Dynamic positioning method resistant to non-line-of-sight interference and base station layout method for positioning

Abstract

The invention belongs to the technical field of wireless positioning, and discloses a non-line-of-sight interference-resistant dynamic positioning method and a base station layout method for positioning. Comprising the following steps: calculating original distance measurement results of each initial base station in a label and a sub-region, obtaining current shielding probability and current signal intensity of the initial base station and current geometric accuracy factors of different initial base station combinations in an initial base station set, selecting candidate base stations, calculating information gain of each candidate base station pair, and obtaining the information gain of each candidate base station pair; selecting candidate base stations with information gains meeting preset requirements to construct a target base station subset, inputting an original ranging result of the target base station subset into a Kalman filter based on a motion state space model for processing, and compensating an NLOS time delay error through a sliding window mean value to obtain a positioning result of a label and an NLOS time delay error estimation value, the problem that the positioning result is not accurate enough in a complex environment is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of wireless positioning technology, in particular to a dynamic positioning method against non-line-of-sight interference and a base station layout method for positioning. BACKGROUND

[0002] In the ultra-wideband (UWB) positioning technology, non-line-of-sight (NLOS) interference is one of the main factors affecting the positioning accuracy. When there are obstacles (such as partitions, cables, metal structures) or electromagnetic interference in the signal propagation path, the signal will be reflected, refracted or scattered, resulting in a significant NLOS time delay error in the measured time delay, and thus causing the positioning result to deviate from the true position.

[0003] The traditional UWB positioning method usually uses a fixed model or simple filtering to process the NLOS time delay error, and the positioning accuracy of the tag in a dynamically changing complex environment is not enough. SUMMARY

[0004] In view of the above defects of the prior art, the present application provides a dynamic positioning method against non-line-of-sight interference and a base station layout method for positioning to solve the problem of how to improve the positioning accuracy in a complex environment.

[0005] To achieve the above object, the first aspect of the present application provides a dynamic positioning method against non-line-of-sight interference, comprising: S1, calculating original ranging results of a tag and each initial base station in real time through bidirectional signal interaction between the tag and each initial base station in an initial base station set, the initial base station set including all base stations in a sub-region where the tag is located; S2, grouping the initial base stations to obtain different initial base station combinations, calculating current geometric dilution of precision of different initial base station combinations, and obtaining current occlusion probability and current signal strength of each initial base station, and selecting candidate base stations from the initial base station set in combination with corresponding current geometric dilution of precision threshold, occlusion probability threshold and signal strength threshold; S3, grouping each candidate base station to obtain different candidate base station pairs, calculating information gain of different candidate base station pairs, and selecting candidate base station pairs whose information gain meets preset requirements to construct a target base station subset; S4, establishing a state vector including position coordinates, velocity components and NLOS time delay error of the tag, constructing a state transition equation based on motion characteristics of the tag, and constructing a distance observation equation based on position coordinates of all target base stations in the target base station subset to form a motion state space model of the tag, the state transition equation being used to describe dynamic change relationship of the state vector, and the distance observation equation being used to associate measured distance of the tag with the state vector; S5, inputting original ranging results of the target base station subset into a Kalman filter based on the motion state space model for processing, compensating NLOS time delay error of the tag through sliding window mean of NLOS time delay error estimate value, and dynamically adjusting process noise covariance matrix in the Kalman filter according to three-dimensional acceleration of the tag collected in real time by an inertial measurement unit to obtain tag positioning result of a current period and NLOS time delay error estimate value, the NLOS time delay error estimate value being used to update sliding window mean to compensate NLOS time delay error of a next period, and realizing dynamic positioning of the tag.

[0006] The second aspect of the present application provides a base station layout method for positioning, comprising:

[0007] In the case that the positioning area is a tunnel structure, the positioning area is an activity area of the tag:

[0008] If the tunnel is in a straight line form, each sub-region is arranged in a right triangular space on both side walls and the top of the tunnel according to a first interval;

[0009] If the tunnel is in an S-shaped form, a third preset number of base stations are arranged in a preset range before and after the inflection point, and the inboard base stations in the curved section are installed at a preset angle;

[0010] If the tunnel is in a multi-fold line form, reference base stations are arranged at each fold inflection point, and mobile base stations are arranged on the top of the tunnel in each sub-region.

[0011] The beneficial effects of the present application are: through the bidirectional signal interaction between the tag and each initial base station in the initial base station set, the original ranging results of the tag and each initial base station in the sub-area are calculated in real time, the current blocking probability and the current signal strength of the initial base station are obtained, and the current geometric dilution of precision of different initial base station combinations is obtained, the candidate base stations are selected according to the corresponding blocking probability threshold, signal strength threshold and current geometric dilution of precision threshold, the information gain of each candidate base station pair is calculated, the candidate base station pair whose information gain meets the preset requirement is selected to construct the target base station subset, the motion state space model of the tag is constructed, and the Kalman filtering process is realized. The original ranging results of all target base stations are input into the Kalman filter based on the motion state space model for processing, the NLOS time delay error is compensated through the sliding window mean value, the three-dimensional acceleration of the tag is collected according to the inertial measurement unit, the noise covariance matrix in the Kalman filtering process is dynamically adjusted, the tag positioning result and the NLOS time delay error estimation value of the current period are obtained, and the NLOS time delay error estimation value is used to update the sliding window mean value to compensate the NLOS time delay error of the next period, so as to realize the dynamic positioning of the tag. In this way, the ranging error and the calculation complexity can be reduced by fusing the ranging technology, the filtering algorithm and the intelligent base station selection strategy, the NLOS time delay error can be suppressed to realize the low-time-delay dynamic positioning, and the positioning accuracy and stability in complex environment can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 is a kind of anti non line of sight interference dynamic positioning method flow chart provided by the embodiment of the present application;

