FIG. 4 is a simulation effect diagram of an embodiment regarding the overall performance of the system.
  • a uncertainty-aware situation-aware system based on reliability management includes a context information collection module, a context information processing module, a knowledge base module, a context information response module, a context information application module, and a context information retrieval/ a subscription module, a context information correction module, and a user feedback module;
  • the scenario information collection module, the context information processing module, and the knowledge base module are sequentially connected, and the knowledge base module, the context information application module, the context information retrieval/subscription module, and the context information correction module are connected end to end, the knowledge base module, the context information response module, and the scenario.
  • the information application modules are sequentially connected, and the user feedback modules are respectively connected to the knowledge base module and the context information application module;
  • the situation information collection module is responsible for periodically collecting original scene information through multiple physical sensors, virtual sensors and logic sensors, and transmitting the collected original scene information to the scene information processing module, the original scene information is multi-source scene information, and more
  • the source scenario information is a certain scenario information collected by multiple sensors; for example, the location information of the user may be collected by using multiple sensors such as Bluetooth, WIFI, infrared, and Zigbee;
  • Scenario information processing module responsible for processing original situation information from the scene information collection module
  • Knowledge Base Module responsible for storing user feedback information, context information fusion reasoning information, context information retrieval/subscription correction information, scenario application information, and providing various context application information for the context information response module to provide the context information processing module Various information;
  • the scenario information application module is responsible for displaying the comprehensive information from the context information response module and the user feedback module on the context aware system interface, and transmitting the context information to the context information retrieval/subscription module;
  • Scenario information retrieval/subscription module according to the context information application module for the retrieval of the context information, the corresponding context information is retrieved in the knowledge base module, and the relevant subscription information is sent to the context information according to the subscription requirement of the context information application module for the context information. Correction module;
  • the scenario information response module according to the scenario information application module needs the context information, retrieves the corresponding context information in the knowledge base module, and sends the scenario information required by the scenario information application module to the scenario information application module;
  • the scenario information correction module is responsible for correcting the scenario information sent by the context information retrieval/subscription module, and transmitting the corrected scenario information to the knowledge base module;
  • User feedback module responsible for storing the user's situation information in certain environments into the knowledge base module, and providing the scenario information application module with the required application scenario information.
  • the scenario information processing module includes a scenario information modeling unit, a multi-algorithm incompleteness elimination unit, an algorithm result inconsistency elimination unit, each source inconsistency elimination unit, a credibility management unit, a reliability management unit, and a context information fusion reasoning.
  • Scenario information collection module scenario information modeling unit, multi-algorithm incompleteness elimination unit, algorithm inconsistency elimination unit, credibility management unit, reliability management unit, source inconsistency elimination unit, context information fusion reasoning
  • the units are connected in sequence, and the adaptive management unit is respectively connected with a multi-algorithm incompleteness elimination unit, a credibility management unit, each source inconsistency elimination unit, a context information fusion inference unit, and each source inconsistency elimination unit connection credibility management.
  • the context information fusion inference unit is connected to the knowledge base module, and the knowledge base module is connected to the adaptive management unit;
  • the scenario information modeling unit is responsible for modeling the multi-source scenario information collected by the scenario information collection module according to the scenario information modeling mode in the knowledge base module, and the modeling mode is “scenario perception type + context perception information + context perception accuracy
  • the context-aware type is the type of context-aware information, such as the context-aware information “bedroom”, and its perception type is “location”.
  • the context-aware information is the original scene information collected by each sensor, and the scene-aware accuracy is the sensor's inherent perceptual accuracy. For example, "perceive type - user location” + "perception information - bedroom” + “perceptual accuracy - 90%” sends the modeled situation information to the multi-algorithm incomplete elimination unit;
  • Multi-algorithm incompleteness elimination unit responsible for eliminating the incompleteness of scene information by using multiple algorithms at the same time, obtaining the result of incompleteness elimination of multiple algorithms, and sending the result of incompleteness elimination of multiple algorithms to the inconsistency elimination unit of each algorithm result;
  • Completeness means that the original scene information collected by a sensor is missing;
  • various algorithms include neural network, evidence theory, maximum expectation algorithm, voting election, fuzzy set theory algorithm;
  • Algorithm inconsistency elimination unit responsible for using the voting election algorithm to eliminate inconsistencies in the incompleteness of multiple algorithms sexuality, get complete situation information with higher accuracy, and send complete situation information with higher accuracy to the credibility management unit; inconsistency is inconsistent between the results of using various algorithms to eliminate the incompleteness of situation information.
  • the credibility management unit uses the received user feedback information to determine the correctness of the complete and accurate context information, and then calculates and stores the credibility of the corresponding source information of the complete and accurate information:
  • the user feedback information refers to the scenario information that the user actively returns according to the environment; the correctness of the context information is the correct scenario information determined based on the user feedback information;
  • the Reliability Management Unit Combine the accuracy of each sensor and the credibility of the source to manage the reliability of each source:
  • the sensor is a sensor used when collecting the source; the sensor accuracy is a static nominal value, and the reliability is a dynamic evaluation value; the reliability of each source is sent to each source inconsistency eliminating unit;
  • Each source inconsistency elimination unit uses a reliability-based evidence theory algorithm to eliminate inconsistencies between sources.
  • Inconsistency refers to inconsistencies between scene information collected by different sensors at the same time, for example, The infrared sensor collects the current location information of the user as “bedroom”, and the Zigbee sensor collects the user's location information as “living room”, and sends the source inconsistency elimination result to the credibility management unit, and the sources are inconsistent.
  • the result of the sexual elimination is sent to the context information fusion inference unit;
  • Scenario information fusion reasoning unit using ontology reasoning, rule-based reasoning, evidence theory or Bayesian network inference method, inferring advanced scenarios based on the collected source inconsistency elimination results and historical information in the adaptive management unit Information, the advanced context information after fusion reasoning is stored in the knowledge base module; the historical information refers to the rule system-based rule engine and rule set in the adaptive management unit; the advanced context information refers to the available users after the inference fusion Or scenario information for various device applications;
  • Adaptive management unit responsible for providing user feedback information and correction information after context information retrieval/subscription for the multi-algorithm incompleteness elimination unit, providing user feedback information for the credibility management unit, and providing a rule-based system for the context information fusion inference unit The rule engine and rule set, and make appropriate adjustments to various parameters according to the current situation information, so that the adaptive ability of the context-aware system is better.
  • the context information collection module includes a plurality of physical sensors, virtual sensors, and logic sensors.
  • the reliability management-based uncertainty elimination scenario-aware system described in Embodiment 1 is based on the uncertainty management work method of reliability management, as shown in FIG. 2, taking a typical scenario of scenario-aware computing, a smart home, as an example.
  • WIFI, Bluetooth, Infrared and Zigbee are used to collect information about people's location.
  • the scene information obtained by WIFI, Bluetooth, Infrared and Zigbee are I WIFI , I Bluetooth , I Infrared , I Zigbee , including Proceed as follows:
  • the modeling mode is “scenario perception type + context awareness information + context perception accuracy ";
  • I WIFI "Sense Type - User Location” + “Perception Information - Bedroom” + “Perception Accuracy - 90%”
  • I Bluetooth "Perception Type - User Location” + “Perception Information - Bedroom” + “Perception Accuracy -92%”
  • I Zigbee "Sense Type - User Location” + “Perception Information - Bedroom” + “Perceived Accuracy - 94%”
  • I Infrared "Perception Type - User Location” + “Perception Information - Bedroom” + " Perceptual accuracy -96%”;
  • I WIFI "Sense Type - User Location” + “Perception Information - Living Room” + “Perception Accuracy - 90%”
  • I Bluetooth "Perception Type - User Location” + “Perception Information - Outdoor” + “Perception Accuracy -92%”
  • I Zigbee "Sense Type - User Location” + “Perception Information - Bedroom” + “Perceived Accuracy - 94%”
  • I Infrared "Perception Type - User Location” + “Perception Information - NaN” + " Perceptual accuracy -96%”;
  • I WIFI "Sense Type - User Location” + “Perception Information - Bedroom” + “Perception Accuracy - 90%”
  • I Bluetooth "Perception Type - User Location” + “Perception Information - Living Room” + “Perception Accuracy -92%”
  • I Zigbee "Sense Type - User Location” + “Perception Information - Bedroom” + “Perceived Accuracy - 94%”
  • I Infrared "Perception Type - User Location” + “Perception Information - Bedroom” + " Perceptual accuracy -96%”;
  • the time interval of each sensing device is fixed and the interval is appropriate, and the timeliness is high;
  • the current location information is displayed as "outdoor", which belongs to inaccurate context information.
  • the detected I WIFI context information is complete context information, and proceeds to step S02';
  • the scene information collected by other sensing devices outside the I WIFI is incomplete, and proceeds to step S03;
  • the user automatically adjusts the current location situation information to adjust the processing of the uncertainty situation information by the situation sensing system, so that the system is more stable and reliable;
  • the scene information will be incomplete based on various scenarios from the information storage in the horizontal direction (the same sensor at different times) and the vertical (the same sensor at the same time). Elimination of travel, such as neural networks, maximum expectation algorithms, voting elections, evidence theory, fuzzy set theory and other algorithms;
  • step S02' If there is user feedback information at the current moment, such as "current location is bedroom”, the missing location information is added to "bedroom”, and then complete situation information is obtained, and the process returns to step S02';
  • the results obtained by the algorithms are all "bedrooms", that is, the results of the algorithms are consistent, and the process returns to step S02';
  • the neural network elimination result is “bedroom”
  • the maximum expected algorithm elimination result is “living room”
  • the evidence theory elimination result is “bedroom”
  • the fuzzy set theory elimination result is The "living room” and the voting election elimination result are “bedrooms”, and the simple inconsistency elimination (voting election) of various algorithm results is obtained, and the complete situation information with higher quality is obtained, and the incomplete position information is finally completed, that is, t1 time I Zigbee location information is "living room”, t3 moment I Bluetooth location information is "living room”, I infrared position information is "bedroom”;
  • the position information collected by each sensor at time t2 is the same, and the reliability calculation is directly performed.
  • the received feedback information is used to judge the correctness of the multi-source context information, and then the reliability of the corresponding information source is evaluated: Dynamic calculation and storage of information sources, Note: The correctness of location information is judged based on feedback information);
  • step S11 Determining whether the current time user actively feedbacks his or her location information, and if so, returns to step S10, otherwise, proceeds to step S11;
  • the reliability result of the source is used to eliminate the inconsistency of each sensor in the step S11, so that the judgment result is true and reliable;
  • FIG. 3 is a simulation effect diagram of the existing situational awareness system

