CN109738365A - A kind of gas leakage source searching method based on TDLAS sensor - Google Patents
A kind of gas leakage source searching method based on TDLAS sensor Download PDFInfo
- Publication number
- CN109738365A CN109738365A CN201811601199.8A CN201811601199A CN109738365A CN 109738365 A CN109738365 A CN 109738365A CN 201811601199 A CN201811601199 A CN 201811601199A CN 109738365 A CN109738365 A CN 109738365A
- Authority
- CN
- China
- Prior art keywords
- optical path
- sensor
- entropy
- source
- subsequent time
- 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
Abstract
The present invention relates to a kind of gas leakage source searching methods based on TDLAS sensor, the Dynamic Programming of sensor optical path are carried out in such a way that ground type mobile robot cooperates holder, including the following steps: [1] initialization of map and environmental parameter；[2] the smell Packet capturing number k and wind vector at current position r are acquiredInformation updates posterior probability using Bayesian inference；[3] current time global map information entropy is calculated, if entropy is lower than certain threshold value, then enable sensor optical path centered on maximum probability point, certain length is that radius does 360 degree of scannings, pass through the smell packet flux progress source acknowledgement calculated on the enclosed closed curve of scanning optical path；If entropy is still higher than given threshold, continue the search-path layout of subsequent time；[4] sensor optical path of subsequent time is planned.
Description
Technical field
The present invention relates to a kind of robot autonomous search source of leaks algorithm based on TDLAS gas concentration sensor, the calculation
Method realizes the accurate positionin to odor source with the information taxis way of search of additional geometrical constraint.
Background technique
With industrialized continuous development, the factory of some high-risk harmful toxic matters of storage is due to supervising the evil caused accidentally
Property gas leakage accidents frequently occur.Leakage once occur, gently then cause peculiar smell to disturb residents, cause fear, it is heavy then cause the people give birth to
Order the heavy losses of property safety.Therefore in the case where toxic or pernicious gas leaks, rapidly and accurately seeking is to letting out
The position of drain-source rationally disposes and rescues to the later period most important.
Various kinds of sensors is carried using robot, staff is can replace and enters danger zone and carry out leakage investigation.It examines
Consider bionics smell to trace to the source inherent mechanism, robot is needed when carrying out active path planning according to sensing data to going through
The data at history moment are reasonably utilized, rather than unilateral decision is only carried out according to the data at current time.For this
Problem, Vergassola etc. propose a kind of information taxis (Infotaxis) strategy for maximizing global information gain, the algorithm
Current information and historical information are effectively merged by Bayesian inference；This method is independent of DIFFUSION IN TURBULENCE model, energy simultaneously
Effectively overcome the problems, such as source search { Vergassola M, Villermaux E, the Shraiman B I. under turbulent environment
‘Infotaxis’as a Strategy for Searching without Gradients[J].Nature,2007,445
(7126):406.}.The essence that the algorithm maximizes Map Information Volume is more likely to robot to unknown area at search initial stage
The collection of domain information, but this search pattern is being searched for for carrying the robot of single-contact formula sensor without plume
It can be taken longer time in terms of overlay area.Simultaneously near gas source or the higher region of local concentration, robot is easy to produce
The raw self-trapping problem (turning back repeatedly in the region) in part, and then lead to the decline of source positioning accuracy.
TDLAS (tunable diode laser absorption spectroscopy) is that a kind of novel have highly selective optical gas concentration
Sensor, the output valve of the sensor are to sum to all gas concentration in optical path, therefore this line integral type sensor is being supervised
Survey the advantage for having single-contact formula sensor incomparable in range.Lilienthal team is carried using ground robot
The mode of TDLAS sensor carries out the dense distribution of gas and rebuilds { Arain M A, Trincavelli M, Cirillo M, et
al.Global coverage measurement planning strategies for mobile robots
equippedwith a remote gas sensor[J].Sensors,2015,15(3):6845-6871.}.But due to gas
Concentration distribution reconstruction tasks are higher to the demand of sampled data (i.e. more sampled datas, reconstructed results are more accurate), this grinds
The purpose for studying carefully middle robot path planning is maximization spatial coverage, this is apparently not a kind of efficient search pattern,
Both it had been unable to give full play advantage of the TDLAS sensor in measurement range, has not also combined the important informations such as wind speed and direction sufficiently.
In conclusion current gas leakage source searching algorithm is limited to the measurement range of single-contact formula sensor, but
Lack the effective use to TDLAS sensor again.