[0013] Figure 2 is a positioning coordinate diagram provided by the embodiment of the present application;

[0014] Figure 3 is another anti non line of sight interference dynamic positioning flow chart provided by the embodiment of the present application. DETAILED DESCRIPTION

[0015] The present application will be further described below in conjunction with the drawings and embodiments, it should be noted that in the description of the present application, the orientation or position relationship indicated by the terms "upper", "lower", "left", "right", "inner", "outer" and the like is based on the orientation or position relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, and cannot be understood as indicating or implying that the indicated device or element must have a particular orientation, be constructed and operated in a particular way, therefore cannot be understood as limiting the present application. The terms "first", "second", "third" and the like are only for the purpose of description, and cannot be understood as indicating or implying relative importance.

[0016] As Figure 1Fig. 1 is a schematic diagram of a dynamic positioning method against non-line-of-sight interference provided by an embodiment of the present application, comprising:

[0017] S1, calculating the original ranging result of the tag and each initial base station in real time through the bidirectional signal interaction between the tag and each initial base station in the initial base station set, wherein the initial base station set comprises all base stations in the sub-region where the tag is located.

[0018] Specifically, the tag can represent a target to be positioned, can be a terminal device or other device with signal transmission and sending, and can be used for operator positioning in special scenarios, for example, when laying cables, tunnel construction or cave operation, the operator carries the tag, and positioning is performed through signal transmission between the tag and the base station. The sub-region where the tag is located can be first divided based on the overall positioning region, and the sub-region is initially set offline, for example, every 50 m is a sub-region, and 4-6 base stations are deployed in each sub-region. Then, the sub-region is updated according to the activity adaptability of the tag, that is, the sub-region can change with the position change of the tag. The activity region of the tag is the overall region for positioning, and the six base stations closest to the tag are divided into the initial base station set in the sub-region. The sub-region where the tag is located is adaptively updated based on the change of the initial base station set.

[0019] Further, the original ranging result of the tag and the initial base station in the sub-region can be obtained, wherein the step S1 of obtaining the original ranging result comprises:

[0020] S11, obtaining ranging data through bidirectional signal interaction between the tag and each initial base station, wherein the ranging data comprises distance data and time data;

[0021] S12, removing outliers in the distance data, and performing median filtering processing on the distance data with the window size being a preset value to obtain the original ranging result, wherein the outliers include distance data exceeding a distance threshold and distance data corresponding to time data exceeding a time threshold.

[0022] Specifically, first, the bidirectional ranging is performed through the symmetric double-sided two-way ranging (SDS-TWR) technology to measure the round-trip time of the signal between the initial base station and the tag, and to calculate the relative distance observation value, that is, the original ranging result. The bidirectional ranging can reduce the influence of clock synchronization error in one-way ranging, and provide stable data support for subsequent positioning.

[0023] Further, first, the tag can send a ranging packet (containing a sending time stamp t1 and a sub-region ID j where the initial base station is located) to the initial base station in the sub-region at time t a, the initial base station receives the ranging packet at time t2, waits for a time interval t3, and then sends a ranging packet to the tag at time t4. The tag receives the ranging packet at time t5, waits for a time interval t6, and then sends a ranging packet to the initial base station at time t7. The initial base station receives the ranging packet at time t8, and the ranging packet sent by the tag at time t9. The ranging packet sent by the initial base station at time t4 and the ranging packet sent by the tag at time t7 are the same, and the ranging packet sent by the initial base station at time t8 and the ranging packet sent by the tag at time t9 are the same. After that, the response packet (response information contains timestamp t2, round = t4 - t1 , the original distance is calculated, and the distance data d round = c x t round / 2, where c is the speed of light c = 0.3 m / ns (0.3 meters per nanosecond).

[0024] According to the calculated distance data and time data, it is judged whether there is an abnormal value, if there is, the abnormal value is eliminated, the abnormal value can represent the distance data exceeding the distance threshold and the time data exceeding the time threshold, the distance threshold can represent the maximum positioning range in the overall positioning area, for example, in the tunnel, the maximum positioning range is 200m (for example, the distance between the base station at the tunnel entrance and the tunnel entrance in the opposite direction in the straight tunnel), the time threshold corresponding to the one-way propagation of the signal is 200m ÷ (0.3m / ns) = 333.3ns, so the round-trip time threshold is 2x333.3ns = 666.6ns, then the distance threshold is 200m, and the time threshold is 666.6ns, the distance data corresponding to the time data exceeding 666.6ns in the ranging data and the distance data exceeding 200m are eliminated, the distance data after eliminating the abnormal value can be obtained, of course, the time threshold and the distance threshold can be determined according to the actual situation, which is not limited here, and then the distance data after eliminating the abnormal value can be subjected to median filtering processing, the window size of the median filtering processing can be 3, that is, the median value of the distance data after eliminating the abnormal value for three times in succession is taken, and the burst noise (such as random jump caused by multipath effect) is suppressed, and the actual effective original ranging result is obtained. In this way, it can be ensured that the obtained original ranging result is real and effective, and accurate distance data is provided for subsequent positioning result acquisition, and the accuracy of the positioning result is improved.