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Telephonic Communication Services (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Selective Calling Equipment (AREA)

Abstract

一种基于可信度管理的不确定性消除情景感知系统及其工作方法，该系统包括情景信息采集模块、情景信息处理模块、知识库模块、情景响应模块、情景信息应用模块、情景信息检索/订阅模块、情景信息校正模块以及用户反馈模块，其中情景处理模块中包括多算法不完备性消除单元、各算法结果不一致性消除单元、可信度管理单元、可靠性管理单元、各信源不一致性消除单元、情景信息融合推理单元以及自适应管理单元。能够得到更加可靠的情景信息，使该情景感知系统的精确性、可靠性、自适应性显著提高。

Description

一种基于可靠性管理的不确定性消除情景感知系统及其工作方法 技术领域
本发明提出了一种基于可靠性管理的不确定性消除情景感知系统及其工作方法，属于情景感知的技术领域。
背景技术
随着无线传感技术、人机交互技术和智能计算技术的日趋完善，情景感知技术得到了迅速的发展，进而情景感知系统得以融入人们的日常生活中。情景感知系统是以人为中心的计算系统，该计算系统中各传感设备可以自动感知情景以及情景的变化，向用户提供与当前情景相关的服务。
理想的情景信息应该是精确、完备和一致的确定性信息，但在实际的情景感知应用中，同一情景信息可以从不同的信息源通过不同的方式获取，由于采集时间间隔的抖动，传感器本身存在的精确度和可靠性问题，以及网络传输过程中的丢包、时延等一系列问题，可能导致采集到的原始情景信息存在不精确性(与真实情景信息相差很大)、不完备性(情景信息在某时刻有缺失)和不一致性(各传感器采集的情景信息存在冲突)等不确定性问题。
原始情景信息通常需要融合推理为高级情景信息才能被应用程序和设备利用，而原始情景信息的质量对情景信息融合推理的结果起着至关重要的作用，所以有效利用原始情景信息的基础就是消除原始情景信息的不确定性问题，进而提高情景信息融合推理的精确性和可靠性。
现有的情景感知系统往往只针对不确定性问题中的某一方面(不一致性)进行消除，没有考虑全面的不确定性消除方案；并且在情景信息不完备性消除方面往往采用单一消除算法，致使处理精确度低；在情景信息不一致性消除方面，多采用单一的信源可信度评估机制来消除原始情景信息不一致性，致使可靠性低。因此如何完善不确定性消除情景感知系统功能，并提高不完备性消除的精度以及不一致性消除的可靠性，使系统做出正确和可靠的决策，成为情景感知技术面临的挑战。
发明内容
针对现有技术的不足，本发明提供一种基于可靠性管理的不确定性消除情景感知系统；
本发明还提供了上述系统的工作方法；
本系统在不精确性原始情景信息的处理方面，将其看作不完备情景信息来处理，进而转化为对不完备情景信息的处理；本系统在不完备性原始情景信息处理方面，同时在横向和纵向上(横向是指同一传感器的不同时刻，纵向是指不同传感器的同一时刻)采用多种算法(神经网络、证据论、最大期望算法、投票选举、模糊集理论等算法)来消除原始情景信息不完备性，并通过对各算法结 果进行简单不一致性消除(采用投票选举算法)，得到精确度更高的完备情景信息；
本系统在不一致性原始情景信息的处理方面，采用基于可靠性管理的证据理论方法对情景信息的不一致性进行有效消除进而得到高可靠性情景信息，该方法通过结合各传感器精度和情景信息的可信度两个参数计算得到可靠性信息；
本系统能够有效地消除原始情景信息存在的不精确性、不完备性、不一致性等不确定性问题。
本发明的技术方案为：
一种基于可靠性管理的不确定性消除情景感知系统，包括情景信息采集模块、情景信息处理模块、知识库模块、情景信息响应模块、情景信息应用模块、情景信息检索/订阅模块、情景信息校正模块以及用户反馈模块；
所述情景信息采集模块、所述情景信息处理模块、所述知识库模块依次连接，所述知识库模块、所述情景信息应用模块、所述情景信息检索/订阅模块、所述情景信息校正模块依次首尾连接，所述知识库模块、所述情景信息响应模块、所述情景信息应用模块依次连接，所述用户反馈模块分别连接所述知识库模块、所述情景信息应用模块；
所述情景信息采集模块：负责通过多个物理传感器、虚拟传感器和逻辑传感器周期性地采集原始情景信息，并将采集到的原始情景信息发送至所述情景信息处理模块，所述原始情景信息为多源情景信息，所述多源情景信息为通过多个传感器采集到的某一情景信息；例如，可以通过蓝牙、WIFI、红外、Zigbee等多个传感器采集用户的位置信息；
所述情景信息处理模块：负责对来自所述情景信息采集模块的原始情景信息进行处理；
所述知识库模块：负责存储用户反馈信息、情景信息融合推理信息、情景信息检索/订阅校正信息、情景应用信息，同时为所述情景信息响应模块提供各种情景应用信息，为所述情景信息处理模块提供所需要的各种信息；
所述情景信息应用模块：负责将来自所述情景信息响应模块、所述用户反馈模块中的综合信息显示在情景感知系统界面上，并将情景信息发送给情景信息检索/订阅模块；
所述情景信息检索/订阅模块：根据所述情景信息应用模块对情景信息的检索需求，在所述知识库模块中检索相应的情景信息，根据所述情景信息应用模块对情景信息的订阅需求，将相关订阅信息发送到所述情景信息校正模块中；
所述情景信息响应模块：根据所述情景信息应用模块对情景信息的需求，在所述知识库模块中检索相应的情景信息，将所述情景信息应用模块所需的情景信息发送给所述情景信息应用模块；
所述情景信息校正模块：负责对所述情景信息检索/订阅模块发送来的情景信息进行校正，将校正后的情景信息发送到知识库模块；
所述用户反馈模块：负责将用户在某些环境下的情景信息存入所述知识库模块，并向所述情景信息应用模块提供所需应用情景信息。
根据本发明优选的，所述情景信息处理模块包括情景信息建模单元、多算法不完备性消除单元、各算法结果不一致性消除单元、各信源不一致性消除单元、可信度管理单元、可靠性管理单元、情景信息融合推理单元以及自适应管理单元；
所述情景信息采集模块、所述情景信息建模单元、所述多算法不完备性消除单元、所述各算法结果不一致性消除单元、所述可信度管理单元、所述可靠性管理单元、所述各信源不一致性消除单元、所述情景信息融合推理单元依次连接，所述自适应管理单元分别连接所述多算法不完备性消除单元、所述可信度管理单元、所述各信源不一致性消除单元、所述情景信息融合推理单元，所述各信源不一致性消除单元连接所述可信度管理单元；所述情景信息融合推理单元连接所述知识库模块，所述知识库模块连接所述自适应管理单元；
