CN113283516B - Multi-sensor data fusion method based on reinforcement learning and D-S evidence theory - Google Patents
Multi-sensor data fusion method based on reinforcement learning and D-S evidence theory Download PDFInfo
- Publication number
- CN113283516B CN113283516B CN202110605802.5A CN202110605802A CN113283516B CN 113283516 B CN113283516 B CN 113283516B CN 202110605802 A CN202110605802 A CN 202110605802A CN 113283516 B CN113283516 B CN 113283516B
- Authority
- CN
- China
- Prior art keywords
- data
- sensor data
- fusion
- adopting
- state
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The invention discloses a multi-sensor data fusion method based on reinforcement learning and a D-S evidence theory, which comprises the following steps: step one, inputting data of multiple sensors; step two, establishing a Markov decision model; thirdly, the Q-learning algorithm realizes conflict data resolution; and step four, realizing multi-sensor data fusion by adopting a D-S evidence theory. The method based on the combination of reinforcement learning and D-S evidence theory is adopted to quickly and efficiently realize the multi-sensor data fusion, and the effective multi-sensor data after conflict resolution is obtained by adopting the reinforcement learning to perform real-time online processing on conflict data and fault data in the multi-sensor, so that the problem of high data conflict is solved; and secondly, uncertain multi-source data can be fused well by adopting a D-S evidence theory fusion method, and the fusion performance is improved.
Description
Technical Field
The invention relates to the field of multi-sensor data fusion, in particular to a method for realizing multi-sensor data fusion based on reinforcement learning and a D-S evidence theory, which realizes real-time online effective fusion of multi-sensor data.
Background
With the application of modern science and technology to industrial equipment, the equipment is complicated, and the complex operation condition of the equipment cannot be accurately reflected by single sensor information. And because the device is easily interfered by the environment, the obtained data may have fault data, and the accurate decision of the complex equipment cannot be realized. And the multi-sensor can simultaneously reflect different characteristics of the system, and the multi-source data fusion is favorable for improving the system performance, so that the credibility of the result is high.
The multi-sensor data fusion technology is an important data processing technology which makes full use of multi-sensor data in different time and space and analyzes, sequences and fuses the data to improve the system performance, and plays a key role in actual production and application.
Among the multi-sensor data fusion models and methods, the D-S evidence theory is one of the most effective methods that can process uncertain information. The evidence theory also provides a Dempster combination rule, and the evidence can be fused under the condition of no prior information. However, a counterintuitive judgment results when there is a high degree of conflict between the evidences. In addition, in practice, there may be temporal inconsistency in the data obtained by multiple sensors, so that the method cannot realize real-time fusion of online data. The reinforcement learning is realized in a trial and error mode, and autonomous real-time interaction with the external environment can be realized without system prior information. Therefore, the real-time online fusion of the multi-sensor data is realized by adopting a method combining reinforcement learning and a D-S evidence theory.
Disclosure of Invention
In order to realize real-time online data fusion of multiple sensors, the invention provides an intelligent multi-sensor data fusion method based on reinforcement learning and a D-S evidence theory, and solves the problems of conflicting evidence and online real-time fusion.
