Vehicle three-dimensional passive positioning method based on information entropy link weight model

Abstract

The invention discloses a vehicle three-dimensional passive positioning method based on an information entropy link weight model, and the method can complete the passive positioning of a three-dimensional target for a vehicle based on a wireless sensor network. The method comprises four parts of wireless sensor network arrangement, three-dimensional positioning space voxel division, information entropy-based link weight distribution and three-dimensional positioning image reconstruction for description. Compared with a traditional three-dimensional wireless positioning technology needing target prior information or an active three-dimensional wireless positioning technology, the three-dimensional target positioning method provided by the invention does not need any prior information of the target, is high in environmental adaptability and accurate in positioning result, and is a more convenient and universal positioning means.

Description

Technical Field

[0001] The invention belongs to the technical field of three-dimensional target positioning, and in particular relates to a three-dimensional passive positioning method for a wireless sensor network information entropy link weight model of a vehicle. Background Art

[0002] The target passive positioning technology using wireless sensor networks was proposed by the United States in 2007. Since then, domestic and foreign scholars have conducted extensive research on this technology. At present, the mainstream research is mainly focused on the positioning algorithm of the two-dimensional plane, while there are few studies on three-dimensional target positioning. The technical principle of two-dimensional positioning is to use the signal reception strength changes of communication network nodes to complete the positioning and tracking of the target, including fingerprint positioning algorithm, geometric filtering algorithm, particle filtering algorithm, wireless tomography algorithm, etc. The fingerprint positioning algorithm records the offline training database by placing the target at different positions. When the target enters the monitoring area, the communication link signal reception strength at this time is matched with the offline database to complete the positioning; the geometric filtering algorithm screens out the links with large changes in the communication link signal reception strength in the monitoring area, and then uses the method of assigning weights to different links to obtain the intersection of valid links as the target position; the particle filtering algorithm uses the target's previous moment position information as a priori estimation, and estimates the target's next moment position, thereby obtaining the weight distribution of the adjustment particles in the monitoring area, and performs particle weight filtering processing, thereby establishing a state transition model to obtain the target position; the wireless tomography technology divides the space in the monitoring area into grids, and according to the link occlusion of the wireless sensor network, the target position information is obtained by image reconstruction through the target's shadow occlusion fading model.

[0003] At present, the two-dimensional positioning technology using wireless sensor networks is relatively mature at home and abroad, but there are still some problems in using wireless sensor networks for three-dimensional target positioning: (1) When performing three-dimensional target positioning, the target needs to be trained with data in advance, and there are few actual use scenarios; (2) Some mainstream positioning algorithms cannot be positioned in harsh outdoor environments and have poor environmental adaptability; (3) The active positioning method requires the target to carry a "tag" and cannot locate passive targets. Therefore, it is very necessary to explore a convenient, practical and adaptable three-dimensional positioning algorithm. Among various positioning algorithms, wireless tomography technology has great development potential due to its self-organizing node layout, no need for any prior information about the target, and strong environmental adaptability. The core key problem of this algorithm is how to accurately weight the communication link, which is exactly the key problem solved by the present invention. Summary of the invention

[0004] The purpose of the present invention is to provide a vehicle three-dimensional passive positioning method based on an information entropy link weight model to solve the problem of inaccurate link weight assignment in traditional three-dimensional target positioning using wireless tomography technology of a three-dimensional wireless sensor network.

[0005] The technical solution to achieve the purpose of the present invention is: the vehicle three-dimensional passive positioning method based on the information entropy link weight model described in the present invention deploys wireless sensor network nodes in the target area, and the nodes communicate with each other to form a three-dimensional communication link monitoring network, and then divides the target area into three-dimensional spatial voxel units, and assigns corresponding weight values ​​according to the information entropy of each link to obtain a voxel weight matrix, and finally reconstructs the voxel weight matrix to generate a spatial positioning result of the three-dimensional target.