所述情景信息建模单元：负责将情景信息采集模块采集来的多源情景信息按照所述知识库模块中的情景信息建模方式进行建模，建模模式为“情景感知类型+情景感知信息+情景感知精度”，所述情景感知类型为所述情景感知信息的类型，如情景感知信息“卧室”，其感知类型为“位置”，所述情景感知信息为各传感器采集的原始情景信息，所述情景感知精度为传感器固有的感知精度，例如“感知类型-用户位置”+“感知信息-卧室”+“感知精度-90％”将建模好的情景信息发送到所述多算法不完备消除单元；
所述多算法不完备性消除单元：负责同时采用多种算法消除情景信息的不完备性，得到多算法不完备性消除结果，将多算法不完备性消除结果发送到所述各算法结果不一致性消除单元；所述不完备性是指某一传感器所采集的原始情景信息有缺失；所述多种算法包括神经网络、证据论、最大期望算法、投票选举、模糊集理论算法；
所述各算法不一致性消除单元：负责采用投票选举算法消除多算法不完备性消除结果中存在的不一致性，得到精确性较高的完备情景信息，将精确性较高的完备情景信息发送给所述可信度管理单元；所述不一致性为采用各种算法消除情景信息不完备性的结果之间存在不一致。
所述可信度管理单元：利用接收到的用户反馈信息，判断精确性较高的完备情景信息的正确性，继而计算并存储精确性较高的完备情景信息对应信源的可信度：
Figure PCTCN2017104792-appb-000001
所述用户反馈信息是指用户根据所处环境所主动返回的情景信息；情景信息的正确性是基于用户反馈信息来判断的正确的情景信息；
所述可靠性管理单元：结合各传感器精度，以及信源的可信度，对各信源进行可靠性管理：
Figure PCTCN2017104792-appb-000002
所述传感器为采集该信源时所用传感器；所述传感器精度为静态标称值，可信度为动态评估值；将各信源的可靠性发送给所述各信源不一致性消除单元；
所述各信源不一致性消除单元，采用基于可靠性的证据理论算法消除各信源间的不一致性，所述不一致性是指同一时刻通过不同的传感器所采集到的情景信息之间存在的不一致性，例如，红外传感器采集到用户当前的位置信息为“卧室”，而Zigbee传感器采集到用户的位置信息为“客厅”，将各信源不一致性消除结果发送到所述可信度管理单元，同时将各信源不一致性消除结果发送到情景信息融合推理单元；
所述情景信息融合推理单元：运用本体推理、基于规则的推理、证据论或贝叶斯网络推理方法，根据采集到的各信源不一致性消除结果和自适应管理单元中的历史信息，推理出高级情景信息，将融合推理后的高级情景信息存储到知识库模块中；所述历史信息是指自适应管理单元中基于规则系统的规则引擎和规则集；所述高级情景信息是指经过推理融合后得到的可供用户或各种设备应用的情景信息；
所述自适应管理单元：负责为所述多算法不完备性消除单元提供用户反馈信息和情景信息检索/订阅后的校正信息，为所述可信度管理单元提供用户反馈信息，为所述情景信息融合推理单元提供基于规则系统的规则引擎和规则集，并根据当前情景信息对各种参数做出适当调整，使情景感知系统的自适应能力更好。
根据发明优选的，所述情景信息采集模块包括多个物理传感器、虚拟传感器和逻辑传感器。
一种利用上述系统基于可靠性管理的不确定性消除工作方法，包括步骤如下：
S01：情景信息的收集和建模
收集多个物理传感器、虚拟传感器和逻辑传感器采集的情景信息，将情景信息按照知识库模块中的情景信息建模模式进行建模，建模模式为“情景感知类型+情景感知信息+情景感知精度”；
S02：检测各情景信息是否完备
检测已经建模好的每个传感器采集的情景信息是否完备，即情景信息有无缺失，如果完备，进入步骤S02’，否则，进入步骤S03；
S02’：信息存储
进行情景信息存储，以备步骤S06、步骤S09使用；
S03：计算不完备率
计算已建模好的每个传感器采集到的情景信息的不完备率：
Figure PCTCN2017104792-appb-000003
S04：判断不完备性是否在可控范围内
判断情景信息的不完备性是否在可补齐范围内，如果是，进入步骤S05’，否则，进入步骤S05；例如，系统设定不完备性的可控范围为15％，当不完备率为10％时，小于不完备性的可控范围15％，即在可控范围内；
S05：删除情景信息
直接将该情景信息删除；
S05’：判断是否存在用户反馈
检测当前时刻是否存在用户反馈，如果存在，进入步骤S06’，否则，进入步骤S06；
S06：多算法消除不完备性
将来自信息存储的各情景信息同时在横向和纵向上采用多种算法对情景信息不完备性进行消除，横向是指同一传感器的不同时刻，纵向是指不同传感器的同一时刻，多种算法包括：神经网络、最大期望算法、投票选举、证据论、模糊集理论算法；
S06’：直接补齐不完备性信息
根据用户反馈信息补齐不完备性情景信息，得到完备信息，返回步骤S02’；
S07：判断各算法结果是否存在不一致性
判断通过多种算法对情景信息不完备性消除的结果是否存在不一致性，若各算法结果一致，则返回步骤S02’，否则，进入步骤S08；
S08：各算法结果不一致性消除
采用投票选举策略对各算法结果进行简单不一致性消除，得到质量较高的完备情景信息；
S09：判断各传感器是否存在不一致性
判断各传感器采集到的情景信息是否存在不一致性，如果存在，进入步骤S10’，否则，进入步骤S10；
S10：计算可信度
直接进行可信度计算：
Figure PCTCN2017104792-appb-000004
进入步骤S12，情景信息的正确性是基于用户反馈信息来判断的，在未出现情景信息不一致性前，该信源的可信度为1；
如果m是一个基本信度分配函数，则
Figure PCTCN2017104792-appb-000005
B≠φ；A上所有子集总的信度Bel(A)的公式如式(Ⅰ)所示：
Figure PCTCN2017104792-appb-000006
所定义的函数Bel是一个信任函数；
设m1，m2是对应同一识别框架Θ上的两个基本信度分配函数，焦元分别为A1,A2,...,Ak和B1,B2,...,Bk，则两个信度函数的合成法则如式(Ⅱ)所示：
Figure PCTCN2017104792-appb-000007
m(A)反映了m1和m2对应的两个证据对命题A的联合支持程度，
Figure PCTCN2017104792-appb-000008
表示证据的冲突程度，当K＝0时，称为完全不冲突；当0＜K＜1时，称为非完全冲突；K＝1时，称为完全冲突，证据组合规则满足交换律和结合律，对于多个证据的组合可重复运用式(2)对多证据进行信度合成；
S10’：判断是否存在用户反馈
判断当前时刻是否存在用户反馈信息，如果是，返回步骤S10，否则，进入步骤S11；
S11：各传感器不一致性消除
