High-precision positioning method and system for sound indoor TDMA positioning architecture

Abstract

The present invention discloses a high-precision positioning method and system for an acoustic indoor TDMA positioning architecture. The method extracts the distance and relative motion speed information from a base station to a target in an acoustic signal, constrains the motion model of the target using a B-spline broadcast value, and then jointly estimates the target position within a single positioning cycle of the TDMA architecture based on a new particle swarm optimization algorithm. This method can not only improve the positioning accuracy of the target, but also effectively express the motion attributes and motion trajectory of the target, thereby solving the problem that the current positioning system based on the TMDA architecture cannot meet the positioning requirements of fast-moving targets and has a large positioning deviation during the positioning process.

Description

Technical Field

[0001] The present invention relates to the technical field of indoor target positioning, and in particular to a high-precision positioning method and system oriented to a sound indoor TDMA positioning architecture. Background Art

[0002] With the popularity of smart phones and smart mobile terminals (such as sweeping robots, smart watches, etc.) in daily life, location information-based services have spawned a variety of smart applications. Location information is also the cornerstone of the promotion and application of unmanned systems in the future. In indoor scenes, satellite positioning systems are greatly restricted. In order to achieve positioning in satellite positioning denied environments, domestic and foreign scholars have proposed many positioning technologies, such as those based on UWB, Bluetooth, WIF, RFID, IMU, sound and other technologies. Among them, sound-based positioning technology has attracted the attention of domestic and foreign researchers due to its high positioning accuracy, compatibility with various commercial smart mobile terminals, and low application cost.

[0003] Sound-based positioning technology uses only microphones and speakers as sensing elements, making it compatible with the widest range of devices. Since the sampling frequency of commercial smart mobile terminals is usually less than 48KHz, near-ultrasonic positioning systems (signal frequency band between 16kHz-24kHz) are the most common type. Based on the above advantages, this technology is imperceptible to the human ear, avoiding sound pollution to the environment. Common positioning system architectures include distance-based positioning architectures and angle-based and fingerprint-based positioning architectures, among which distance-based positioning architectures are more common.

[0004] The distance-based positioning architecture includes positioning base stations and positioning tags (including customized tags and smart mobile terminals, etc.). The tag position is estimated by estimating the distance between the base station and the tag. When the base station and the tag can be synchronized, the TOA three-sided positioning architecture can be used, otherwise the TDOA hyperbolic positioning architecture can be used. The accuracy and performance of the positioning system depends on the distance estimation accuracy. Since there is a relatively serious reverberation in the indoor scene, and the indoor acoustic channel is a typical fading channel with certain frequency selection characteristics, increasing the signal bandwidth can improve the ranging performance, thereby improving the positioning accuracy and stability of the system. In practical applications, it is necessary to increase the frequency domain bandwidth of the signal as much as possible.

[0005] Due to the limitations of human hearing and the sampling frequency of commercial smart mobile terminals, the frequency domain bandwidth of the sound signal is limited to between 16kHz and 24kHz. When the number of base stations is N, it is necessary to accommodate N signals in the range of 16kHz to 24kHz, which can be achieved by using the frequency division multiple access (FDMA) architecture to divide the sound channel into N sub-channels. In order to accommodate more signals, it is necessary to minimize the frequency domain bandwidth of the signal. Although the FDMA architecture can increase the positioning frequency of the system to a certain extent, the narrow signal frequency domain bandwidth will reduce the stability of the system. Therefore, in practical applications, most sound positioning systems use the time division multiple access (TDMA) architecture, or a composite architecture of TDMA and FDMA. Under the TDMA architecture, each base station broadcasts sound signals in different time slices, and each sound signal can use the entire sound frequency domain bandwidth, so it has better positioning accuracy and stability, but the positioning frequency is low. Considering that there is a serious reverberation phenomenon in the indoor space, the length of the time slice is required to be greater than the reverberation time of the environment. The system needs to poll all base stations for a single positioning, broadcast the ranging signal in turn, and then estimate the target position. Therefore, the time required for a single positioning is N time slices long.