Summary of the invention
The purpose of the present invention is to propose to a kind of gas leakage source searching methods based on TDLAS gas concentration sensor, both
TDLAS sensor space can be given full play to and cover big advantage, and can be believed with effective integration other sensors (such as anemobiagraph)
Breath, and then improve search efficiency, overcome part self-trapping.Technical solution is as follows:
A kind of gas leakage source searching method based on TDLAS sensor, with ground type mobile robot cooperation holder
Mode carries out the Dynamic Programming of sensor optical path, including the following steps:
[1] initialization of map and environmental parameter；It each grid in region to be searched is set carves at the beginning and be identified as gas
The probability P in taste source0(r)；The parameters such as initialization source release rate R, diffusibility of gases D；
[2] the smell Packet capturing number k and wind vector at current position r are acquiredInformation, after being updated using Bayesian inference
Probability is tested, is predicted first when gaseous diffusion source is in sensor light pathWhen upper, each point be can be detected in grating map
Concentration valueSecondly the Poisson distribution coefficient of likelihood function in Bayesian inference model is sought using the prediction concentrations valueWherein dt is time interval, and q is that smell Bao Congyuan sets out to the experience that disappears
Average length, a are the effective search radius of sensor；Last normalized similarity processing, updates the posterior probability at current time
It is distributed Pt(r)；
[3] current time global map information entropy: S is calculatedt=∑ Pt(r)·lnPt(r), if entropy is lower than certain threshold
Value, then enable sensor optical path centered on maximum probability point, certain length be r radius do 360 degree of scannings, pass through calculate scanning light
Smell packet flux on the enclosed closed curve in road carries out source acknowledgement, overcomes the problems, such as that local concentration is self-trapping caused by excessively high with this；If
Entropy is still higher than given threshold, then continues the search-path layout of subsequent time；
[4] sensor optical path of subsequent time is planned；According to target point undetermined, sensor detection route undetermined is determined, with
Current time probabilistic distribution estimation optical path is moved to the Posterior probability distribution and its corresponding entropy of each sensor detection route undetermined
Value, and choose comentropy and reduce undetermined search coverage of the maximum route as subsequent time sensor optical path；
[5] robot makes sensor optical path be moved to specified region, and start next by displacement and cloud platform rotation
The iterative calculation at moment.
The main advantages of the present invention and characteristic be embodied in following aspects:
1. the present invention has given full play to advantage of the TDLAS gas concentration sensor in investigative range, cooperate cloud platform rotation
Robot is effectively reduced to the exploration time of no plume overlay area, improves search efficiency.
2. the present invention judges whether there is gaseous diffusion source and is being closed by calculating the smell packet flux on scanning closed curve
In curve, robot is effectively prevented near gas source or concentration upper zone falls into local convergence.
Detailed description of the invention
Fig. 1 is total algorithm flow chart of the invention.
Fig. 2 indicates that the rasterizing map of different moments, shade represent the probability size that grid is odor source.
Fig. 3 is the geometrical constraint that algorithm increases newly, i.e. sensor detection range becomes multiple points on same straight line.
Fig. 4 shows that the smell packet flux difference by odor source in or beyond closed curve can carry out source acknowledgement.
Fig. 5 is that sensor subsequent time optical path plans schematic diagram, and wherein optical path two-end-point respectively has 5 kinds of candidate items.
Specific embodiment
The present invention is described in detail below with reference to the accompanying drawings and embodiments.Embodiment is with technical side of the present invention
The specific implementation carried out premised on case gives detailed embodiment and process.But claims hereof protection scope
It is not only restricted to the description of following embodiments.Algorithm flow is as shown in Figure 1, steps are as follows:
[1] map and environmental parameter initialization.Discrete region to be searched is turned to the grating map of M × N, each grid table
Show the probability for being identified as odor source, and carve to meet at the beginning and be uniformly distributed, i.e., any grid is that the probability of odor source is 1/
(M×N)；Initialization source release rate R, diffusibility of gases D, robot size a, smell packet service life τ, the partial parameters are for subsequent
The building of likelihood function in Bayesian inference.
[2] smell Packet capturing number k and wind vector of the acquisition robot at current position rMore using Bayesian inference
New each grid is identified as the probability of odor source.By taking Fig. 2 as an example, it is assumed that previous moment map probability distribution as shown in the left diagram,
After a Bayesian inference updates, it is as shown on the right that probability distribution will appear variation.
Firstly, calculating when gaseous diffusion source is in sensor light path, each point be can be detected dense in grating map
Angle value c.Geometrical constraint is introduced herein as shown in figure 3, can be by line segmentUpper concentration line integral dismantling is the L products equidistantly put
Point, L=4 in legend, equidistant point is (3,1), (4,2), (5,3), (6,4).In line segmentThere are under conditions of gas source
On, any point on line segmentIt is uniformly distributed for the probability obedience of gas source, i.e.,Therefore line source is pre-
Survey the weighted sum that concentration distribution is represented by each point source prediction concentrations:
Wherein c (r | r0) can solve to obtain comprising Bessel function K by convective-diffusion equation0Analytic expression.