采用证据理论方法进行不一致性消除；
步骤S12：计算可靠性
在可信度的基础上，结合传感器本身的精度进行可靠性计算：
Figure PCTCN2017104792-appb-000009
并将该信源的可靠性结果用于所述步骤S11各传感器不一致性消除：使判决结果真实可靠；
设Θ为识别框架，如果对于任何一个属于Θ的子集A满足
Figure PCTCN2017104792-appb-000010
则称m为识别框架Θ上的基本信度分配函数；
Figure PCTCN2017104792-appb-000011
m(A)称为子集A的基本信度值，m(A)反映了对A本身的信度大小，即A的可靠性。
本发明的有益效果为：
1、本系统功能更完善，能够有效地消除原始情景信息存在的不精确性、不完备性、不一致性等不确定性问题，实用性强；
2、高精确性：本发明所述方法通过对已经建模好的情景信息采用多种算法在横向和纵向上得到不完备信息值，并对各算法结果采用投票选举的方法进行简单不一致性消除，得到精度更高的完备情景信息；
3、高可靠性：利用信源可信度与传感器固有精度相结合的可靠性管理方法消除各传感器之间的不一致性，得到更加可靠的情景信息，同时用户也可以根据自身需求、环境状态的变化主动反馈信息，使得该情景感知系统的精确性、可靠性、自适应性显著提高。
附图说明
图1是本发明所述一种基于可靠性管理的不确定性消除情景感知系统的框图；
图2是本发明一种基于可靠性管理的不确定性消除方法的处理流程图；
图3是现有情景感知系统的仿真效果图；
图4是实施例关于本系统总体性能的仿真效果图。
具体实施方式
下面结合说明书附图和实施例对本发明作进一步限定，但不限于此。
实施例1
一种基于可靠性管理的不确定性消除情景感知系统，如图1所示，包括情景信息采集模块、情景信息处理模块、知识库模块、情景信息响应模块、情景信息应用模块、情景信息检索/订阅模块、情景信息校正模块以及用户反馈模块；
情景信息采集模块、情景信息处理模块、知识库模块依次连接，知识库模块、情景信息应用模块、情景信息检索/订阅模块、情景信息校正模块依次首尾连接，知识库模块、情景信息响应模块、情景信息应用模块依次连接，用户反馈模块分别连接知识库模块、情景信息应用模块；
情景信息采集模块：负责通过多个物理传感器、虚拟传感器和逻辑传感器周期性地采集原始情景信息，并将采集到的原始情景信息发送至情景信息处理模块，原始情景信息为多源情景信息，多源情景信息为通过多个传感器采集到的某一情景信息；例如，可以通过蓝牙、WIFI、红外、Zigbee等多个传感器采集用户的位置信息；
情景信息处理模块：负责对来自情景信息采集模块的原始情景信息进行处理；
知识库模块：负责存储用户反馈信息、情景信息融合推理信息、情景信息检索/订阅校正信息、情景应用信息，同时为情景信息响应模块提供各种情景应用信息，为情景信息处理模块提供所需要 的各种信息；
情景信息应用模块：负责将来自情景信息响应模块、用户反馈模块中的综合信息显示在情景感知系统界面上，并将情景信息发送给情景信息检索/订阅模块；
情景信息检索/订阅模块：根据情景信息应用模块对情景信息的检索需求，在知识库模块中检索相应的情景信息，根据情景信息应用模块对情景信息的订阅需求，将相关订阅信息发送到情景信息校正模块中；
情景信息响应模块：根据情景信息应用模块对情景信息的需求，在知识库模块中检索相应的情景信息，将情景信息应用模块所需的情景信息发送给情景信息应用模块；
情景信息校正模块：负责对情景信息检索/订阅模块发送来的情景信息进行校正，将校正后的情景信息发送到知识库模块；
用户反馈模块：负责将用户在某些环境下的情景信息存入知识库模块，并向情景信息应用模块提供所需应用情景信息。
情景信息处理模块包括情景信息建模单元、多算法不完备性消除单元、各算法结果不一致性消除单元、各信源不一致性消除单元、可信度管理单元、可靠性管理单元、情景信息融合推理单元以及自适应管理单元；
情景信息采集模块、情景信息建模单元、多算法不完备性消除单元、各算法结果不一致性消除单元、可信度管理单元、可靠性管理单元、各信源不一致性消除单元、情景信息融合推理单元依次连接，自适应管理单元分别连接多算法不完备性消除单元、可信度管理单元、各信源不一致性消除单元、情景信息融合推理单元，各信源不一致性消除单元连接可信度管理单元；情景信息融合推理单元连接知识库模块，知识库模块连接自适应管理单元；
情景信息建模单元：负责将情景信息采集模块采集来的多源情景信息按照知识库模块中的情景信息建模方式进行建模，建模模式为“情景感知类型+情景感知信息+情景感知精度”，情景感知类型为情景感知信息的类型，如情景感知信息“卧室”，其感知类型为“位置”，情景感知信息为各传感器采集的原始情景信息，情景感知精度为传感器固有的感知精度，例如“感知类型-用户位置”+“感知信息-卧室”+“感知精度-90％”将建模好的情景信息发送到多算法不完备消除单元；
多算法不完备性消除单元：负责同时采用多种算法消除情景信息的不完备性，得到多算法不完备性消除结果，将多算法不完备性消除结果发送到各算法结果不一致性消除单元；不完备性是指某一传感器所采集的原始情景信息有缺失；多种算法包括神经网络、证据论、最大期望算法、投票选举、模糊集理论算法；
各算法不一致性消除单元：负责采用投票选举算法消除多算法不完备消除结果中存在的不一致 性，得到精确性较高的完备情景信息，将精确性较高的完备情景信息发送给可信度管理单元；不一致性为采用各种算法消除情景信息不完备性的结果之间存在不一致。
可信度管理单元：利用接收到的用户反馈信息，判断精确性较高的完备情景信息的正确性，继而计算并存储精确性较高的完备情景信息对应信源的可信度：
Figure PCTCN2017104792-appb-000012
用户反馈信息是指用户根据所处环境所主动返回的情景信息；情景信息的正确性是基于用户反馈信息来判断的正确的情景信息；
可靠性管理单元：结合各传感器精度，以及信源的可信度，对各信源进行可靠性管理：
Figure PCTCN2017104792-appb-000013
传感器为采集该信源时所用传感器；传感器精度为静态标称值，可信度为动态评估值；将各信源的可靠性发送给各信源不一致性消除单元；
各信源不一致性消除单元，采用基于可靠性的证据理论算法消除各信源间的不一致性，不一致性是指同一时刻通过不同的传感器所采集到的情景信息之间存在的不一致性，例如，红外传感器采集到用户当前的位置信息为“卧室”，而Zigbee传感器采集到用户的位置信息为“客厅”，将各信源不一致性消除结果发送到可信度管理单元，同时将各信源不一致性消除结果发送到情景信息融合推理单元；