[0006] In practical applications, for moving targets, the target is no longer a "point" during the positioning cycle, but an application trajectory. The target position estimated by the traditional positioning method has a large deviation and cannot effectively characterize the target's motion properties and trajectory. Therefore, the positioning system based on the TMDA architecture can meet the positioning requirements of stationary targets and slow-moving targets, but will inevitably introduce large positioning deviations for fast-moving targets. Summary of the invention

[0007] Based on the problems raised by the above background technology, the purpose of the present invention is to provide a high-precision positioning method and system for the sound indoor TDMA positioning architecture. By extracting the distance and relative motion speed information from the base station to the target in the acoustic signal, the motion model of the target is constrained using B-spline broadcast values, and then based on a new particle swarm optimization algorithm, the target position within a single positioning cycle of the TDMA architecture is jointly estimated. This can not only improve the positioning accuracy of the target, but also effectively express the motion attributes and motion trajectory of the target, solving the problem that the current positioning system based on the TMDA architecture cannot meet the positioning needs of fast-moving targets and has large positioning deviations during the positioning process.

[0008] The present invention is achieved through the following technical solutions:

[0009] A first aspect of the present invention provides a high-precision positioning method for a sound indoor TDMA positioning architecture, comprising the following steps:

[0010] The moving target receives the acoustic signal broadcasted by the base station in a fixed time sequence in a round-robin manner;

[0011] The acoustic signal is estimated using a time delay estimation method and a speed estimation method to obtain a measurement information set;

[0012] Using the B-spline interpolation method to perform motion calculation on the position set of the moving target, and combining the result of the motion calculation with the measurement information set to construct a nonlinear least squares problem;

[0013] The improved particle swarm optimization algorithm is used to solve the nonlinear least squares problem and obtain the position of the moving target at each moment in a fixed time series.

[0014] In the above technical solution, for a positioning scenario with N base stations, the N base stations broadcast in a fixed order and at a fixed time sequence, and the microphone of the moving target collects the broadcasted sound signals in real time. After the signal broadcast of the N base stations is completed, the collected sound signals are estimated using the delay estimation method and the speed estimation method to obtain a measurement information set.

[0015] The measurement information set includes distance measurement values ​​and relative motion speed measurement values. In the existing method for simultaneously estimating distance and speed, the distance measurement values ​​and the relative motion speed measurement values ​​are directly used for estimation, which are interfered by noise and have large errors in the distance measurement values ​​and the relative motion speed measurement values. Based on the distance measurement values, the position estimation is performed using the existing maximum relief method based on grid division, and the obtained position estimation result also has large errors.

[0016] Therefore, after obtaining the measurement information set, this method converts it into a high-dimensional nonlinear least squares problem: first, the B-spline interpolation method is used to perform motion calculation on the position of the moving target. The purpose of this motion calculation is to analyze the motion trajectory of the moving target; then the results of the motion trajectory analysis of the moving target are combined with the distance measurement value and the relative motion speed measurement value in the measurement information set to jointly construct a nonlinear least squares problem. The nonlinear least squares problem is to estimate the position of the moving target within a single positioning cycle of the TDMA architecture. By constructing a nonlinear least squares problem, the motion attributes and motion trajectory of the moving target are effectively expressed, overcoming the problem of inaccurate positioning caused by the existing estimation using only distance measurement values ​​and relative motion speed measurement values, ignoring the motion attributes and motion trajectory of the moving target.

[0017] In order to solve the nonlinear least squares problem and obtain the position information, this method uses an improved particle swarm optimization algorithm to solve it, and realizes the joint estimation of the moving target position passed by the moving target in a single cycle of the TDMA architecture to obtain the position information. It can not only improve the positioning accuracy of the moving target, but also effectively express the motion attributes and motion trajectory of the moving target.

[0018] In an optional embodiment, estimating the acoustic signal using a time delay estimation method and a speed estimation method comprises the following steps:

[0019] Extracting the flight time and relative motion speed in the time slice from the acoustic signal;

[0020] The flight time and the relative movement speed are estimated by using a time delay estimation method and a speed estimation method to obtain a distance measurement value and a relative movement speed measurement value;

[0021] The time slice, the distance measurement value and the relative motion speed measurement value are combined to obtain a measurement information set.