Secondly, it is assumed that in limited time interval dt, the probability that k smell packet is captured at the r of position obeys Poisson distribution
(i.e. likelihood function in Bayesian inference model) is then distributed according to line source prediction concentrationsIt can calculate in the period
Poisson distribution coefficient
Finally, updating the posteriority at current time as the prior probability at current time using the posterior probability at previous moment
Probability distribution, and pass through normalized:
[3] current time global map information entropy is calculated:If entropy is higher than setting threshold
Value, then continue the search-path layout of subsequent time；If entropy be lower than certain threshold value, as shown in figure 4, enable sensor optical path with
Centered on maximum probability point, certain length be that radius does 360 degree of scannings, and calculate the gas on the enclosed closed curve l of scanning optical path
Taste Bao Tongliang.
For infinitesimal a certain on closed curveProvide that its direction is directed toward outside closed curve along radius, detection in infinitesimal
Quantity to smell packet is kl, using wind vector direction as the direction of motion of smell packetThen on closed curve l
Smell packet flux are as follows:
Left figure shows that, when gas source is inside closed curve, flux should be larger positive value in Fig. 4；Right figure table in Fig. 4
Bright, when gas source is outside closed curve, flux should level off to zero as far as possible.
[4] sensor optical path of subsequent time is planned.5 are respectively moved to current time probabilistic distribution estimation optical path two-end-point
The Posterior probability distribution of a target point undetermined (upper and lower, left and right, original place) and its corresponding entropy, and choose comentropy and reduce maximum
Undetermined search coverage of the route as subsequent time sensor optical path.Since two-end-point respectively has 5 kinds of candidates (such as Fig. 5 institute
Show), therefore optical path candidate item shares 5 × 5=25 kind.
[5] robot makes sensor optical path be moved to specified region, and start next by displacement and cloud platform rotation
The iterative calculation at moment.
Claims (1)
1. a kind of gas leakage source searching method based on TDLAS sensor, with the side of ground type mobile robot cooperation holder
The Dynamic Programming of formula progress sensor optical path, including the following steps:
[1] initialization of map and environmental parameter；It each grid in region to be searched is set carves at the beginning and be identified as odor source
Probability P0(r)；The parameters such as initialization source release rate R, diffusibility of gases D；
[2] the smell Packet capturing number k and wind vector at current position r are acquiredIt is general to update posteriority using Bayesian inference for information
Rate is predicted first when gaseous diffusion source is in sensor light pathWhen upper, the detectable concentration of each point in grating map
ValueSecondly the Poisson distribution coefficient of likelihood function in Bayesian inference model is sought using the prediction concentrations valueWherein dt is time interval, and q is that smell Bao Congyuan sets out to the experience that disappears
Average length, a are the effective search radius of sensor；Last normalized similarity processing, updates the posterior probability at current time
It is distributed Pt(r)；
[3] current time global map information entropy: S is calculatedt=∑ Pt(r)·lnPt(r), if entropy is lower than certain threshold value,
Enable sensor optical path centered on maximum probability point, certain length be radius do 360 degree of scannings, by calculate scanning optical path enclosed
Smell packet flux on closed curve carries out source acknowledgement, overcomes the problems, such as that local concentration is self-trapping caused by excessively high with this；If entropy is still
Higher than given threshold, then continue the search-path layout of subsequent time；
[4] sensor optical path of subsequent time is planned；According to target point undetermined, sensor detection route undetermined is determined, with current
Moment probabilistic distribution estimation optical path is moved to the Posterior probability distribution and its corresponding entropy of each sensor detection route undetermined, and
It chooses comentropy and reduces undetermined search coverage of the maximum route as subsequent time sensor optical path；
[5] robot makes sensor optical path be moved to specified region, and start subsequent time by displacement and cloud platform rotation
Iterative calculation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811601199.8A CN109738365B (en) | 2018-12-26 | 2018-12-26 | Gas leakage source searching method based on TDLAS sensor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811601199.8A CN109738365B (en) | 2018-12-26 | 2018-12-26 | Gas leakage source searching method based on TDLAS sensor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109738365A true CN109738365A (en) | 2019-05-10 |
| CN109738365B CN109738365B (en) | 2021-10-01 |
Family
ID=66361311
