Research On Intelligent Landmine Anti-removal And Network Key Technologies

Landmines are a relatively inexpensive war defense weapon.For a long time,in almost all large-scale ground operations confrontation,landmines have highlighted the unique advantages of flexible operations,resulting in a large number of enemy casualties and equipment destruction.In recent years,all the world's military powers have developed mine warfare equipment as the focus of the engineering warfare equipment,especially emphasizing the technical cooperation with the development of the main battle equipment system,focusing on improving its intelligence and information warfare level,and developing into a wireless Intelligent mine with communication,target detection and independent decision making.The intelligent mine is deployed in a large-scale area in the enemy war zone.It works in an unmanned status for a long time.It relies on the battery to supply power.It monitors the targets in the war area through its own sensors and detection systems and attacks the target in an optimization way.Compared with traditional mines,intelligent mines have the ability to perceive the battlefield environment,and can exchange battlefield information with each other through wireless networks,and then have decision-making reconnaissance and attack on targets in the battlefield,thus maximizing the life cycle,with a small number of long-term blockade of the battlefield area.At present,there is no complete research system for intelligent and networked mines.At the same time,the particularity of the intelligent mine application environment puts high demands on various technologies on the intelligent mine.There are still problems left to be solved such as complex mine action recognition,time synchronization with outdoor environmental temperature changes,intelligent mine self-localization problem of large-scale low-beacon node density,and robust visual target tracking problem.To this end,this paper studies the key technologies of landmines considering unmanned,large-scale,limited energy,and complex environment,such as mine defense and network.Aiming at the problem of single lightning protection and poor recognition accuracy in the existing anti-discharge technology,the geomagnetic sensor and gyroscope are added on the basis of the acceleration sensor to capture the displacement and attitude change information of the intelligent mine;17 frequency and time domains are adopted.The signal quantization method is used to extract the feature information contained in the intelligent mine motion data,and the mine action feature vector space is constructed.Aiming at the computational problem caused by the high-dimensional feature space,a feature space-weighted feature selection algorithm is proposed.The dimension reduction of feature space is carried out.The intelligent radar multi-state support vector machine classifier based on directed acyclic graph is constructed.The data collection test of typical mine action is carried out.The accuracy of mine action recognition is analyzed with the number of features and the parameter changes.The law proves that the proposed feature selection algorithm can effectively improve the accuracy of mine action recognition.Aiming at the problem of time synchronization error increase of intelligent lightning network caused by frequent changes of ambient temperature,a dynamic estimation model of temperature compensation for intelligent lightning clock frequency shift is proposed.The problem of mismatching sampling rate of temperature data and clock frequency shift data is proposed.The two data are fused by the weighted summation of Almon function.For the problem of increasing synchronization error caused by temperature data and frequency shift data noise,the Kalman system equation of clock synchronization is constructed,and the Kalman filter is used to localize the node.The clock is compensated.For the problem that the synchronization algorithm brought by the fixed period time synchronization has large energy consumption,large channel occupation and poor adaptability,a resynchronization decision method for minimizing the risk of clock synchronization failure is proposed,which makes the node change according to the ambient temperature.The situation adaptively adjusts the node clock synchronization communication cycle;the time synchronization algorithm performance verification test is carried out.The intelligent mine time synchronization method proposed in this paper effectively improves the time synchronization period and has higher time synchronization precision than the traditional method.Aiming at the limited computing power of intelligent lightning,the traditional position estimation method has nonlinearity and the received signal strength indicates large noise.Based on the least square position estimation method,a weighted linear least squares position estimation method is proposed.The algorithm solves the problem of high beacon node density in the network.The idea of iterative update is used to construct the intelligent mine location algorithm.The location unknown node is used as the location reference of other nodes.At the same time,the node makes the positioning decision based on the spatial positional relationship and the location reliability of the neighbor reference nodes.In order to evaluate the positional reliability of the node,based on the ellipse fitting of the spatial point distribution,the linearization degree,dispersion degree and offset of the reference node are proposed.The degree quantization index is used to evaluate the node position estimation.The performance of the intelligent mine location algorithm under different beacon nodes and channel noise conditions is simulated.Aiming at the problem that the existing intelligent thunder target detection and tracking method is bulky and the target motion representation ability is poor,an image-based target detection and tracking method is proposed.For the intelligent mine battlefield application,there are strong shadow and illumination changes.Based on LBP,a dynamic adaptive time domain pixel difference background model based on ST-LBSP is proposed.For the problem of target occlusion and background motion,the minimum pixel distance between image blocks and image block and target historical data are calculated.The standardized difference squared distance between the target image blocks is integrated.At the same time,the image block is re-segmented by calculating the optical flow vector,intra-regional difference and inter-regional difference of the target image block after integration.A multi-target tracker based on structured support vector machine is constructed.The performance of the proposed algorithm was verified on the standard visual target tracker data set CDW2014.