[0022] In an optional embodiment, the motion calculation of the position set of the moving target is performed using a B-spline interpolation method, including the following steps:

[0023] Acquire a position set of a moving target, and fit the position set using a B-spline interpolation method to obtain a motion trajectory of the moving target;

[0024] Calculating the motion trajectory of the moving target based on the second type curve integration method to obtain the moving distance of the moving target;

[0025] Calculating the movement speed of the moving target at the position based on the movement trajectory, calculating the movement speed of the moving target at the position, and obtaining a relative movement speed calculation value;

[0026] Wherein, a nonlinear least squares problem is constructed by combining the movement distance, the calculated value of the relative movement speed and the measurement information set.

[0027] In an optional embodiment, a nonlinear least squares problem is constructed by combining the movement distance, the relative movement speed calculation value and the measurement information set, including the following steps:

[0028] Calculate the distance between the moving target and the base station to obtain a distance calculation value;

[0029] Extracting distance measurement values ​​and relative motion speed measurement values ​​from the measurement information set;

[0030] A nonlinear least squares problem is constructed using the difference between the distance calculation value and the distance measurement value, the relative motion speed calculation value and the relative motion speed measurement value, and the motion distance between the two base stations.

[0031] In an optional embodiment, an improved particle swarm optimization algorithm is used to solve the nonlinear least squares problem, comprising the following steps:

[0032] The positioning space spanned by the base station coordinates is set as the search space, and a number of particle clusters are randomly scattered in the search space in a uniform distribution manner;

[0033] Using the nonlinear least squares problem as a fitness function, performing particle exploration on a plurality of particle clusters based on the fitness function to obtain a local optimum and a global optimum;

[0034] Particle swarm development is performed based on the local optimal value and the global optimal value to obtain a position estimation value.

[0035] In an optional embodiment, performing particle exploration on a plurality of particle clusters based on the fitness function comprises the following steps:

[0036] Randomly select a particle cluster as an exploration particle cluster, and calculate the fitness value of each particle in the exploration particle cluster based on the fitness function;

[0037] Performing local optimization in the exploration particle cluster based on the fitness value, and taking the position corresponding to the minimum fitness value in the exploration particle cluster as the local optimal value of the exploration particle cluster;

[0038] Randomly assigning the local optimal value to the global optimal value pointed to by the exploration particle cluster;

[0039] Based on the global optimal value, each particle in the exploration particle cluster is iteratively updated until the number of iterations exceeds the exploration iteration threshold or the change of the local optimal value and the global optimal value is less than the exploration change threshold.

[0040] In an optional embodiment, performing particle swarm development based on the local optimal value and the global optimal value comprises the following steps:

[0041] Performing local optimization among the local optimal values ​​of several particle clusters and the global optimal value to obtain a local optimal value of the particle cluster;

[0042] Randomly assign the local optimal value of the particle swarm to the global optimal value of the particle swarm pointed to by several particle clusters;

[0043] Iteratively update a plurality of particle clusters based on the global optimal value of the particle swarm until the number of iterations exceeds a development iteration threshold or the changes of the local optimal value of the particle swarm and the global optimal value of the particle swarm are less than a development change threshold.

[0044] A second aspect of the present invention provides a high-precision positioning system for a sound indoor TDMA positioning architecture, comprising:

[0045] A data acquisition module, used for a moving target to receive acoustic signals broadcasted by a base station in a fixed time sequence in a round-robin manner;

[0046] An estimation module, used to estimate the acoustic signal using a time delay estimation method and a speed estimation method to obtain a measurement information set;

[0047] A construction module, for performing motion calculation on a set of positions of a moving target using a B-spline interpolation method, and combining the result of the motion calculation with the measurement information set to construct a nonlinear least squares problem;

[0048] The solution module is used to solve the nonlinear least squares problem using an improved particle swarm optimization algorithm to obtain the position of the moving target at each moment in a fixed time series.