情景信息融合推理单元：运用本体推理、基于规则的推理、证据论或贝叶斯网络推理方法，根据采集到的各信源不一致性消除结果和自适应管理单元中的历史信息，推理出高级情景信息，将融合推理后的高级情景信息存储到知识库模块中；历史信息是指自适应管理单元中基于规则系统的规则引擎和规则集；高级情景信息是指经过推理融合后得到的可供用户或各种设备应用的情景信息；
自适应管理单元：负责为多算法不完备性消除单元提供用户反馈信息和情景信息检索/订阅后的校正信息，为可信度管理单元提供用户反馈信息，为情景信息融合推理单元提供基于规则系统的规则引擎和规则集，并根据当前情景信息对各种参数做出适当调整，使情景感知系统的自适应能力更好。
情景信息采集模块包括多个物理传感器、虚拟传感器和逻辑传感器。
实施例2
实施例1所述的基于可靠性管理的不确定性消除情景感知系统基于可靠性管理的不确定性消除工作方法，如图2所示，以情景感知计算的典型场景—智能家庭为例。在智能家庭中通过WIFI、蓝牙、红外和Zigbee 4种方法来采集关于人的位置信息，由WIFI、蓝牙、红外和Zigbee获取的 情景信息分别为IWIFI、I蓝牙、I红外、IZigbee，包括步骤如下：
S01：情景信息的收集和建模
收集多个物理传感器、虚拟传感器和逻辑传感器采集的情景信息，将情景信息按照知识库模块中的情景信息建模模式进行建模，建模模式为“情景感知类型+情景感知信息+情景感知精度”；
建模好的情景信息为：t1时刻：IWIFI＝“感知类型-用户位置”+“感知信息-客厅”+“感知精度-90％”，I蓝牙＝“感知类型-用户位置”+“感知信息-客厅”+“感知精度-92％”，IZigbee＝“感知类型-用户位置”+“感知信息-NaN(缺失)”+“感知精度-94％”，I红外＝“感知类型-用户位置”+“感知信息-卧室”+“感知精度-96％”；
t2时刻：IWIFI＝“感知类型-用户位置”+“感知信息-卧室”+“感知精度-90％”，I蓝牙＝“感知类型-用户位置”+“感知信息-卧室”+“感知精度-92％”，IZigbee＝“感知类型-用户位置”+“感知信息-卧室”+“感知精度-94％”，I红外＝“感知类型-用户位置”+“感知信息-卧室”+“感知精度-96％”；
t3时刻：IWIFI＝“感知类型-用户位置”+“感知信息-客厅”+“感知精度-90％”，I蓝牙＝“感知类型-用户位置”+“感知信息-户外”+“感知精度-92％”，IZigbee＝“感知类型-用户位置”+“感知信息-卧室”+“感知精度-94％”，I红外＝“感知类型-用户位置”+“感知信息-NaN”+“感知精度-96％”；
t4时刻：IWIFI＝“感知类型-用户位置”+“感知信息-卧室”+“感知精度-90％”，I蓝牙＝“感知类型-用户位置”+“感知信息-客厅”+“感知精度-92％”，IZigbee＝“感知类型-用户位置”+“感知信息-卧室”+“感知精度-94％”，I红外＝“感知类型-用户位置”+“感知信息-卧室”+“感知精度-96％”；
各传感设备采集的时间间隔固定且间隔适宜，时效性高；
S02：检测各情景信息是否完备
t3时刻显示当前位置信息为“户外”，属于不精确情景信息，在对其进行处理时，当作不完备信息来处理，检测到的IWIFI情景信息为完备情景信息，进入步骤S02’；除了IWIFI外其他传感设备采集的情景信息都存在不完备性，进入步骤S03；
S02’：信息存储
进行情景信息存储，以备步骤S06、步骤S09使用；
S03：计算不完备率
计算已建模好的每个传感器采集到的情景信息的不完备率：可以计算出各传感设备采集情景信息的不完备率分别为：IWIFI＝10％、I蓝牙＝8％、I红外＝6％、IZigbee＝4％；
S04：判断不完备性是否在可控范围内
各传感设备采集情景信息的不完备率分别为：IWIFI＝10％、I蓝牙＝8％、I红外＝6％、IZigbee＝4％，系统设置的不完备的可控范围为15％，都在可控范围内，进入步骤S05’；
S05’：判断是否存在用户反馈
检测当前时刻是否存在用户反馈，如果存在，进入步骤S06’，否则，进入步骤S06；
通过用户自动反馈自己当前的位置情景信息，来调整该情景感知系统对不确定性情景信息的处理，使系统更加稳定可靠；
S06：多算法消除不完备性
当前用户没有主动反馈自己当前的情景位置信息，则将根据来自信息存储的各种情景信息同时在横向(同一传感器不同时刻)和纵向(不同传感器同一时刻)上采用多种算法对情景信息不完备行进性消除，如神经网络、最大期望算法、投票选举、证据论、模糊集理论等算法；
S06’：直接补齐不完备性信息
如果当前时刻存在用户反馈信息，如“当前位置为卧室”，则将缺失位置信息补齐为“卧室”，进而得到完备的情景信息，返回步骤S02’；
S07：判断各算法结果是否存在不一致性
通过判断经过多种算法对位置信息不完备消除的结果是否一致，各算法得出的结果都为“卧室”，即各算法结果一致，返回步骤S02’；
S08：各算法结果不一致性消除
若通过多种算法消除不完备性的结果存在不一致性，即神经网络消除结果为“卧室”、最大期望算法消除结果为“客厅”、证据理论消除结果为“卧室”、模糊集理论消除结果为“客厅”、投票选举消除结果为“卧室”，则对各种算法结果进行简单不一致消除(投票选举)，得到质量较高的完备情景信息，最终将不完备位置信息补齐，即t1时刻IZigbee位置信息为“客厅”，t3时刻I蓝牙位置信息为“客厅”，I红外位置信息为“卧室”；
S09：判断各传感器是否存在不一致性
可以判断出，各传感器采集的位置信息在t1、t3、t4存在不一致性，进入步骤S10’；
S10：计算可信度
各传感器在t2时刻采集的位置信息一致，直接进行可信度计算，利用接收到的反馈信息判断多源情景信息的正确性，继而评估对应信息源的可信度：
Figure PCTCN2017104792-appb-000014
对信息源进行动态的计算和存储，注：位置信息的正确性是基于反馈信息来判断的)；
S10’：判断是否存在用户反馈
判断当前时刻用户是否主动反馈自己的位置信息，如果是，返回步骤S10，否则，进入步骤S11；
S11：各传感器不一致性消除
各传感器采集的位置信息在t1、t3、t4存在不一致性，则采用证据理论方法进行不一致性消除；
S12：计算可靠性
此例中，经计算I红外的可信度为92％，传感器本身固有的精度为96％，那么I红外的可靠性为2*0.92*0.96/(0.92+0.96)＝93.96％；并将该信源的可靠性结果用于所述步骤S11各传感器不一致性消除，使判决结果真实可靠；
图3是现有情景感知系统的仿真效果图；图4是本实施例关于本系统总体性能的仿真效果图。通过图3与图4对比可以看出，和现有情景感知系统相比，本发明所述情景感知系统的更加精确可靠；在用户反馈率为0.1，情景信息数量N＝1000时，现有情景感知系统的正确率为97.25％，本发明所述情景感知系统的正确率为99.34％。