The technical scheme adopted by the invention for solving the technical problem comprises the following steps:
the method comprises the following steps: multi-sensor data input
The multi-sensor data obtained is represented as: { D 1 ,D 2 ,…,D i In which D is i Data representing the ith sensor in the systemThere are a total of i sensors. In addition, the data expression of the sensor is a Basic Probability Assignment function (BPA). If new sensor data is added during the measurement process, the new sensor data is written into the data, and the data is expressed as: { D 1 ,D 2 ,…,D i ,D New }。
Step two: markov decision model construction
When the reinforcement learning is adopted to carry out self-adaptive conflict resolution on the multi-sensor data, a Markov Decision Process (MDP) model needs to be constructed. The MDP model comprises a system state, an action and a reward function, and specifically comprises the following steps:
(1) Action set A: because the information quantity of the multisource sensors is different, the system should make different action selections under different sensor information, so that when high conflict information exists, conflict resolution is carried out on the high conflict information to ensure the accuracy and validity of a fusion result, and therefore an action set is defined as follows: a = { a = 1 ,a 2 } = { reserved, deleted }, and the system can take actions of reservation or deletion according to actual situations;
(2) A state set S: when the fusion system takes a certain action under different sensor information at different moments, the state of the system is transferred, and a fusion result at the current moment is defined as the state of the system, namely:
wherein m is t And m t+1 Represents the fusion result of taking different actions at the current moment, a t+1 Indicating the action currently being taken. The system state set is represented as: s = { S = 1 ,s 2 ,…,s t ,…}。
(3) The reward function R: which represents the reward value or penalty value given by the system in a certain state s and a certain action a during the operation of the system.
In the multi-sensor data fusion algorithm, different actions are selected by setting a reward function value, and finally a more accurate and effective fusion result is obtained under the condition of the same data. Best pass in reinforcement learningThe large accumulated reward value can achieve the optimal action, so the reward function setting is very important. In the invention, a reward function is set according to the quality of the fusion result in a certain state, the quality of the fusion result is evaluated by adopting the Deng entropy which is a reliability measurement mode among evidences. Deng entropy E (m) of different states of system t ) Is defined as follows:
wherein, Θ represents the recognition framework, a is a subset in the recognition framework, and the corresponding BPA at time t is m t (A) And | a | represents the potential of a.
The higher the Dung entropy value is, the larger the information content contained by BPA is, which indicates that the fusion result in the current state is better. When E (m) t+1 )≥E(m t ) Now, the new state s is explained t+1 Advantageously, a positive reward should be given at this time. Otherwise, the new state s is illustrated t+1 Is negative, a penalty should be given at this time. The reward function at time t +1 is thus defined as:
step three: q-learning algorithm for realizing conflict data resolution
The purpose of realizing conflict resolution in multi-sensor data fusion is to find an optimal strategy pi: s → A, i.e. a = pi (S), so that the system makes the best strategy selection under the condition of conflict data, and effective conflict resolution is realized. The strategy is selected by repeatedly exploring and trial-and-error the environment and the intelligent agent, and finally the strategy of adding the maximum value of the immediate reward and the future reward of the system under a certain strategy is taken as the optimal strategy. The method specifically comprises the following steps: at time t, the multi-sensor data fusion system receives a certain state s t And an instant prize R t Determining the action currently to be performed (retaining or removing the current evidence) based on the reward value, and then the intelligent fusion system transitions to the next state s t+1 And generates a reward R t+1 . This is achieved byThe process is realized by adopting a Q-learning algorithm.
The Q-learning algorithm achieves the best strategy by maximizing the cumulative discount reward value, with the Q-value function being at a certain state s t And a certain action a t The following accumulated award values are expressed as:
where γ represents a discount factor.
And (3) selecting an optimal strategy by adopting an epsilon-greedy algorithm, wherein the optimal strategy is specifically represented as follows:
wherein, pi * (a | s) represents an optimal strategy; ε represents the probability of exploration.
And updating the Q value function by adopting the following formula:
wherein, alpha is (0, 1)]Denotes the learning rate, s t+1 Indicating the next state.
The multi-sensor data fusion system obtains the optimal action at each moment through repeated updating interaction with the environment, and finally eliminates high-conflict BPA to realize self-adaptive online data processing.