[0006] Compared with the prior art, the present invention has the following significant advantages:

[0007] (1) The existing three-dimensional wireless tomography positioning technology based on wireless sensor networks is still immature. The present invention explores an effective method to perform three-dimensional passive positioning of large-volume targets such as vehicles, providing a technical reference for subsequent three-dimensional wireless positioning technology.

[0008] (2) Compared with the traditional three-dimensional wireless positioning technology that requires prior information of the target or active, the three-dimensional target positioning method provided by the present invention does not require any prior information of the target and is a more convenient and universal positioning method. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 This is an overall flow chart of a vehicle three-dimensional passive positioning method based on an information entropy link weight model proposed by the present invention.

[0010] Figure 2 This is a schematic diagram of the deployment of a wireless sensor network for a vehicle three-dimensional passive positioning method based on an information entropy link weight model proposed in the present invention.

[0011] Figure 3 A schematic diagram of voxel division of a three-dimensional positioning space for a vehicle three-dimensional passive positioning method based on an information entropy link weight model proposed in the present invention.

[0012] Figure 4 This is a schematic diagram of the effective area of ​​communication link positioning for a vehicle three-dimensional passive positioning method based on an information entropy link weight model proposed in the present invention.

[0013] Figure 5 A schematic diagram of vehicle three-dimensional positioning results of a vehicle three-dimensional passive positioning method based on an information entropy link weight model proposed in the present invention.

[0014] Figure 6This is a schematic diagram of the projection on the XOY plane of the three-dimensional positioning result of the vehicle three-dimensional passive positioning method based on the information entropy link weight model proposed by the present invention.

[0015] Figure 7 This is a schematic diagram of the projection on the XOZ plane of the three-dimensional positioning result of the vehicle three-dimensional passive positioning method based on the information entropy link weight model proposed by the present invention. DETAILED DESCRIPTION

[0016] The present invention is further described in detail below in conjunction with the accompanying drawings.

[0017] Combination Figure 1 The present invention proposes a three-dimensional passive positioning method for vehicles based on an information entropy link weight model. The design concept is as follows: first, wireless sensor network nodes are deployed in the target area to form a three-dimensional communication link monitoring network. When the node communication link is blocked by a vehicle, the signal strength is reduced due to spatial fading. Then, the target area is divided into three-dimensional spatial voxel units. The communication links with signal fading pass through different voxels. The information entropy of each link is assigned a corresponding weight value, and mapped to obtain a voxel weight matrix. Finally, the voxel weight matrix is ​​reconstructed into a three-dimensional heat map to generate a spatial positioning result of the three-dimensional target.

[0018] Combination Figure 1 The three-dimensional passive positioning method of a vehicle based on an information entropy link weight model of the present invention comprises the following steps:

[0019] The first step is to deploy a wireless sensor network;

[0020] N wireless sensor network nodes are deployed on both sides of the selected road, with each network node spaced M meters apart, for a total of 2N wireless sensor network nodes. In the direction perpendicular to the road surface, the height distribution of the wireless sensor network nodes is set to h meters, h+0.5 meters, h+1 meters, and so on. The communication links of the deployed wireless sensor network form a three-dimensional monitoring network in space. When a vehicle is in the monitoring area, the signal strength of the communication link is attenuated due to the occlusion of the vehicle, laying a data foundation for target positioning.

[0021] The second step is to build a cube model of the three-dimensional monitoring space covered by the wireless sensor network and divide the modeled three-dimensional positioning space voxels as follows:

[0022] The three-dimensional monitoring space covered by the wireless sensor network is innovatively modeled as a cube, and the minimum voxel size of the three-dimensional monitoring space is set to x0mm*y0mm*z0mm, then the entire three-dimensional monitoring space is divided into K voxels of the same size. Theoretically, the smaller the voxel size, the higher the precision of the data and the better the positioning effect. Let the direction along the road be the X-axis, the direction through the road surface be the Y-axis, and the direction perpendicular to the road surface be the Z-axis, then each voxel can be projected on the XOY plane, XOZ plane, and YOZ plane respectively, and the target positioning results can also be reflected on different two-dimensional planes.