Claims (4)

  1. 一种基于可靠性管理的不确定性消除情景感知系统，其特征在于，包括情景信息采集模块、情景信息处理模块、知识库模块、情景信息响应模块、情景信息应用模块、情景信息检索/订阅模块、情景信息校正模块以及用户反馈模块；
    所述情景信息采集模块、所述情景信息处理模块、所述知识库模块依次连接，所述知识库模块、所述情景信息应用模块、所述情景信息检索/订阅模块、所述情景信息校正模块依次首尾连接，所述知识库模块、所述情景信息响应模块、所述情景信息应用模块依次连接，所述用户反馈模块分别连接所述知识库模块、所述情景信息应用模块；
    所述情景信息采集模块：负责通过多个物理传感器、虚拟传感器和逻辑传感器周期性地采集原始情景信息，并将采集到的原始情景信息发送至所述情景信息处理模块，所述原始情景信息为多源情景信息，所述多源情景信息为通过多个传感器采集到的某一情景信息；
    所述情景信息处理模块：负责对来自所述情景信息采集模块的原始情景信息进行处理；
    所述知识库模块：负责存储用户反馈信息、情景信息融合推理信息、情景信息检索/订阅校正信息、情景应用信息，同时为所述情景信息响应模块提供各种情景应用信息，为所述情景信息处理模块提供所需要的各种信息；
    所述情景信息应用模块：负责将来自所述情景信息响应模块、所述用户反馈模块中的综合信息显示在情景感知系统界面上，并将情景信息发送给情景信息检索/订阅模块；
    所述情景信息检索/订阅模块：根据所述情景信息应用模块对情景信息的检索需求，在所述知识库模块中检索相应的情景信息，根据所述情景信息应用模块对情景信息的订阅需求，将相关订阅信息发送到所述情景信息校正模块中；
    所述情景信息响应模块：根据所述情景信息应用模块对情景信息的需求，在所述知识库模块中检索相应的情景信息，将所述情景信息应用模块所需的情景信息发送给所述情景信息应用模块；
    所述情景信息校正模块：负责对所述情景信息检索/订阅模块发送来的情景信息进行校正，将校正后的情景信息发送到知识库模块；
    所述用户反馈模块：负责将用户在某些环境下的情景信息存入所述知识库模块，并向所述情景信息应用模块提供所需应用情景信息。
  2. 根据权利要求1所述的一种基于可靠性管理的不确定性消除情景感知系统，其特征在于，所述情景信息处理模块包括情景信息建模单元、多算法不完备性消除单元、各算法结果不一致性消除单元、各信源不一致性消除单元、可信度管理单元、可靠性管理单元、情景信息融合推理单元以及自适应管理单元；
    所述情景信息采集模块、所述情景信息建模单元、所述多算法不完备性消除单元、所述各算法结果不一致性消除单元、所述可信度管理单元、所述可靠性管理单元、所述各信源不一致性消除单元、所述情景信息融合推理单元依次连接，所述自适应管理单元分别连接所述多算法不完备性消除单元、所述可信度管理单元、所述各信源不一致性消除单元、所述情景信息融合推理单元，所述各信源不一致性消除单元连接所述可信度管理单元；所述情景信息融合推理单元连接所述知识库模块，所述知识库模块连接所述自适应管理单元；
    所述情景信息建模单元：负责将情景信息采集模块采集来的多源情景信息按照所述知识库模块中的情景信息建模方式进行建模，建模模式为“情景感知类型+情景感知信息+情景感知精度”，所述情景感知类型为所述情景感知信息的类型，所述情景感知信息为各传感器采集的原始情景信息，所述情景感知精度为传感器固有的感知精度，将建模好的情景信息发送到所述多算法不完备消除单元；
    所述多算法不完备性消除单元：负责同时采用多种算法消除情景信息的不完备性，得到多算法不完备性消除结果，将多算法不完备性消除结果发送到所述各算法结果不一致性消除单元；所述不完备性是指某一传感器所采集的原始情景信息有缺失；所述多种算法包括神经网络、证据论、最大期望算法、投票选举、模糊集理论算法；
    所述各算法不一致性消除单元：负责采用投票选举算法消除多算法不完备消除结果中存在的不一致性，得到精确性较高的完备情景信息，将精确性较高的完备情景信息发送给所述可信度管理单元；所述不一致性为采用各种算法消除情景信息不完备性的结果之间存在不一致。
    所述可信度管理单元：利用接收到的用户反馈信息，判断精确性较高的完备情景信息的正确性，继而计算并存储精确性较高的完备情景信息对应信源的可信度：
    Figure PCTCN2017104792-appb-100001
    所述用户反馈信息是指用户根据所处环境所主动返回的情景信息；情景信息的正确性是基于用户反馈信息来判断的正确的情景信息；
    所述可靠性管理单元：结合各传感器精度，以及信源的可信度，对各信源进行可靠性管理：
    Figure PCTCN2017104792-appb-100002
    所述传感器为采集该信源时所用传感器；所述传感器精度为静态标称值，可信度为动态评估值；将各信源的可靠性发送给所述各信源不一致性消除单元；
    所述各信源不一致性消除单元，采用基于可靠性的证据理论算法消除各信源间的不一致性，所述不一致性是指同一时刻通过不同的传感器所采集到的情景信息之间存在的不一致性，将各信源不 一致性消除结果发送到所述可信度管理单元，同时将各信源不一致性消除结果发送到情景信息融合推理单元；
    所述情景信息融合推理单元：运用本体推理、基于规则的推理、证据论或贝叶斯网络推理方法，根据采集到的各信源不一致性消除结果和自适应管理单元中的历史信息，推理出高级情景信息，将融合推理后的高级情景信息存储到知识库模块中；所述历史信息是指自适应管理单元中基于规则系统的规则引擎和规则集；所述高级情景信息是指经过推理融合后得到的可供用户或各种设备应用的情景信息；