[0025] S2, grouping the initial base stations to obtain different initial base station combinations, calculating the current geometric precision factor of different initial base station combinations, and obtaining the current shielding probability and current signal strength of each initial base station, combining the corresponding current geometric precision factor threshold, shielding probability threshold and signal strength threshold, selecting candidate base stations from the initial base station set.

[0026] ​​Specifically, due to the complexity of the positioning environment, the base station can be blocked, affecting the signal strength of the base station, so it is necessary to select the candidate base station with the best positioning effect for the tag in combination with the environmental state, and construct the target base station subset. Before selecting the target base station subset, the initial base station is first grouped, so as to facilitate the selection of candidate base stations meeting the requirements from a small range. The candidate base station can be screened according to the pre-set current geometric dilution of precision threshold (for example, 4), the blocking probability threshold (for example, 1.0) and the signal strength threshold (for example, -95dBm). The screening condition can be that the current blocking probability of the initial base station is less than the blocking probability threshold, the current signal strength is greater than the signal strength threshold, and the current geometric dilution of precision of the initial base station combination is less than the current geometric dilution of precision threshold. Based on the screening condition, the candidate base station is screened from the initial base station set. In this way, the target base station subset with accurate positioning can be obtained through multi-step screening of the initial base station-candidate base station-target base station, so as to improve the positioning accuracy of the tag in a complex environment.

[0027] When selecting the candidate base station, the current geometric dilution of precision of different initial base station combinations needs to be calculated, and the current blocking probability and the current signal strength of each initial base station need to be obtained. The specific calculation method can be as follows:

[0028] According to the real-time position coordinates of different initial base station combinations, the current geometric dilution of precision is calculated, and the specific calculation formula is as follows: Wherein, H is an observation matrix composed of the position coordinates of each initial base station in the initial base station combination, H T represents the row and column conversion matrix of H, and GDOP is the current geometric dilution of precision of the initial base station combination.

[0029] The static blocking probability of the initial base station is preset based on the installation position of the initial base station, and the static blocking probability is updated in combination with the real-time blocking state of the initial base station to obtain the current blocking probability of the initial base station.

[0030] The initial signal strength of the last preset number of periods is obtained from the real-time recorded initial signal strength, and the mean value of the initial signal strength of the last preset number of periods is determined as the current signal strength.

[0031] Specifically, the signal strength of each initial base station can be collected in real time, and the average of the initial signal strength of the last preset number of periods (for example, the last 10 periods) is taken as the current signal strength. For example, the received signal strength indication (RSSI) is collected every 50 ms, and the average of the last 10 initial signal strengths is taken as the current signal strength. It should be understood that the RSSI value of the tag reflection signal can be actively collected by the base station, or the signal strength of the surrounding base station can be measured by the tag and reported by the tag, or the signal strength can be collected by the base station and the tag together, and the signal strength of the two is combined to enhance the credibility of the initial signal strength. This is not limited here.

[0032] The current shielding probability of the initial base station can be calculated in combination with the shielding state of the initial base station and the static shielding probability. The static shielding probability can be preset based on the installation position of each initial base station. For example, when the initial base station is arranged near a cable rack, the static shielding probability of the initial base station near the cable rack is 0.3, and the static shielding probability of the initial base station arranged in an open area is 0.1. Of course, this is not limited here.

[0033] There can be movable shielding objects other than the shielding of the initial base station detection, such as moving vehicles, construction equipment, etc., and the shielding state can be reflected by the shielding probability. Therefore, it is necessary to update the shielding probability of the initial base station in real time. The static shielding probability can be updated in combination with the real-time shielding state of the initial base station to obtain the current shielding probability. The specific steps can be as follows:

[0034] If the visual sensor detects that the shielding object enters the preset range of the initial base station, the static shielding probability is updated based on a preset probability increment to obtain a first shielding probability, and the first shielding probability is the current shielding probability. When the current signal strength is lower than the preset signal strength, the current shielding probability is set as the forced shielding probability.

[0035] If the shielding object leaves the preset range of the initial base station, the second shielding probability is calculated based on the first shielding probability according to a preset probability decay rule, and the second shielding probability is the current shielding probability.

[0036] Specifically, a visual sensor can be installed in the positioning area to obtain the shielding state of the initial base station in real time. The visual sensor can be a camera, and the installation position and number are determined according to the specific situation. Here, it is not limited as long as the shielding state around the base station can be clearly collected.

[0037] When the shielding object enters the preset range of the initial base station, it means that the shielding object enters the signal path of the initial base station, for example, the shielding object enters within 2 m from the initial base station. At this time, the preset probability increment (for example, ΔP) can be used according to the preset probability increment. =+0.6) Update the static occlusion probability. That is, if the obstruction enters within 2m of the initial base station, the static occlusion probability is increased by 0.6. Taking the static occlusion probability of 0.1 as an example, if the initial base station is installed in an open area, the first occlusion probability is 0.1+0.6=0.7.