[0023] The third step is to allocate link weights based on information entropy. Information entropy is used to calculate the contribution of different communication links to the three-dimensional positioning of the vehicle and calculate the voxel weight matrix, as follows:

[0024] Due to the differences in the spatial distribution of links, the amount of target information that can be provided by all communication links in the three-dimensional monitoring space is different. For example, if some communication links in the space are not blocked by vehicles under any circumstances or the blocking degree is negligible, the target information that can be provided by these communication links is almost zero, and the link weight is low; while if some communication links are blocked by vehicles to a large extent, they can provide more target information and have a higher weight. Based on this weight distribution consideration, information entropy is used to calculate the contribution of different communication links to the three-dimensional positioning of the vehicle:

[0025]

[0026] Wherein, a represents the number of states into which the communication link is divided, P(x) represents the probability that the communication link is blocked, and I represents the information entropy value provided by the communication link.

[0027] If the effective area of ​​the i-th communication link in the XOY plane is S i , the total area of ​​the effective area projected on the XOY plane is S, then the probability that the i-th communication link is blocked is defined as P i for:

[0028]

[0029] In fact, the probability of a communication link being blocked is directly related to the effective area. The relationship between the link weight value and the effective area obtained by the innovative research of the present invention is as follows:

[0030]

[0031] Among them, α and σ l are adjustment parameters, w i is the link weight value.

[0032] According to the above link weight values, the weights of all communication links can be calculated to obtain the weight matrix of the communication links:

[0033] W=[w1,w2,...w i ,...w n ] T

[0034] Among them, w i is the weight of the i-th communication link, i=1,2,…n.

[0035] When the weight value of the communication link is determined, all communication links passing through the vehicle are projected onto the XOY plane and the XOZ plane. On the XOY plane projection, there is a communication link between the two wireless sensor nodes. If the weight of this communication link is p, the voxel projections passed by this communication link all obtain the weight p, while the weight of the projected pixels not passed by this communication link is 0.

[0036] By weight mapping all links and voxels, the following voxel weight matrix can be obtained:

[0037]

[0038] Where n is the number of communication links; s i is the state of communication link i; u i is the sum of the voxel weights generated by the communication link projection, which is an r-dimensional vector; C u is the spatial covariance matrix, which is an r-dimensional matrix, y u Represents the weight of each voxel.

[0039] In the fourth step, according to the voxel weight matrix y u , perform three-dimensional positioning image reconstruction and generate the three-dimensional positioning result of the vehicle, as follows:

[0040] According to the voxel weight matrix y u , each voxel in the matrix has obtained a corresponding weight value, and the spatial position of each voxel is also determined. The voxel weight matrix can be converted into a three-dimensional heat map, and voxels with high voxel weights can be given a higher color display result, so as to generate a three-dimensional positioning result of the vehicle.

[0041] Example 1

[0042] Combination Figure 1 The three-dimensional passive positioning method of a vehicle based on an information entropy link weight model of the present invention comprises the following steps:

[0043] In the first step, the present invention designs the following road for vehicle travel: Figure 2The wireless sensor network topology shown in the figure has 9 wireless sensor network nodes deployed on both sides of the selected road, with a 2-meter interval between each network node, and a total of 18 wireless sensor network nodes deployed. In the vertical direction of the road surface, the height distribution of the wireless sensor network nodes is set to 0.5 meters, 1 meter, 1.5 meters, and so on. A three-dimensional space monitoring area of ​​16 meters * 4 meters * 1.5 meters is formed.