Step four: multi-sensor data fusion by evidence theory
After conflict resolution processing is carried out on sensor data, effective fusion of multi-source data is realized by adopting Dempster combination rules in an evidence theory, and the method specifically comprises the following steps:
wherein the content of the first and second substances,
the method has the advantages that the method combining the evidence theory and the reinforcement learning can efficiently and quickly realize the data fusion of the multiple sensors; according to the method, the conflict data and the fault data in the multiple sensors are processed in real time on line by adopting reinforcement learning, so that the effective multiple sensor data after conflict resolution is obtained, and the problem of high data conflict is solved; the D-S evidence theory fusion method adopted by the invention can well fuse uncertain multi-source data and improve the fusion performance.
Drawings
FIG. 1 is a diagram of an overall model of an implementation of the present invention;
FIG. 2 is vehicle fault data;
FIG. 3 is a multi-sensor data fusion result.
Detailed Description
The invention is further illustrated with reference to the following figures and examples. An example of a vehicle system fault diagnosis is given here, with experimental data from [1 ]]。[1]Shows that there are three types of faults in the vehicle system (here denoted as F) 1 ，F 2 And F 3 And is shown), low oil pressure, air leakage of the air intake system and blocking of the electromagnetic valve are respectively realized. Furthermore, it passes through five sensors C 1 ,C 2 ,C 3 ,C 4 ,C 5 And acquiring automobile fault data. The invention describes the implementation steps of the proposed method in connection with the vehicle system fault data.
The method comprises the following steps: multi-sensor data input
The multi-sensor data obtained from the automotive system is represented as: { D 1 ,D 2 ,D 3 ,D 4 ,D 5 Indicates that there are 5 sensors in the system. In the vehicle system fault data, C 5 The sensors are in a failure state, and the other sensors work normally. Specific data are shown in fig. 2. Wherein, m (F) 1 ) Representative failure type is F 1 Confidence value of m (F) 2 ) Representative failure type is F 2 Confidence value of m (F) 3 ) Representative failure type is F 3 M (Θ) represents an indeterminate confidence value for the fault type, Θ = { F = { 1 ,F 2 ,F 3 }. The invention aims to judge which type of fault occurs in an automobile system according to multi-sensor data.
Step two: markov decision model construction
The method comprises the steps of conducting self-adaptive conflict resolution on automobile system data through reinforcement learning, and firstly conducting Markov Decision Process (MDP) model construction. The MDP model comprises a system state, an action and a reward function, and specifically comprises the following steps:
(1) Action set A: because the information content of 5 sensors is different, the system should make different action selections under different sensor information, so that when high conflict information exists, conflict resolution is carried out on the high conflict information to ensure the accuracy and validity of a fusion result, and therefore the action set is as follows: a = { a = 1 ,a 2 } = { keep, delete }.
(2) A state set S: when the fusion system takes a certain action, the state of the system is transferred, and the invention defines the fusion result at the current moment as the state of the system, namely:
wherein m is t And m t+1 Indicating a fusion result of taking different actions at the current moment, a t+1 Indicating the action currently being taken. The set of system states is represented as: s = { S = 1 ,s 2 ,…,s t ,…}。
(3) The reward function R: which represents the reward value or penalty value given by the system in a certain state s and a certain action a during the operation of the system. In the invention, a reward function is set according to the quality of the fusion result in a certain state, the quality of the fusion result is evaluated by adopting the Deng entropy which is a reliability measurement mode among evidences. Deng entropy E (m) of different states of system t ) Is defined as:
wherein, Θ represents the recognition framework, a is a subset in the recognition framework, and the corresponding BPA at time t is m t (A) And | a | represents the potential of a.