[0038] Furthermore, when the initial base station is severely obstructed, it will affect the signal quality between the initial base station and the tag. Therefore, the current signal strength can be compared with a preset signal strength indicating that the initial base station is severely obstructed to determine whether the initial base station is severely obstructed. This preset signal strength can be obtained through base station obstruction testing. Therefore, if the comparison result between the current signal strength and the preset signal strength indicates that the initial base station is severely obstructed, the current obstruction probability of the initial base station can be set to a forced obstruction probability (e.g., ...). =1.0, meaning the initial base station's forced obstruction probability is 1). As an example, the preset signal strength is -95dBm. When the current signal strength is lower than the preset signal strength (i.e., RSSI is less than -95dBm), it is determined that the initial base station is severely obstructed. In this case, the current obstruction probability can be set as the forced obstruction probability. In this situation, the line-of-sight signal strength is usually strong, greater than or equal to -85dBm, while the non-line-of-sight signal strength is attenuated to less than or equal to -95dBm. Therefore, -95dBm can be used as a critical value to distinguish between line-of-sight and non-line-of-sight signal strength. When the signal strength is lower than this critical value, it indicates that the signal is severely attenuated due to obstruction, and the positioning error increases significantly.

[0039] If the obstruction moves away from the preset range of the initial base station, it means that the obstruction of the initial base station will gradually be reduced, but there is still a possibility of obstruction. Therefore, the obstruction probability of the initial base station can be gradually reduced. The second obstruction probability is calculated based on the preset probability decay law (e.g., decaying by 0.1 every 5 seconds from the preset range of the initial base station). The second obstruction probability is the current obstruction probability, until the static obstruction probability is restored.

[0040] This allows for real-time updates of the initial base station's occlusion probability based on its real-time occlusion status.

[0041] To screen candidate base stations, it is also necessary to calculate the current geometric precision factor for different initial base station combinations. The current geometric precision factor can be used to evaluate the positioning accuracy of different initial base station combinations and can be calculated using the following formula: Where H is the observation matrix composed of the coordinates of each initial base station in the initial base station combination (in two-dimensional positioning, each row is in the form of [x]). i -x0,y i -y0]), H TThe row-column conversion matrix representing H, GDOP is the current geometric dilution of precision, the larger the current geometric dilution of precision, the larger the positioning error, so the current geometric dilution of precision threshold (for example, 4) can be set to select appropriate base stations, when GDOP>4, the positioning error increases significantly, so the initial base station combination data with GDOP>4 needs to be removed. Wherein, the initial base station combination for calculating the current geometric dilution of precision can be a random combination of the initial base station, or it can be grouped based on the initial position of the initial base station, and the nearest two or several initial base stations are selected as a group. Different groups can contain the same initial base station information, but the initial base station information will not be exactly the same, and the purpose is to select the required candidate base station, which will not be expanded here.

[0042] In this way, the current geometric dilution of precision, the current blocking probability and the current signal strength for screening candidate base stations can be obtained, and the candidate base stations are screened in combination with the parameter thresholds.

[0043] S3, grouping the candidate base stations to obtain different candidate base station pairs, calculating the information gain of different candidate base station pairs, and selecting the candidate base station pair with information gain meeting the preset requirement to construct a target base station subset.

[0044] Specifically, the candidate base station with information gain meeting the preset requirement can be selected as the target base station to construct the target base station subset.

[0045] Specifically, the following steps are included:

[0046] S31, setting the initial state covariance matrix of the candidate base station as a diagonal matrix, and the initial state covariance matrix elements include the position error, velocity error and NLOS delay error of the tag.

[0047] S32, updating the initial state covariance matrix by using the position coordinates of each candidate base station pair to obtain the updated state covariance matrix corresponding to each candidate base station pair.

[0048] S33, calculating the information gain corresponding to different candidate base station pairs according to the initial state covariance matrix and each updated state covariance matrix.

[0049] S34, using a greedy algorithm to iteratively select the candidate base station in the candidate base station pair with the maximum information gain as the target base station until the information gain increment is less than the preset increment, or the number of target base stations reaches the maximum allowed number of base stations, to obtain the target base station subset.

[0050] Specifically, first, the initial state covariance matrix Pk is constructed, and the initial state covariance matrix is set as a diagonal matrix. The initial state covariance matrix can represent the initial position error, velocity error and NLOS error of the tag, for example, the initial position error is ±1m, the velocity error is ±0.1m / s, and the NLOS delay error is ±10ns.

[0051] wherein the calculation formula of information gain is: wherein det represents the matrix determinant, represents the inverse matrix of Pk, P k represents the initial state covariance matrix at k moment, is the updated state covariance matrix after adding the position information of candidate base station pair (B i ,B j ), IG represents the information gain corresponding to candidate base station pair (B i ,B j ), and the information gain can be used to evaluate the contribution of different candidate base station pairs to the positioning accuracy of the tag. The greater the IG value is, the more effective the introduced observation data is.

[0052] wherein the calculation method of updated state covariance matrix (P k |B i ,B j ) can be: P k |B i ,B j = (P k ) -1 + H T i,j R - 1 H i,j ) -1 , H i,j is an observation matrix composed of the coordinates of candidate base stations B i and B j , and R is an observation noise covariance matrix wherein the construction method of the observation matrix can be: for three-dimensional positioning, the row vector corresponding to each candidate base station is [x i -x0,y i -y0,z i -z0], wherein (x i ,y i ,z i ) is the three-dimensional position coordinates of the candidate base station, and (x0, y0, z0) is the predicted position of the tag. This will not be described in detail here, so that the information gain corresponding to different candidate base station pairs can be calculated.