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811601199.8A Active CN109738365B (en) | 2018-12-26 | 2018-12-26 | Gas leakage source searching method based on TDLAS sensor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109738365B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110514567A (en) * | 2019-08-28 | 2019-11-29 | 哈尔滨工业大学 | Gas source searching method based on comentropy |
| CN112950905A (en) * | 2021-02-01 | 2021-06-11 | 航天科技控股集团股份有限公司 | Gas station early warning system and method based on Internet of things |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102034030A (en) * | 2010-12-28 | 2011-04-27 | 杭州电子科技大学 | Method for cooperatively positioning dangerous odor source by multi-robot system |
| CN106940704A (en) * | 2016-11-25 | 2017-07-11 | 北京智能管家科技有限公司 | A kind of localization method and device based on grating map |
| CN108109162A (en) * | 2018-01-08 | 2018-06-01 | 中国石油大学(华东) | A kind of multiscale target tracking merged using self-adaptive features |
| CN108398660A (en) * | 2018-01-08 | 2018-08-14 | 国网江苏省电力有限公司 | A kind of terminal device localization method and system based on Wi-Fi cloud platform systems |
-
2018
- 2018-12-26 CN CN201811601199.8A patent/CN109738365B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102034030A (en) * | 2010-12-28 | 2011-04-27 | 杭州电子科技大学 | Method for cooperatively positioning dangerous odor source by multi-robot system |
| CN106940704A (en) * | 2016-11-25 | 2017-07-11 | 北京智能管家科技有限公司 | A kind of localization method and device based on grating map |
| CN108109162A (en) * | 2018-01-08 | 2018-06-01 | 中国石油大学(华东) | A kind of multiscale target tracking merged using self-adaptive features |
| CN108398660A (en) * | 2018-01-08 | 2018-08-14 | 国网江苏省电力有限公司 | A kind of terminal device localization method and system based on Wi-Fi cloud platform systems |
Non-Patent Citations (3)
| Title |
|---|
| ALEJANDRO R. GARCIA RAMIREZ ET AL.: "An Infotaxis Based Odor Navigation Approach", 《ISSNIP BIOSIGNALS AND BIOROBOTICS CONFERENCE 2011》 * |
| MASSIMO VERGASSOLA ET AL.: "‘Infotaxis’ as a strategy for searching without gradients", 《NATURE》 * |
| 宋程等: "基于认知差异的多机器人协同信息趋向烟羽源搜索方法", 《控制与决策》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110514567A (en) * | 2019-08-28 | 2019-11-29 | 哈尔滨工业大学 | Gas source searching method based on comentropy |
| CN110514567B (en) * | 2019-08-28 | 2021-10-29 | 哈尔滨工业大学 | Gas source searching method based on information entropy |
| CN112950905A (en) * | 2021-02-01 | 2021-06-11 | 航天科技控股集团股份有限公司 | Gas station early warning system and method based on Internet of things |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109738365B (en) | 2021-10-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dai et al. | Fast frontier-based information-driven autonomous exploration with an mav | |
| CN106843216B (en) | A kind of biology excitation complete traverse path planing method of robot based on backtracking search | |
| CN106814737B (en) | A kind of SLAM methods based on rodent models and RTAB Map closed loop detection algorithms | |
| Flaspohler et al. | Information-guided robotic maximum seek-and-sample in partially observable continuous environments | |
| Ai et al. | Coverage path planning for maritime search and rescue using reinforcement learning | |
| CN106056643B (en) | A kind of indoor dynamic scene SLAM method and system based on cloud | |
| CN108764560A (en) | Aircraft scene trajectory predictions method based on shot and long term Memory Neural Networks | |
| Wang et al. | Efficient autonomous exploration with incrementally built topological map in 3-D environments | |
| Nasrollahzadeh et al. | Optimal motion sensor placement in smart homes and intelligent environments using a hybrid WOA-PSO algorithm | |
| CN110926477A (en) | Unmanned aerial vehicle route planning and obstacle avoidance method | |
| CN109917394B (en) | Short-term intelligent extrapolation method based on weather radar | |
| CN109738365A (en) | A kind of gas leakage source searching method based on TDLAS sensor | |
| Niska et al. | Neural networks for the prediction of species-specific plot volumes using airborne laser scanning and aerial photographs | |
| Wang et al. | An intelligent UAV path planning optimization method for monitoring the risk of unattended offshore oil platforms | |
| Velez et al. | Modelling observation correlations for active exploration and robust object detection | |
| Cai et al. | A prior information‐based coverage path planner for underwater search and rescue using autonomous underwater vehicle (AUV) with side‐scan sonar | |
| CN117687416B (en) | Path planning method and system for river network water safety detection device | |
| Prokop et al. | Neuro-heuristic pallet detection for automated guided vehicle navigation | |
| Sahyoun et al. | Dynamic plume tracking using mobile sensors | |
| Wu et al. | An autonomous coverage path planning algorithm for maritime search and rescue of persons-in-water based on deep reinforcement learning | |
| Xu et al. | Trajectory prediction for autonomous driving with topometric map | |
| Sun et al. | Study on safe evacuation routes based on crowd density map of shopping mall | |
| Lawrance et al. | Fast marching adaptive sampling | |
| Huang et al. | Stochastic planning for asv navigation using satellite images | |
| Agishev et al. | Trajectory optimization using learned robot-terrain interaction model in exploration of large subterranean environments |
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 |