    所述自适应管理单元：负责为所述多算法不完备性消除单元提供用户反馈信息和情景信息检索/订阅后的校正信息，为所述可信度管理单元提供用户反馈信息，为所述情景信息融合推理单元提供基于规则系统的规则引擎和规则集，并根据当前情景信息对各种参数做出适当调整，使情景感知系统的自适应能力更好。
  3. 根据权利要求1所述的一种基于可靠性管理的不确定性消除情景感知系统，其特征在于，所述情景信息采集模块包括多个物理传感器、虚拟传感器和逻辑传感器。
  4. 权利要求2所述的一种基于可靠性管理的不确定性消除情景感知系统的工作方法，其特征在于，包括步骤如下：
    S01：情景信息的收集和建模
    收集多个物理传感器、虚拟传感器和逻辑传感器采集的情景信息，将情景信息按照知识库模块中的情景信息建模模式进行建模，建模模式为“情景感知类型+情景感知信息+情景感知精度”；
    S02：检测各情景信息是否完备
    检测已经建模好的每个传感器采集的情景信息是否完备，即情景信息有无缺失，如果完备，进入步骤S02’，否则，进入步骤S03；
    S02’：信息存储
    进行情景信息存储，以备步骤S06、步骤S09使用；
    S03：计算不完备率
    计算已建模好的每个传感器采集到的情景信息的不完备率：
    Figure PCTCN2017104792-appb-100003
    S04：判断不完备性是否在可控范围内
    判断情景信息的不完备性是否在可补齐范围内，如果是，进入步骤S05’，否则，进入步骤S05；
    S05：删除情景信息
    直接将该情景信息删除；
    S05’：判断是否存在用户反馈
    检测当前时刻是否存在用户反馈，如果存在，进入步骤S06’，否则，进入步骤S06；
    S06：多算法消除不完备性
    将来自信息存储的各情景信息同时在横向和纵向上采用多种算法对情景信息不完备性进行消除，横向是指同一传感器的不同时刻，纵向是指不同传感器的同一时刻，多种算法包括：神经网络、最大期望算法、投票选举、证据论、模糊集理论算法；
    S06’：直接补齐不完备性信息
    根据用户反馈信息补齐不完备性情景信息，得到完备信息，返回步骤S02’；
    S07：判断各算法结果是否存在不一致性
    判断通过多种算法对情景信息不完备性消除的结果是否存在不一致性，若各算法结果一致，则返回步骤S02’，否则，进入步骤S08；
    S08：各算法结果不一致性消除
    采用投票选举策略对各算法结果进行简单不一致性消除，得到质量较高的完备情景信息；
    S09：判断各传感器是否存在不一致性
    判断各传感器采集到的情景信息是否存在不一致性，如果存在，进入步骤S10’，否则，进入步骤S10；
    S10：计算可信度
    直接进行可信度计算：
    Figure PCTCN2017104792-appb-100004
    进入步骤S12，情景信息的正确性是基于用户反馈信息来判断的，在未出现情景信息不一致性前，该信源的可靠性为1；
    如果m是一个基本信度分配函数，则
    Figure PCTCN2017104792-appb-100005
    A上所有子集总的信度Bel(A)的公式如式(Ⅰ)所示：
    Figure PCTCN2017104792-appb-100006
    所定义的函数Bel是一个信任函数；
    设m1，m2是对应同一识别框架Θ上的两个基本信度分配函数，焦元分别为A1,A2,...,Ak和B1,B2,...,Bk，则两个信度函数的合成法则如式(Ⅱ)所示：
    Figure PCTCN2017104792-appb-100007
    m(A)反映了m1和m2对应的两个证据对命题A的联合支持程度，
    Figure PCTCN2017104792-appb-100008
    表示证据的冲突程度，当K＝0时，称为完全不冲突；当0＜K＜1时，称为非完全冲突；K＝1时，称为完全冲突，证据组合规则满足交换律和结合律，对于多个证据的组合可重复运用式(2)对多证据进行信度合成；
    S10’：判断是否存在用户反馈
    判断当前时刻是否存在用户反馈信息，如果是，返回步骤S10，否则，进入步骤S11；
    S11：各传感器不一致性消除
    采用证据理论方法进行不一致性消除；
    步骤S12：计算可靠性
    在可信度的基础上，结合传感器本身的精度进行可靠性计算：
    Figure PCTCN2017104792-appb-100009
    并将该信源的可靠性结果用于所述步骤S11各传感器不一致性消除：
    设Θ为识别框架，如果对于任何一个属于Θ的子集A满足
    Figure PCTCN2017104792-appb-100010
    则称m为识别框架Θ上的基本信度分配函数；
    Figure PCTCN2017104792-appb-100011
    m(A)称为子集A的基本信度值，m(A)反映了对A本身的信度大小，即A的可靠性。
PCT/CN2017/104792 2016-12-30 2017-09-30 一种基于可靠性管理的不确定性消除情景感知系统及其工作方法 WO2018120962A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
AU2017386444A AU2017386444B2 (en) 2016-12-30 2017-09-30 Reliability management-based uncertainty elimination context awareness system and working method thereof