[0049] In an optional embodiment, the building block includes:

[0050] A fitting unit, used to obtain a position set of a moving target, and fit the position set using a B-spline interpolation method to obtain a motion trajectory of the moving target;

[0051] A motion distance calculation unit, used for calculating the motion trajectory of the motion target based on a second type curve integration method to obtain the motion distance of the motion target;

[0052] The relative motion speed calculation unit is used to calculate the motion speed of the motion target at the position based on the motion trajectory, and calculate the motion speed of the motion target at the position to obtain a relative motion speed calculation value.

[0053] In an optional embodiment, the building block further includes:

[0054] A distance calculation unit is used to calculate the distance between the moving target and the base station to obtain a distance calculation value;

[0055] An extraction unit, configured to extract a distance estimation value and a relative motion speed estimation value from the measurement information set;

[0056] The difference calculation unit is used to construct a nonlinear least squares problem using the difference between the distance calculation value and the distance estimation value, the relative motion speed calculation value and the relative motion speed estimation value, and the motion distance between two base stations.

[0057] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0058] 1. In order to solve the problem that the sound indoor positioning system based on TDMA architecture lacks high-precision positioning means for moving targets, a position estimation method based on distance and relative motion speed information is proposed. Compared with the traditional positioning method, this method can estimate the N positions that the moving target has passed during the entire positioning cycle, which can not only improve the positioning accuracy of the target, but also effectively express the motion attributes and motion trajectory of the target, and has good practical application value;

[0059] 2. The problem of estimating the position of moving targets in the sound indoor positioning system based on the TDMA architecture is converted into a high-dimensional nonlinear least squares problem. The goal is to find a set of combinations of N positions so that the least squares problem has a minimum value. First, the position combination is fitted based on the B-spline interpolation method to obtain the motion trajectory of the target under the position combination. Then, based on the trajectory, the distance from each position to the corresponding base station is calculated, the motion speed of each position is calculated, and the trajectory between each position is integrated to obtain the trajectory length. Finally, the solution to the nonlinear least squares problem is a set of combinations of N positions, so that the sum of the difference between the calculated distance and relative motion speed and the measured value and the difference between the motion trajectory lengths of each position is minimized.

[0060] 3. A new particle swarm optimization algorithm is proposed. The algorithm is designed based on a random distance star topology structure and has two stages to solve the results. In the first stage, in order to improve the activity of particles, the local optimal value of each particle cluster is randomly assigned to the global optimal value of the cluster in the particle update stage, thereby improving the search range of the particle swarm optimization algorithm in the numerical space. In the second stage, it is a standard particle swarm optimization process with a topological structure to improve the convergence speed of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:

[0062] Figure 1 It is a schematic diagram of the moving target positioning process based on TDMA architecture in the prior art;

[0063] Figure 2The overall technical framework of the TDMA positioning architecture provided in Example 1 of the present invention;

[0064] Figure 3 A comparison diagram of positioning results of three base stations in a scenario provided in Example 3 of the present invention;

[0065] Figure 4 A comparison diagram of the cumulative probability distribution of positioning errors provided in Example 3 of the present invention. DETAILED DESCRIPTION

[0066] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0067] The schematic diagram of the existing moving target positioning process based on TDMA architecture is as follows: Figure 1 As shown in FIG. 1 , when the target moves, the target continuously receives ranging acoustic signals from N base stations at different positions within a single positioning time, which brings a large deviation to the target position estimation, and the deviation is related to the target's movement speed.

[0068] Base stations 1 to 4 follow a fixed time slice t 1 ~t 4 The sound signals are broadcasted in sequence. The positioning time is T s =t 4 -t 1 During this period, the target moves along the trajectory shown in the figure, and the target position to be estimated is no longer a "point" but a "trajectory". 1 At time t, base station 1 broadcasts the measurement sound signal, records the timestamp, the distance between the target and the base station, and the actual position of the target (t 1 , r 1 , p 1 ). In the same way, record T s The relevant information from base station 2 to base station 4 during the period is (t 2 , r 2 , p 2 ), (t 3 , r 3 , p 3 ) and (t 4 , r 4 , p 4 ).