[0053] After the information gain is calculated, a greedy algorithm is used, and each time the candidate base station pair with the maximum IG is selected, until the IG increment is less than a preset increment (for example, the IG increment is less than 0.5), or the number of target base stations reaches the maximum allowed number of base stations (such as 4), the selection of the target base station is terminated, and a target base station subset is constructed. The maximum allowed number of base stations can be set after considering the positioning accuracy requirement, environmental complexity and hardware processing capacity, and this will not be specifically limited here, and it is reasonable.

[0054] It should be noted that when calculating the information gain and the current geometric dilution of precision, an observation matrix is constructed, the observation matrix constructed by the position coordinates of the initial base station combination is used to evaluate the positioning potential of the initial base station combination, and the observation matrix constructed based on the position coordinates of the candidate base station pair is used to evaluate the positioning potential of the corresponding candidate base station pair, both of which are used to evaluate the positioning accuracy of the base station. In this way, the dynamic selection from the initial base station to the target base station subset is completed, and then the motion state space model for precise positioning can be constructed based on the target base station subset.

[0055] S4, a state vector including the position coordinates of the tag, the velocity component, and the NLOS time delay error is established, a state transition equation is constructed based on the motion characteristics of the tag, and a distance observation equation is constructed based on the coordinate positions of all target base stations in the target base station subset, forming a motion state space model of the tag, wherein the state transition equation is used to describe the dynamic change relationship of the state vector, and the distance observation equation is used to associate the measured distance of the tag with the state vector.

[0056] S5, the original ranging results of the target base station subset are input into the Kalman filter based on the motion state space model for processing, the NLOS time delay error is compensated by the sliding window mean of the NLOS time delay error estimate value, and the process noise covariance matrix in the Kalman filter is dynamically adjusted according to the three-dimensional acceleration of the tag collected by the inertial measurement unit in real time, so as to obtain the tag positioning result in the current period and the NLOS time delay error estimate value, which is used to update the sliding window mean to compensate the NLOS time delay error in the next period, so as to realize the dynamic positioning of the tag.

[0057] Steps S4 and S5 are the construction of the motion state space model and the processing of the Kalman filter, and through the combination of the Kalman filter and the motion state space model, the NLOS time delay error can be effectively estimated and suppressed, the positioning error can be reduced, and the positioning accuracy in complex environments can be improved.

[0058] Among them, the design of the state vector contains the two-dimensional / three-dimensional position coordinates (x, y, z) of the tag, the velocity component (v x ,v y ,v z ) and the NLOS time delay error (δ), that is, the state vector can be X=[x, y, z, v x ,v y ,v z ,δ] TWherein, the position and the velocity are used to represent the physical motion state of the tag, and the NLOS delay error explicitly models the signal propagation delay deviation caused by the environment as a state variable, so that the NLOS delay error can be estimated as a dynamic variable, and the real-time correction is performed through the filtering algorithm to improve the adaptability to the shielding environment, and the accuracy and real-time performance of the positioning result of the target base station, wherein the two-dimensional and three-dimensional determination is based on the actual situation of the target base station and the coordinate axis.

[0059] Embodiments of the present application take uniform speed as an example, i.e. acceleration is 0, and a state transition equation is constructed for realizing state prediction, which can be x k+1 = Fx k-1 + w k Combined with the motion characteristics of the tag (uniform acceleration, variable acceleration, etc.), it is constructed taking the sampling period T=0.1s as an example, wherein the state transition matrix

[0060] F= .

[0061] The state prediction is based on the state vector (position (x, y, z), velocity (v x , v y , v z ), NLOS delay error δ) of the last moment, and the motion state at the current moment is predicted through the state transition matrix (uniform acceleration model, assuming that the acceleration is 0, the velocity is linearly transmitted, and δ is slow time-varying). The position component in the matrix is calculated by “last position + velocity × time” (for example ), the velocity component remains unchanged ( ), and δ is unchanged. In this way, the position, velocity and NLOS delay error of the tag can be recursively updated in the same framework.

[0062] Wherein the process noise w k : the size of the uncertainty of the tag motion process is described through the noise covariance matrix Q, i.e. w k , and the value will be adjusted according to the motion state of the tag: the three-dimensional acceleration of the tag is collected through the inertial measurement unit, and when the absolute value of the acceleration exceeds 0.1m / s 2 , it can be judged that the tag is in a dynamic motion state (such as moving and turning); the noise parameter position noise variance is dynamically adjusted ;

[0063] Wherein: the NLOS delay error variance (increases when dynamic), ( is the maximum value of the three-axis acceleration).

[0064] When the tag moves more violently (with greater acceleration), the noise covariance of position, velocity, and NLOS error is increased, making the filtering algorithm more sensitive to "sudden disturbances" and thus adapting to the uncertainty in dynamic environments. Conversely, when the tag is stationary or moving slowly, the noise covariance is reduced to improve the stability of positioning.

[0065] Accordingly, the distance observation equation is constructed using the raw ranging results obtained from a subset of observations of the target base station. It correlates the measured distance to the tag with the predicted location coordinates and NLOS delay error, thereby correcting the predicted state of the tag and obtaining its localization result. The raw ranging results used to construct the distance observation equation can be the raw ranging results corresponding to the target base station, ensuring greater accuracy of the measured distance obtained from the equation.