The larger the value of the Dunnex is, the larger the information content contained in the BPA is, which indicates that the fusion result in the current state is better. When E (m) t+1 )≥E(m t ) Now, the new state s is explained t+1 Advantageously, a positive reward should be given at this time. Otherwise, the new state s is illustrated t+1 Is negative, a penalty should be given at this time. The reward function at time t +1 is thus defined as:
step three: q-learning algorithm for realizing conflict data resolution
The purpose of realizing conflict resolution in multi-sensor data fusion is to find an optimal strategy pi: s → A, i.e. a = pi (S), so that the system makes the best strategy selection under the condition of conflict data, and effective conflict resolution is realized. The strategy is selected by repeatedly exploring and trial-and-error the environment and the intelligent agent, and finally the strategy of adding the maximum value of the immediate reward and the future reward of the system under a certain strategy is taken as the optimal strategy. The method specifically comprises the following steps: at time t, the multi-sensor data fusion system receives a certain state s t And an instant prize R t Determining the action currently to be performed (retaining or removing the current evidence) based on the reward value, and then the intelligent fusion system transitions to the next state s t+1 And generates a reward R t+1 . The process is realized by adopting a Q-learning algorithm.
The Q-learning algorithm achieves the best strategy by maximizing the cumulative discount reward value, with the Q-value function being at a certain state s t And a certain action a t The following jackpot value is expressed as:
where γ represents a discount factor.
And (3) adopting an epsilon-greedy algorithm to select an optimal strategy, which is specifically represented as:
wherein, pi * (a | s) represents an optimal strategy; ε represents the probability of exploration.
And updating the Q value function by adopting the following formula:
wherein, alpha is (0, 1)]Indicates the learning rate, s t+1 Indicating the next state.
The multi-sensor data fusion system obtains the optimal action at each moment through repeated updating interaction with the environment, and finally eliminates high-conflict BPA to realize self-adaptive online data processing.
Step four: multi-sensor data fusion realized by adopting D-S evidence theory
After conflict resolution processing is carried out on sensor data, effective fusion of multi-source data is realized by adopting Dempster combination rules in an evidence theory, and the method specifically comprises the following steps:
wherein the content of the first and second substances,
finally, the multi-sensor data fusion method based on reinforcement learning and D-S evidence theory is adopted, simulation analysis is carried out according to the data in the figure 2, simulation comparison is carried out with the traditional Dempster combination rule and the Yager combination rule, and the fusion result is shown in figure 3. As can be seen from FIG. 3, the fusion method provided by the invention has relatively high fusion precision, and can realize efficient, accurate and intelligent fusion under multi-sensor data. The main reasons are: the reinforcement learning eliminates conflict data and fault data in the multi-sensor data, and avoids negative influence of the data on the fusion result. In addition, the method has the advantages that whether the data arrival time is consistent or not does not need to be considered, and the online fusion of the data can be effectively realized.
The foregoing shows and describes the general principles and broad features of the present invention and advantages thereof. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Reference to the literature
[1]Yan X H,Zhu J H,Kuang M C,et al.Missile aerodynamic design using reinforcement learning and transfer learning.Sci China Inf Sci,2018,61:119204.
Claims (1)
1. A multi-sensor data fusion method based on reinforcement learning and D-S evidence theory is characterized by comprising the following steps:
the method comprises the following steps: multi-sensor data input
The multi-sensor data obtained is represented as: { D 1 ,D 2 ,…,D i In which D is i Data representing the ith sensor, wherein the system has i sensors in total; in addition, the data expression of the sensor is a basic probability assignment function BPA; if new sensor data is added during the measurement process, the new sensor data is written into the data, and the data is expressed as: { D 1 ,D 2 ,…,D i ,D New }；
Step two: markov decision model construction
When the multi-sensor data is subjected to self-adaptive conflict resolution by adopting reinforcement learning, firstly, a Markov decision MDP model is constructed, wherein the MDP model comprises a system state, an action and a reward function, and specifically comprises the following steps:
(1) Action set A: define the action set as: a = { a = 1 ,a 2 } = { reserved, deleted }, and the system can take actions of reservation or deletion according to actual situations;
(2) A state set S: defining the fusion result at the current moment as the state of the system, namely:
wherein m is t And m t+1 Indicating a fusion result of taking different actions at the current moment, a t+1 Representing the currently taken action, the system state set is represented as: s = { S = 1 ,s 2 ,…,s t ,s t+1 ,…}；
(3) The reward function R: a reward function is set according to the quality of a fusion result in a certain state, the fusion result quality is evaluated by adopting the Dung entropy, and the Dung entropy E (m) of the system in different states t ) Is defined as:
wherein, Θ represents the recognition framework, a is a subset in the recognition framework, and the corresponding BPA at time t is m t (A) And | a | represents the potential of a;
the higher the value of the Dung entropy is, the larger the information content contained by BPA is, which indicates that the fusion result in the current state is better; when E (m) t+1 )≥E(m t ) Now, the new state s is explained t+1 Advantageously, a positive reward should be given; conversely, a penalty should be given; the reward function at time t +1 is thus defined as:
step three: q-learning algorithm for realizing conflict data resolution
Q-learning algorithmObtaining an optimal strategy by maximizing the cumulative discount reward value, the Q value function being at a certain state s t And a certain action a t The following accumulated award values are expressed as:
wherein γ represents a discount factor;
and (3) adopting an epsilon-greedy algorithm to select an optimal strategy, which is specifically represented as:
wherein, pi * (a | s) represents an optimal strategy; epsilon represents the exploration probability;
and updating the Q value function by adopting the following formula:
wherein, α ∈ (0, 1)]Indicates the learning rate, s t+1 Indicating a next state;
the multi-sensor data fusion system obtains the optimal action at each moment through repeated updating interaction with the environment, and finally eliminates high-conflict BPA to realize self-adaptive online data processing;
step four: multi-sensor data fusion realized by adopting D-S evidence theory
After conflict resolution processing is carried out on sensor data, effective fusion of multi-source data is realized by adopting Dempster combination rules in evidence theory, and the method specifically comprises the following steps:
wherein the content of the first and second substances,
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110605802.5A CN113283516B (en) | 2021-06-01 | 2021-06-01 | Multi-sensor data fusion method based on reinforcement learning and D-S evidence theory |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110605802.5A CN113283516B (en) | 2021-06-01 | 2021-06-01 | Multi-sensor data fusion method based on reinforcement learning and D-S evidence theory |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113283516A CN113283516A (en) | 2021-08-20 |
| CN113283516B true CN113283516B (en) | 2023-02-28 |
Family
ID=77282908