[0069] Usually, after a positioning process is completed, based on the distance location information {r 1 , r 2 , r 3 , r 4} Estimate the target position. If the positioning algorithm is f(·), then the estimated target position is So:

[0070]

[0071] Due to the use of t 1 ~t 4 The distance measurement information at the moment {r 1 , r 2 , r 3 , r 4}, even if the distance measurement information does not contain measurement errors, the estimated With {p 1 , p 2 , p 3 , p 4}There are also large deviations.

[0072] The error depends on the target's speed and the size of the time slice Δt. In practical applications, due to the lack of effective countermeasures, Directly as p 4 This greatly reduces the positioning accuracy and stability of the system based on the TDMA architecture.

[0073] To solve this problem, ShuaiCao proposed a TDMA correction method in 2022. The main idea of ​​this method is to convert the movement of the target into the movement of the base station. Specifically, the base station is moved according to the estimated target movement speed, so that the distance between the moved base station and the current target position is equal to the path length of the audio emitted by the base station to the target, and then the position is estimated using the coordinates of the moved base station to achieve error correction. The core idea of ​​this method is to estimate the target at p by measuring the distance of N base stations. N The location of the place.

[0074] However, through Figure 1 It can be seen that when the number of base stations is N, at T s In time, the target moves N times in total, and needs to pass {p 1 , p 2 ,…,p N} to describe the target's trajectory. Especially when the target moves at a high speed, only using p N Unable to accurately describe the target in T s The motion properties of a target over time.

[0075] Example 1

[0076] Based on the above defects of the existing moving target positioning based on TDMA architecture, the embodiment 1 of the present invention proposes a technical framework such as Figure 2 A high-precision positioning method for an indoor sound TDMA positioning architecture is shown, which comprises the following steps:

[0077] The moving target receives the acoustic signal broadcasted by the base station in a fixed time sequence in a round-robin manner;

[0078] The acoustic signal is estimated using a time delay estimation method and a speed estimation method to obtain a measurement information set;

[0079] Using the B-spline interpolation method to perform motion calculation on the position set of the moving target, and combining the result of the motion calculation with the measurement information set to construct a nonlinear least squares problem;

[0080] The improved particle swarm optimization algorithm is used to solve the nonlinear least squares problem and obtain the position of the moving target at each moment in a fixed time series.

[0081] It should be noted that for a positioning scenario with N base stations, the N base stations broadcast in a fixed order and at a fixed time sequence, and the microphone of the moving target collects the broadcasted acoustic signals in real time. After the signals of the N base stations are broadcast, the collected acoustic signals are estimated using the delay estimation method and the speed estimation method to obtain a measurement information set.

[0082] The measurement information set includes distance measurement values ​​and relative motion speed measurement values. In the existing method for simultaneously estimating distance and speed, the distance measurement values ​​and the relative motion speed measurement values ​​are directly used for estimation, which are interfered by noise and have large errors in the distance measurement values ​​and the relative motion speed measurement values. Based on the distance measurement values, the position estimation is performed using the existing maximum relief method based on grid division, and the obtained position estimation result also has large errors.

[0083] Therefore, after obtaining the measurement information set, this method converts it into a high-dimensional nonlinear least squares problem: first, the B-spline interpolation method is used to perform motion calculation on the position of the moving target. The purpose of this motion calculation is to analyze the motion trajectory of the moving target; then the results of the motion trajectory analysis of the moving target are combined with the distance measurement value and the relative motion speed measurement value in the measurement information set to jointly construct a nonlinear least squares problem. The nonlinear least squares problem is to estimate the position of the moving target within a single positioning cycle of the TDMA architecture. By constructing a nonlinear least squares problem, the motion attributes and motion trajectory of the moving target are effectively expressed, overcoming the problem of inaccurate positioning caused by the existing estimation using only distance measurement values ​​and relative motion speed measurement values, ignoring the motion attributes and motion trajectory of the moving target.