[0066] The distance observation equation for the target base station is: ,in, This represents the measured distance to the target base station at time K, (x k y k , z k (x) represents the three-dimensional coordinates of the label at time k, (x) b y b , z b () represents the fixed coordinates of the target base station. For the observed noise, the variance R = 0.1m 2 , This represents the NLOS delay error at time k.

[0067] More specifically, step S5 includes:

[0068] S51, calculate the predicted state of the tag according to the state transition equation. The predicted state includes the tag's position prediction result, velocity prediction result, and NLOS delay error prediction result.

[0069] S52, for the target base station, calculate the residual, where the formula for calculating the residual is: ,in( ) represents the predicted position of the label at time k. This represents the NLOS delay error prediction result at time k, (x b y b , z b () represents the fixed coordinates of the target base station. This represents the measured distance between the target base station and the tag at time k. Residuals represent the deviation between the predicted state and the measured result.

[0070] S53, updating the predicted state based on the residual of each target base station, outputting an updated state vector, the updated state vector including the tag positioning result of the current period and the NLOS delay error estimation value.

[0071] In this way, the measured distance of the target base station can be subtracted from the "geometric distance calculated based on the predicted position" and the "NLOS delay error prediction result" to obtain the residual.

[0072] For example, the predicted state is: (state transition equation)

[0073] Target base station coordinates:

[0074] Measured distance: (distance observation equation)

[0075] Predicted distance:

[0076] Residual:

[0077] Thus, it is shown that the predicted distance and the measured distance have a deviation of 0.2 m, and the predicted state of the tag needs to be corrected based on this deviation to obtain the optimal state estimation, i.e., the positioning result. It should be noted that the above coordinates are obtained based on a three-dimensional coordinate constructed in advance, and the origin of the three-dimensional coordinate is determined according to the situation. In this application, the tunnel is taken as an example, as shown in Figure 2 , the midpoint of the landing at the tunnel entrance can be taken as the origin, the right side of the hole facing inside can be taken as the x direction, the vertical hole plane inside the hole can be taken as the y direction, and the vertical to the ground upward can be taken as the z direction.

[0078] Through explicit modeling of the NLOS delay error, real-time updating of the NLOS delay error can be realized, and the NLOS delay error estimation value is obtained. The steps of compensating the NLOS delay error by using the sliding window average include:

[0079] A sliding window with a length of n is established to store the NLOS delay error estimation values of the last n periods;

[0080] The average of the NLOS delay error estimation values of the last n periods is taken as the compensation value of the original ranging result in the current period;

[0081] The compensation formula is:, wherein represents the original ranging result of the target base station subset after compensation at time k+1, is the original ranging result of the target base station at time k+1, and 0.3 is the speed of light, represents the average of the NLOS delay error estimation values of n periods, .

[0082] A sliding window of length n (e.g., 5 periods) can be established to store the NLOS delay error estimates of the last 5 periods , and the average value is taken as the compensation value of the next period, , the NLOS delay error of the next period is compensated based on the above compensation formula, where represents the original ranging result directly measured by the base station, and this distance contains the excess part of the signal blocked (non-line-of-sight interference), represents the average value of the NLOS delay error estimates of the last five times, which represents the time the signal is blocked (NLOS delay error unit is ns (nanosecond)) in each measurement, because the speed of light 0.3 m / ns represents that the signal can travel 0.3 m in 1 ns, so is the false distance corresponding to the time spent in this period, and the original ranging result obtained in the next period is subtracted from the false distance to obtain the compensated original ranging result, so that the original ranging result after compensation in the next period is closer to the true distance through the update and compensation of the NLOS delay error, and a more accurate positioning result can be obtained.

[0083] In summary, as shown in Figure 3 , Figure 3 is another anti-non-line-of-sight interference dynamic positioning process diagram provided by the embodiments of the present application, which can first measure the initial original ranging result between the base station and the tag through SDS-TWR two-way ranging, then obtain the current blocking probability of the base station according to the blocking state of the base station, select the target base station subset with the best positioning effect in combination with the information gain, then construct a motion state space model of the tag to update the Kalman filter, and output the positioning result of the current period and the current NLOS delay error estimate value in combination with the compensation of the NLOS error average value, wherein the positioning result can include the position coordinates and velocity components of the tag. The specific scheme of this step has been described in the foregoing, and will not be expanded here. After obtaining the positioning result, the positioning can be continued, or the operator can choose to end the positioning, i.e., to turn off the equipment or terminate the program, to end the positioning process.

[0084] In this way, by explicitly modeling the NLOS error, the traditional limitation of first removing and then compensating is broken through, and through dynamic base station selection, the dynamic estimation of the state space model combined with environmental error compensation, the robustness of ultra-wideband positioning in complex scenes can be broken through, and the accuracy and stability of ultra-wideband positioning in complex scenes can be improved. And through Kalman filtering and motion state space model, the NLOS delay error can be effectively estimated and suppressed. Compared with the traditional method, the positioning error can be reduced by 30%-50%. By fusing the blocking probability, signal strength, geometric precision factor, and information gain of the base station, the application stability of the base station can be improved, and the positioning stability in the blocking scene can be ensured.