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110605802.5A Active CN113283516B (en) | 2021-06-01 | 2021-06-01 | Multi-sensor data fusion method based on reinforcement learning and D-S evidence theory |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113283516B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114997341A (en) * | 2022-08-01 | 2022-09-02 | 白杨时代(北京)科技有限公司 | Information fusion processing method and device |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103557884A (en) * | 2013-09-27 | 2014-02-05 | 杭州银江智慧城市技术集团有限公司 | Multi-sensor data fusion early warning method for monitoring electric transmission line tower |
| CN103942554A (en) * | 2014-05-07 | 2014-07-23 | 苏州搜客信息技术有限公司 | Image identifying method and device |
| CN104408324A (en) * | 2014-12-11 | 2015-03-11 | 云南师范大学 | D-S evidence theory based multi-sensor information fusion method |
| CN104598986A (en) * | 2014-12-12 | 2015-05-06 | 国家电网公司 | Big data based power load prediction method |
| CN104679991A (en) * | 2015-01-27 | 2015-06-03 | 吉林大学 | Ordered proposition-oriented novel method of information fusion |
| CN105516164A (en) * | 2015-12-22 | 2016-04-20 | 中国科学院长春光学精密机械与物理研究所 | P2P botnet detection method based on fractal and self-adaptation fusion |
| CN107967487A (en) * | 2017-11-27 | 2018-04-27 | 重庆邮电大学 | A kind of colliding data fusion method based on evidence distance and uncertainty |
| CN108804741A (en) * | 2018-04-04 | 2018-11-13 | 西北工业大学 | D-S evidence theory shell combination Algorithm of Firepower Allocation under efficiency maximal condition |
| CN109086470A (en) * | 2018-04-08 | 2018-12-25 | 北京建筑大学 | A kind of method for diagnosing faults based on fuzzy preference relation and D-S evidence theory |
| CN109532719A (en) * | 2018-11-23 | 2019-03-29 | 中汽研（天津）汽车工程研究院有限公司 | One kind being based on electric car combined of multi-sensor information |
| CN110188882A (en) * | 2018-12-28 | 2019-08-30 | 湖南大学 | A kind of high conflicting evidence fusion method based on fuzzy reasoning |
| CN110796194A (en) * | 2019-10-29 | 2020-02-14 | 中国人民解放军国防科技大学 | Target detection result fusion judgment method for multi-sensor information |
| CN111065089A (en) * | 2019-11-05 | 2020-04-24 | 东华大学 | Internet of vehicles bidirectional authentication method and device based on crowd sensing |
| CN111582399A (en) * | 2020-05-15 | 2020-08-25 | 吉林省森祥科技有限公司 | Multi-sensor information fusion method for sterilization robot |
| CN112435275A (en) * | 2020-12-07 | 2021-03-02 | 中国电子科技集团公司第二十研究所 | Unmanned aerial vehicle maneuvering target tracking method integrating Kalman filtering and DDQN algorithm |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101996157B (en) * | 2010-10-23 | 2013-08-21 | 山东科技大学 | Multisource information fusion method in evidence high-conflict environment |
| US11562565B2 (en) * | 2019-01-03 | 2023-01-24 | Lucomm Technologies, Inc. | System for physical-virtual environment fusion |
| CN111931806A (en) * | 2020-06-23 | 2020-11-13 | 广州杰赛科技股份有限公司 | Equipment fault diagnosis method and device for multi-sensor data fusion |
-
2021
- 2021-06-01 CN CN202110605802.5A patent/CN113283516B/en active Active
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103557884A (en) * | 2013-09-27 | 2014-02-05 | 杭州银江智慧城市技术集团有限公司 | Multi-sensor data fusion early warning method for monitoring electric transmission line tower |
| CN103942554A (en) * | 2014-05-07 | 2014-07-23 | 苏州搜客信息技术有限公司 | Image identifying method and device |
| CN104408324A (en) * | 2014-12-11 | 2015-03-11 | 云南师范大学 | D-S evidence theory based multi-sensor information fusion method |
| CN104598986A (en) * | 2014-12-12 | 2015-05-06 | 国家电网公司 | Big data based power load prediction method |
| CN104679991A (en) * | 2015-01-27 | 2015-06-03 | 吉林大学 | Ordered proposition-oriented novel method of information fusion |
| CN105516164A (en) * | 2015-12-22 | 2016-04-20 | 中国科学院长春光学精密机械与物理研究所 | P2P botnet detection method based on fractal and self-adaptation fusion |
| CN107967487A (en) * | 2017-11-27 | 2018-04-27 | 重庆邮电大学 | A kind of colliding data fusion method based on evidence distance and uncertainty |
| CN108804741A (en) * | 2018-04-04 | 2018-11-13 | 西北工业大学 | D-S evidence theory shell combination Algorithm of Firepower Allocation under efficiency maximal condition |
| CN109086470A (en) * | 2018-04-08 | 2018-12-25 | 北京建筑大学 | A kind of method for diagnosing faults based on fuzzy preference relation and D-S evidence theory |
