Research On Related Problems Of Plume Detection In Complex Environment

The current research on gas leakage is focused on plume tracking and source location.There are few studies on plume discovery,and most of them belong to global regular scan strategy.The shortcomings of these strategies are that the discovery efficiency is very low,and these strategies not suitable for a real complex environment.In order to solve these problems,this paper uses the idea of indirect search.It makes full use of other information of the environment to classify the search area,and then combines with the search strategy study to improve the plume discovery ability in the complex environment.The main research contents are as follows:1)Research on gas signal extraction algorithm: The completion of the plume discovery task is marked by the extraction of the corresponding gas signal,so the accuracy and stability of the gas signal extraction directly affect the effect of plume discovery.This paper proposes a gas signal extraction algorithm based on statistical characteristics and saliency.Firstly,calculate the saliency of gas spectrum at different scales,then select candidate signal points according to the statistical characteristics of the saliency.Secondly,study the statistical characteristics of the signals and use it to remove the false signal points from the candidate signal points.Finally,gas signals are extracted by baseline fitting and subtracting in the signal area.Experiments on gas spectrum signals with different signal-to-noise ratios and different baseline distortions show that the proposed algorithm has high accuracy and good stability.Its’ mean absolute error is only 8.71% of the Air PLS algorithm,3.52% of the Wavelet algorithm,and 2.01% of the Do G algorithm.And its root mean square of the absolute error is only 13.08%,5.45% and 3.11% of these three algorithms.2)Research on topology-grid hybrid map building: The boundary points that gotten by the RRT environmental exploration are optimized using the idea of mutual attraction between planets so as to reduce the calculation amount of the robot in the process of environmental exploration;Then the optimized boundary points are clustered to generate topology nodes,and the topology nodes are integrated into the grid map to generate a topology-grid hybrid map.Simulation and real experiments prove that the topology-grid hybrid map buliding algorithm in this paper can bulid maps in different environments so as to provide environmental information for plume discovery strategies.3)Research on plume discovery strategy: Firstly,a topological node risk coefficient model is created,which includes the initial environmental risk coefficient and the untraversed time risk coefficient.Then the plume discovery strategy is established,which determines the next traversal target of the robot through the risk coefficient of each node.So that the robot has a certain tendency in the process of plume discovery.Finally,it combines with the A* algorithm which has good path planning effect in a complex environment to realize the traversal path planning,so as to finally realize the plume discovery.The simulation experiments in three different complex environments show that even if excludes the failure of the grid traversal method and the Zigzag traversal method,the plume discovery time of the proposed algorithm is 53.8% and 46.7% shorter than the grid traversal method and the Zigzag traversal method respectively.In the real environment experiment,the algorithm of this paper is also 2.2% and 62.0% shorter than the grid traversal method and the Zigzag traversal method respectively.It proves that the plume discovery strategy proposed in this paper is effective,and can quickly complete the plume discovery task in different complex environments.