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201611251838.3A CN106650941B (zh) 2016-12-30 2016-12-30 一种基于可靠性管理的不确定性消除情景感知系统及其工作方法
CN201611251838.3 2016-12-30

Publications (1)

Publication Number Publication Date
WO2018120962A1 true WO2018120962A1 (zh) 2018-07-05

Family

ID=58836846

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/104792 WO2018120962A1 (zh) 2016-12-30 2017-09-30 一种基于可靠性管理的不确定性消除情景感知系统及其工作方法

Country Status (3)

Country Link
CN (1) CN106650941B (zh)
AU (1) AU2017386444B2 (zh)
WO (1) WO2018120962A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112672299A (zh) * 2020-12-09 2021-04-16 电子科技大学 一种基于多源异构信息融合的传感器数据可信度评估方法
CN113533962A (zh) * 2021-07-29 2021-10-22 上海交通大学 基于多物理信号传感器决策融合的感应电机健康诊断系统

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106650941B (zh) * 2016-12-30 2019-01-25 山东大学 一种基于可靠性管理的不确定性消除情景感知系统及其工作方法
CN107705037B (zh) * 2017-10-26 2021-03-19 山东大学 一种面向上下文信息处理的QoX质量体系系统及其工作方法
CN108846481B (zh) * 2018-06-25 2021-08-27 山东大学 一种基于QoX自适应管理的情景信息不确定性消除系统及其工作方法
CN109982287B (zh) * 2019-03-13 2021-10-15 北京工业大学 基于zigbee无线传感器网络的林业多传感器火警告警系统
CN111062427B (zh) * 2019-12-11 2023-04-18 山东大学 基于多准则决策多模态情景信息不确定性处理方法及系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120059780A1 (en) * 2009-05-22 2012-03-08 Teknologian Tutkimuskeskus Vtt Context recognition in mobile devices
CN104346451A (zh) * 2014-10-29 2015-02-11 山东大学 一种基于用户反馈的情景感知系统及其工作方法和应用
CN105654134A (zh) * 2015-12-30 2016-06-08 山东大学 一种基于有监督自反馈的情景感知系统及其工作方法与应用
CN106650941A (zh) * 2016-12-30 2017-05-10 山东大学 一种基于可靠性管理的不确定性消除情景感知系统及其工作方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103246819B (zh) * 2013-05-20 2016-10-05 山东大学 一种面向普适计算的不一致性上下文消除系统和方法
CN105389405B (zh) * 2015-12-30 2018-11-30 山东大学 一种多推理引擎融合上下文感知系统及其工作方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120059780A1 (en) * 2009-05-22 2012-03-08 Teknologian Tutkimuskeskus Vtt Context recognition in mobile devices
CN104346451A (zh) * 2014-10-29 2015-02-11 山东大学 一种基于用户反馈的情景感知系统及其工作方法和应用
CN105654134A (zh) * 2015-12-30 2016-06-08 山东大学 一种基于有监督自反馈的情景感知系统及其工作方法与应用
CN106650941A (zh) * 2016-12-30 2017-05-10 山东大学 一种基于可靠性管理的不确定性消除情景感知系统及其工作方法

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112672299A (zh) * 2020-12-09 2021-04-16 电子科技大学 一种基于多源异构信息融合的传感器数据可信度评估方法
CN112672299B (zh) * 2020-12-09 2022-05-03 电子科技大学 一种基于多源异构信息融合的传感器数据可信度评估方法
CN113533962A (zh) * 2021-07-29 2021-10-22 上海交通大学 基于多物理信号传感器决策融合的感应电机健康诊断系统
CN113533962B (zh) * 2021-07-29 2022-08-12 上海交通大学 基于多物理信号传感器决策融合的感应电机健康诊断系统

Also Published As

Publication number Publication date
CN106650941A (zh) 2017-05-10
AU2017386444B2 (en) 2021-03-18
CN106650941B (zh) 2019-01-25
AU2017386444A1 (en) 2019-06-27

Similar Documents

Publication Publication Date Title
WO2018120962A1 (zh) 一种基于可靠性管理的不确定性消除情景感知系统及其工作方法
Ding et al. A survey on data fusion in internet of things: Towards secure and privacy-preserving fusion
JP6771105B2 (ja) フィードバックベースの自己適応主客観的重みコンテキストアウェアネスシステムおよびその動作方法。
CN103246819B (zh) 一种面向普适计算的不一致性上下文消除系统和方法
CN103326904A (zh) 一种基于多参数认知的快速网络拓扑推断方法
CN112433139B (zh) 一种提高超级电容循环寿命的方法及装置
JP6400834B2 (ja) 推薦装置、推薦決定方法、およびコンピュータプログラム
CN105654134B (zh) 一种基于有监督自反馈的情景感知系统及其工作方法与应用
Cheng et al. A model for integrating heterogeneous sensory data in IoT systems
Rapti et al. A bio-inspired service discovery and selection approach for IoT applications
CN106507275B (zh) 一种无线传感器网络的鲁棒分布式滤波方法和装置
WO2020001095A1 (zh) 一种基于层次化综合质量指标QoX的上下文不确定性消除系统及其工作方法
CN110243383B (zh) 用户回家导航方法、装置、计算机设备及存储介质
CN108846481B (zh) 一种基于QoX自适应管理的情景信息不确定性消除系统及其工作方法
Riboni et al. Context provenance to enhance the dependability of ambient intelligence systems
JP6864982B2 (ja) 推定装置
CN105867266A (zh) 一种智慧家庭管理装置及管理方法
Cai et al. Designing adaptive passive personal mobile sensing methods using reinforcement learning framework
US12008480B2 (en) Context information uncertainty elimination system and working method based on QoX adaptive management
CN105725991B (zh) 基于智能代理的健康监护系统
TW201616888A (zh) 多態流量網路之可靠度評估系統及其方法
CN117485265B (zh) 车载终端控制方法、装置及车载终端
Athanasiou et al. Towards personalization of Trust Management service for ubiquitous healthcare environment
CN116827975A (zh) 一种基于设备模型的算法应用方法
WO2022004404A1 (ja) 関係性推定システム

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17886639

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2017386444

Country of ref document: AU

Date of ref document: 20170930

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17886639

Country of ref document: EP

Kind code of ref document: A1