[0044] The second step is to combine Figure 3 The present invention divides the selected three-dimensional space monitoring area into spatial voxels. Theoretically, the smaller the voxel size, the higher the data precision and the better the positioning effect. Considering that the smaller the voxel division, the larger the data processing volume, the poorer the real-time positioning performance, the present invention comprehensively considers that the minimum voxel size of the three-dimensional monitoring space is 500mm*200mm*100mm, and the entire three-dimensional monitoring space is divided into 9600 voxels of the same size. Let the direction along the road be the X-axis, the direction through the road surface be the Y-axis, and the direction perpendicular to the road surface be the Z-axis.

[0045] In the third step, due to the differences in the spatial distribution of links, each communication link can provide different target information. Information entropy is used to scientifically calculate the contribution of different communication links to the three-dimensional positioning of the vehicle:

[0046]

[0047] Where a represents the number of states that the communication link is divided into, P(x) represents the probability that the communication link is blocked, and I represents the information entropy value provided by the communication link. Since the communication link has only two states: blocked and unblocked, a=2 is taken.

[0048] Combination Figure 4 In order to calculate the probability P(x) of link being blocked, since the link in XOY projection can intuitively show its contribution, the “effective area” of the communication link is defined as the rectangular area with the communication link as the diagonal in the XOY projection plane. Figure 4 The dashed part in the middle represents the communication link, and the dark area represents the effective area of ​​the communication link. The effective area of ​​the communication link i in the XOY plane is defined as S i , the total area of ​​the effective area projected on the XOY plane is S, then the probability of communication link i being blocked is defined as P i for:

[0049]

[0050] The relationship between the link weight value and the effective area is as follows:

[0051]

[0052] Among them, α and σ l are adjustment parameters, w i The link weight calculation result.

[0053] According to the above link weight values, the weights of all communication links can be calculated to obtain the weight matrix of the communication links:

[0054] W=[w1,w2,...w i ,...w n ] T

[0055] Among them, w i is the weight of the ith communication link.

[0056] After the communication link weight is determined, if the weight of this communication link is p, the voxel projections passed by this communication link all obtain weight p, and the weight of the projection pixels not passed by this communication link is 0. By weight mapping all links and voxels, the following voxel weight matrix can be obtained:

[0057]

[0058] Where n is the number of communication links; s i is the state of communication link i; u i is the sum of the voxel weights generated by the communication link projection, which is an r-dimensional vector; C u is the spatial covariance matrix, which is an r-dimensional matrix, y u Represents the weight of each voxel.

[0059] The fourth step is to use the voxel weight matrix y u , the voxel weight matrix can be converted into a three-dimensional heat map, and voxels with high voxel weights can be given a higher color display result, thereby generating a three-dimensional positioning result of the vehicle. Figure 5 , Figure 6 , Figure 7 The image shows the positioning result of the vehicle at (X = 10 meters, Y = 2 meters), where Figure 5 The three-dimensional spatial positioning results of the target are shown. Figure 6 The two-dimensional positioning result of the target in the XOY plane is displayed, from which the coordinate position of the target on the road surface can be obtained. Figure 7 The two-dimensional positioning results of the target in the XOZ plane are displayed, from which the height information of the target can be obtained.

[0060] In summary, the present invention innovatively proposes a vehicle three-dimensional passive positioning method based on the information entropy link weight model to realize the three-dimensional passive positioning and recognition of vehicle targets, and the recognition results are accurate and intuitive.

Claims

1. A three-dimensional passive positioning method for vehicles based on information entropy link weight model, characterized in that: Here are the steps: The first step is to deploy a wireless sensor network; The second step is to build a cube model of the three-dimensional monitoring space covered by the wireless sensor network and divide the modeled three-dimensional positioning space voxels; The third step is to allocate link weights based on information entropy. Information entropy is used to calculate the contribution of different communication links to the three-dimensional positioning of the vehicle and calculate the voxel weight matrix. The fourth step is to use the voxel weight matrix y u , perform three-dimensional positioning image reconstruction and generate the three-dimensional positioning result of the vehicle.