[0085] The application also provides a base station layout method for positioning, comprising, in the case that the positioning area is a tunnel structure, the positioning area is the active area of the tag:

[0086] If the tunnel is in a straight line form, the sub-areas are divided according to a first interval, and each sub-area is arranged in a triangular space on the two side walls and the top of the tunnel;

[0087] If the tunnel is in an S-shaped form, the sub-areas are divided according to a second interval, and a third preset number of base stations are arranged within a preset distance before and after the inflection point, and the base stations inside the curve are installed at a preset angle;

[0088] If the tunnel is in a multi-fold line form, the sub-areas are divided according to a third interval, and reference base stations are arranged at each fold inflection point, and mobile base stations are arranged on the top of the tunnel in each sub-area.

[0089] Specifically, (1) straight line tunnel, equidistant triangular distribution + height compensation

[0090] Horizontal layout: the sub-areas are divided according to a first interval (such as 50 meters), and a third preset number of base stations (4 base stations, such as 1 on each side 3 meters from the ground, 1 on the top central axis, arranged in order) are arranged in a triangular shape on the two side walls of the tunnel to form a three-dimensional positioning network. In this way, under the straight line structure, the base stations are distributed in a triangular shape, which can optimize the geometric accuracy factor compared to the straight line deployment of base stations, and the top base station can reduce the shielding of horizontal signals by cable racks and other equipment to improve positioning accuracy.

[0091] (2) S-shaped tunnel, additional base stations in the curve + dynamic angle adjustment

[0092] Because of the more shielding of the S-shaped curve, the sub-areas can be divided according to a second interval (for example, 30 meters), and a third preset number (for example, 6) of base stations are arranged in each sub-area, of which a third preset number (for example, 2) are located inside the curve, that is, compared to the first layout method, 2 base stations are additionally arranged within a range of 20m before and after the inflection point of the curve, and are installed at an angle (for example, inclined at 45° to the tunnel axis) to cover the blind area of the curve, and when calculating the information gain to select the candidate base station, the inside base stations of the curve are forced to participate in the settlement, avoiding the failure of ranging due to curve shielding. It can be seen that this base station layout method can cope with the curve shielding situation and avoid the problem of inaccurate ranging results caused by curve shielding.

[0093] (3) Multi-fold line tunnel: inflection point anchoring + hierarchical networking

[0094] That is, fixed reference base stations (anchor points) are arranged at each fold inflection point, packaged with a waterproof shell, and the position coordinates of the tags obtained by the base stations are pre-calibrated by a total station to achieve an error of less than or equal to 1 cm.

[0095] The sub-regions are divided into segments by broken lines (third spacing). The third spacing is not limited and can be the actual length of the broken line segment. Each sub-region forms a hybrid network of "anchor points + mobile base stations". The mobile base stations can be dynamically deployed on the top of the tunnel as the construction progresses.

[0096] The anchor base station can use a directional antenna with the beam pointing in the direction of the broken line extension to reduce NLOS delay errors caused by corner reflections. When the angle of the broken line exceeds 90°, a multipath fusion algorithm can be enabled, combined with NLOS error compensation (δcomp) in Kalman filtering, to correct the ranging value of the base station at the inflection point in real time by 0.3×δcomp. Of course, the NLOS error compensation process is performed in each calculation. Here, it is only emphasized that NLOS error compensation can reduce the impact of the inflection point of the broken tunnel on the positioning accuracy.

[0097] It should also be noted that the computing platform for the above solution can utilize edge computing nodes (such as NVIDIA Jetson Nano, an embedded development board designed by NVIDIA specifically for edge computing and AI applications), integrating UWB base station data reception, Kalman filter calculation, and base station selection algorithms, with a processing latency of ≤50ms. Its data transmission enables tags and base stations to transmit ranging data via 5.8GHz Wi-Fi, and the deployment of relay nodes within the tunnel ensures communication coverage >99%.

[0098] This not only achieves high efficiency in data transmission and computation, but also ensures the accuracy and stability of ultra-wideband positioning results.

[0099] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A dynamic positioning method for resisting non-line-of-sight interference, characterized in that, Includes the following steps: S1, through bidirectional signal interaction between the tag and each initial base station in the initial base station set, the original ranging results between the tag and each initial base station are calculated in real time. The initial base station set includes all base stations in the sub-region where the tag is located. S2, group the initial base stations to obtain different initial base station combinations, calculate the current geometric precision factor of different initial base station combinations, and obtain the current occlusion probability and current signal strength of each initial base station. Combine the corresponding current geometric precision factor threshold, occlusion probability threshold and signal strength threshold to select candidate base stations from the initial base station set. S3, group the candidate base stations to obtain different candidate base station pairs, calculate the information gain of different candidate base station pairs, and select candidate base station pairs whose information gain meets the preset requirements to construct the target base station subset; S4. Establish a state vector including the tag's position coordinates, velocity components, and non-line-of-sight (NLOS) delay error. Construct a state transition equation based on the tag's motion characteristics and a distance observation equation based on the position coordinates of all target base stations in the target base station subset to form a motion state space model of the tag. The state transition equation is used to describe the dynamic change relationship of the state vector, and the distance observation equation is used to associate the measured distance of the tag with the state vector. S5, the original ranging results of the target base station subset are input into the Kalman filter based on the motion state space model for processing. The NLOS delay error of the tag is compensated by the sliding window mean of the NLOS delay error estimate. The process noise covariance matrix in the Kalman filter is dynamically adjusted according to the three-dimensional acceleration of the tag collected in real time by the inertial measurement unit to obtain the tag positioning result and the NLOS delay error estimate for the current period. The NLOS delay error estimate is used to update the sliding window mean to compensate for the NLOS delay error in the next period, thereby realizing the dynamic positioning of the tag.