| CN109532719A (en) * | 2018-11-23 | 2019-03-29 | 中汽研（天津）汽车工程研究院有限公司 | One kind being based on electric car combined of multi-sensor information |
| CN110188882A (en) * | 2018-12-28 | 2019-08-30 | 湖南大学 | A kind of high conflicting evidence fusion method based on fuzzy reasoning |
| CN110796194A (en) * | 2019-10-29 | 2020-02-14 | 中国人民解放军国防科技大学 | Target detection result fusion judgment method for multi-sensor information |
| CN111065089A (en) * | 2019-11-05 | 2020-04-24 | 东华大学 | Internet of vehicles bidirectional authentication method and device based on crowd sensing |
| CN111582399A (en) * | 2020-05-15 | 2020-08-25 | 吉林省森祥科技有限公司 | Multi-sensor information fusion method for sterilization robot |
| CN112435275A (en) * | 2020-12-07 | 2021-03-02 | 中国电子科技集团公司第二十研究所 | Unmanned aerial vehicle maneuvering target tracking method integrating Kalman filtering and DDQN algorithm |
Non-Patent Citations (5)
| Title |
|---|
| Application of multi-sensor information fusion technology in the power transformer fault diagnosis;Yong-Wei Li等;《2009 International Conference on Machine Learning and Cybernetics》;20090825;29-33 * |
| Multi-sensor data fusion based on the belief divergence measure of evidences and the belief entropy;Fuyuan Xiao;《Information Fusion》;20190331;第46卷;23-32 * |
| 基于本体的多源信息融合空间目标识别方法研究;刘斌;《中国博士学位论文全文数据库 工程科技Ⅱ辑》;20200215;第2020年卷(第2期);C032-5 * |
| 多源信息系统的信息融合及其数值特征;车晓雅;《中国优秀硕士学位论文全文数据库 信息科技辑》;20170915;第2017年卷(第9期);I140-22 * |
| 异构数据融合的CNC刀具磨损状态在线识别方法;冯凯等;《现代制造工程》;20201231(第8期);97-104 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113283516A (en) | 2021-08-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gu et al. | A novel lane-changing decision model for autonomous vehicles based on deep autoencoder network and XGBoost | |
| CN113935406B (en) | Mechanical equipment unsupervised fault diagnosis method based on countermeasure flow model | |
| CN109581871B (en) | Industrial control system intrusion detection method of immune countermeasure sample | |
| CN110807257A (en) | Method for predicting residual life of aircraft engine | |
| KR20210002018A (en) | Method for estimating a global uncertainty of a neural network | |
| CN111159642B (en) | Online track prediction method based on particle filtering | |
| CN113283516B (en) | Multi-sensor data fusion method based on reinforcement learning and D-S evidence theory | |
| CN108986142A (en) | Shelter target tracking based on the optimization of confidence map peak sidelobe ratio | |
| CN116610092A (en) | Method and system for vehicle analysis | |
| CN112082639A (en) | Engine state diagnosis method and diagnosis modeling method thereof | |
| Moore et al. | Data-driven extraction of vehicle states from CAN bus traffic for cyberprotection and safety | |
| JP6968282B2 (en) | Causal analysis of powertrain management | |
| Hamed et al. | Fuel consumption prediction model using machine learning | |
| CN113954855A (en) | Self-adaptive matching method for automobile driving mode | |
| CN110320802B (en) | Complex system signal time sequence identification method based on data visualization | |
| CN112329337A (en) | Aero-engine residual service life estimation method based on deep reinforcement learning | |
| Kang et al. | A transfer learning based abnormal can bus message detection system | |
| CN115373366A (en) | Interactive diagnosis system, diagnosis method and storage medium | |
| CN115659256A (en) | Semi-supervised learning hydraulic reversing valve fault diagnosis method based on multi-sensor information | |
| CN117195031A (en) | Electromagnetic radiation source individual identification method based on neural network and knowledge-graph dual-channel system | |
| CN113310689A (en) | Aeroengine transmission system fault diagnosis method based on domain self-adaptive graph convolution network | |
| CN110751054B (en) | Abnormal driving behavior detection system | |
| JP6950647B2 (en) | Data determination device, method, and program | |
| CN112418398A (en) | Safety monitoring method for power information equipment | |
| CN113243021A (en) | Method for training a neural network |
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 |