2. The vehicle three-dimensional passive positioning method based on the information entropy link weight model according to claim 1 is characterized in that: In the first step, the wireless sensor network is deployed as follows: N wireless sensor network nodes are deployed on both sides of the selected road, with each network node spaced M meters apart, for a total of 2N wireless sensor network nodes; In the vertical direction of the road, the height distribution of the wireless sensor network nodes is set to h meters, h+0.5 meters, h+1 meters, and so on; The deployed wireless sensor network forms a three-dimensional monitoring network through communication links in space. When a vehicle is in the monitoring area, the signal strength of the communication link is attenuated due to the obstruction of the vehicle, laying a data foundation for target positioning.

3. The vehicle three-dimensional passive positioning method based on the information entropy link weight model according to claim 2 is characterized in that: In the second step, the three-dimensional monitoring space covered by the wireless sensor network is cube-modeled, and the modeled three-dimensional positioning space voxels are divided as follows: The three-dimensional monitoring space covered by the wireless sensor network is modeled as a cube, and the minimum voxel size of the three-dimensional monitoring space is set to x0*y0*z0, then the entire three-dimensional monitoring space is divided into K voxels of the same size; Let the direction along the road be the X-axis, the direction through the road surface be the Y-axis, and the direction perpendicular to the road surface be the Z-axis. Each voxel forms a projection on the XOY plane, XOZ plane, and YOZ plane respectively, and the target positioning results are reflected on different two-dimensional planes.

4. The vehicle three-dimensional passive positioning method based on the information entropy link weight model according to claim 3 is characterized in that: In the third step, the link weight allocation based on information entropy is used to calculate the contribution of different communication links to the three-dimensional positioning of the vehicle and calculate the voxel weight matrix, which is as follows: Due to the differences in the spatial distribution of links, all communication links in the three-dimensional monitoring space can provide different amounts of target information. Based on the weight distribution considerations, information entropy is used to calculate the contribution of different communication links to the three-dimensional positioning of the vehicle: Where a represents the number of states into which the communication link is divided, P(x) represents the probability that the communication link is blocked, and I represents the information entropy value provided by the communication link; If the effective area of ​​the i-th communication link in the XOY plane is S i , the total area of ​​the effective area projected on the XOY plane is S, then the probability that the i-th communication link is blocked is defined as P i for: The relationship between the link weight value and the effective area is as follows: Among them, α and σl are adjustment parameters, wi is the link weight value; According to the above link weight values, the weights of all communication links can be calculated to obtain the weight matrix W of the communication links: W=[w1,w2,...w i ,...wn] T Among them, w i is the weight of the i-th communication link; i = 1, 2, ... n; When the weight value of the communication link is determined, all communication links passing through the vehicle are projected onto the XOY plane and the XOZ plane. On the XOY plane projection, there is a communication link between two wireless sensor nodes. If the weight of this communication link is p, then the voxel projections passed by this communication link all obtain the weight p, while the weight of the projection pixels not passed by this communication link is 0. By performing weight mapping on all links and voxels, we get the voxel weight matrix: Where n is the number of communication links; s i is the state of the ith communication link; u i is the sum of the voxel weights generated by the projection of the i-th communication link, which is an r-dimensional vector; C u is the spatial covariance matrix, which is an r-dimensional matrix, y u Represents the weight of each voxel.

5. The vehicle three-dimensional passive positioning method based on the information entropy link weight model according to claim 4 is characterized in that: In the fourth step, according to the voxel weight matrix y u , perform three-dimensional positioning image reconstruction and generate the three-dimensional positioning result of the vehicle, as follows: According to the voxel weight matrix y u , each voxel in the matrix has obtained a corresponding weight value, and the spatial position of each voxel is also determined. The voxel weight matrix can be converted into a three-dimensional heat map, and voxels with high voxel weights can be given a higher color display result, that is, the three-dimensional positioning result of the vehicle is generated.

Images