2. The method according to claim 1, characterized in that, Step S1 includes the following steps: S11, ranging data is obtained through bidirectional signal interaction between the tag and each initial base station, the ranging data including distance data and time data; S12, remove outliers from the distance data, and perform median filtering on the distance data with outliers removed, with a window size of a preset value, to obtain the original distance measurement result. The outliers include distance data that exceeds the distance threshold and distance data corresponding to time data that exceeds the time threshold.

3. The method according to claim 1, characterized in that, The calculation of the current geometric precision factor for different initial base station combinations, and the acquisition of the current occlusion probability and current signal strength for each initial base station, includes: Based on the real-time location coordinates of different initial base station combinations, the current geometric precision factor is calculated using the following formula: Where H is the observation matrix composed of the position coordinates of each initial base station in the initial base station combination. Represents the row and column transformation matrix of H, where GDOP is the current geometric precision factor; The static occlusion probability is preset based on the installation location of the initial base station, and the static occlusion probability is updated in combination with the real-time occlusion status of the initial base station to obtain the current occlusion probability. The initial signal strength of the most recent preset number of periods is obtained from the initial signal strength recorded in real time, and the average value of the initial signal strength of the most recent preset number of periods is determined as the current signal strength.

4. The method according to claim 3, characterized in that, The step of updating the static occlusion probability by combining the real-time occlusion status of the initial base station to obtain the current occlusion probability includes: If an obstruction enters the preset range of the initial base station, the static obstruction probability is updated based on the preset probability increment to obtain a first obstruction probability. The first obstruction probability is the current obstruction probability. When the current signal strength is lower than the preset signal strength, the current obstruction probability is set as a forced obstruction probability. If the obstruction moves away from the preset range of the initial base station, then based on the first obstruction probability, a second obstruction probability is calculated according to a preset probability decay law, and the second obstruction probability is the current obstruction probability.

5. The method according to claim 1, characterized in that, Step S3 includes: S31, Set the initial state covariance matrix as a diagonal matrix, wherein the elements of the initial state covariance matrix include the tag's position error, velocity error and NLOS delay error; S32, using the position coordinates of each candidate base station pair, update the initial state covariance matrix to obtain the updated state covariance matrix corresponding to each candidate base station pair; S33, calculate the information gain of different candidate base stations based on the initial state covariance matrix and each updated state covariance matrix; S34, a greedy algorithm is used to iteratively select candidate base stations from the candidate base station pair that maximizes information gain as target base stations, until the information gain increment is less than the preset increment or the number of target base stations reaches the maximum allowed number of base stations, thus obtaining a subset of target base stations.

6. The method according to claim 5, characterized in that, The formula for calculating the information gain is: Where det represents the determinant of the matrix, P represents k The inverse matrix, P k Let the initial state covariance matrix at time k be denoted as . To join the candidate base station pair The updated state covariance matrix of ), where IG represents the candidate base station pair. The corresponding information gain.

7. The method according to claim 5, characterized in that, The expression for the distance observation equation is: ,in This represents the measured distance to the target base station at time K, (x k y k , z k (x) represents the three-dimensional coordinates of the label at time k, (x) b y b , z b () represents the fixed coordinates of the target base station. ~N(0,R) represents the observation noise covariance. This represents the NLOS delay error at time k.

8. The method according to claim 7, characterized in that, Step S5 includes: S51, calculate the predicted state of the tag according to the state transition equation, the predicted state includes the tag's position prediction result, velocity prediction result and NLOS delay error prediction result; S52, for the target base station, calculate the residual, where the formula for calculating the residual is: ,in , , ) represents the predicted position of the label at time k. This represents the NLOS delay error prediction result at time k, (x b y b , z b () represents the fixed coordinates of the target base station. This represents the measured distance between the target base station and the tag at time k. Residuals represent the deviation between the predicted state and the measured result; S53, based on the residuals of each target base station, update the predicted state and output the updated state vector, which includes the tag positioning result of the current period and the estimated value of NLOS delay error.

9. The method according to claim 8, characterized in that, The method of compensating for the NLOS delay error of the label by the sliding window mean of the NLOS delay error estimate includes: Establish a sliding window of length n to store the estimated NLOS delay error values ​​for the most recent n periods; The average of the NLOS delay error estimates over the most recent n periods is used as the compensation value for the original ranging result in the current period. The compensation formula is: ,in, This represents the original ranging result of the target base station subset after compensation at time k+1. This represents the original ranging result of the target base station at time k+1, where 0.3 represents the speed of light in meters per nanosecond. This represents the mean of the NLOS delay error estimates for the most recent n periods.

10. A base station deployment method for positioning, applied to the method described in any one of claims 1-9, characterized in that, include: When the positioning area is a tunnel structure, the positioning area is the active area of ​​the tag: If the tunnel is in a straight line, it is divided into sub-regions according to the first interval, and each sub-region is deployed in an equilateral triangle space on both sides of the tunnel wall and the top. If the tunnel is S-shaped, the sub-regions are divided according to the second spacing, and a third preset number of base stations are set within a preset distance before and after the turning point. The base stations on the inside of the curve are installed at a preset angle. If the tunnel has a multi-segment shape, then sub-regions are divided according to the third interval, a reference base station is deployed at the inflection point of each segment, and a mobile base station is deployed on the top of the tunnel within each sub-region.

Images