[0084] In order to solve the nonlinear least squares problem and obtain the position information, this method uses an improved particle swarm optimization algorithm to solve it, and realizes the joint estimation of the moving target position passed by the moving target in a single cycle of the TDMA architecture to obtain the position information. It can not only improve the positioning accuracy of the moving target, but also effectively express the motion attributes and motion trajectory of the moving target.

[0085] It should be noted that, in the embodiments of the present invention, unless otherwise specified, all vector parameters are column vectors.

[0086] Specifically, the process of a moving target receiving the acoustic signal broadcasted by the base station in a fixed time sequence is as follows:

[0087] N base stations broadcast the measured acoustic signal x(t) in a round-robin manner at intervals of Δt in a fixed order. s , the moving speed of the moving target T is v. If the reverberation caused by indoor multipath propagation is not considered, the sound signal received by the moving target T can be expressed as:

[0088]

[0089] Among them, τ k is the flight time of the acoustic signal broadcast by the kth base station to the moving target T, υ k is the relative speed of the moving target relative to the kth base station, n k (t) is the additive Gaussian noise on the path, and k = 1, 2, ..., N. Then the distance between the kth base station and the target is d k =cτ k , where c is the speed of sound in the atmosphere.

[0090] In formula (2), t k is the time slice when the kth base station broadcasts the measurement signal. In this embodiment, let t 1 =0, that is, the first base station broadcasts the sound signal at time zero, then t k =(k-1)Δt. The base station coordinates are represented by {a k}, correspondingly, at t k At the moment, the position of the moving target is p k , then the calculation formula for the relative speed is as follows:

[0091]

[0092] In an optional embodiment, estimating the acoustic signal using a time delay estimation method and a speed estimation method comprises the following steps:

[0093] Extracting the flight time and relative motion speed in the time slice from the acoustic signal;

[0094] The flight time and the relative movement speed are estimated by using a time delay estimation method and a speed estimation method to obtain a distance measurement value and a relative movement speed measurement value;

[0095] The time slice, the distance measurement value and the relative motion speed measurement value are combined to obtain a measurement information set.

[0096] It should be noted that the data used to estimate the acoustic signal in this embodiment are the flight time and relative movement speed under the time slice. Therefore, the acoustic signal is estimated by using the time delay estimation method and the speed estimation method based on the estimation of the flight time and relative movement speed in the acoustic signal according to formula (2) to obtain the distance measurement value and the relative movement speed measurement value.

[0097] The time slice, distance measurement value and relative motion speed measurement value are combined to obtain the measurement information set D within a measurement period, which is expressed as:

[0098]

[0099] In an optional embodiment, the motion calculation of the position set of the moving target is performed using a B-spline interpolation method, including the following steps:

[0100] Acquire a position set of a moving target, and fit the position set using a B-spline interpolation method to obtain a motion trajectory of the moving target;

[0101] Calculating the motion trajectory of the moving target based on the second type curve integration method to obtain the moving distance of the moving target;

[0102] Calculating the movement speed of the moving target at the position based on the movement trajectory, calculating the movement speed of the moving target at the position, and obtaining a relative movement speed calculation value;

[0103] Wherein, a nonlinear least squares problem is constructed by combining the movement distance, the calculated value of the relative movement speed and the measurement information set.

[0104] It should be noted that the core point of the present invention is the construction and application of a high-dimensional nonlinear least squares problem. First of all, as for the construction, the nonlinear least squares problem constructed by the embodiment of the present invention is as follows:

[0105]

[0106] in, is the distance calculation value, is the calculated value of relative motion speed, is the moving distance of the moving target on the moving trajectory.

[0107] It can be seen from formula (5) that the parameters required to construct the nonlinear least squares problem include: the position set of the moving target, the distance calculation value, the relative motion speed calculation value, the motion distance in the motion trajectory, and the distance measurement value and motion speed measurement value in the measurement information set.

[0108] The purpose of this step is to obtain the distance calculation value, relative motion speed calculation value, and motion distance in the motion trajectory in constructing the nonlinear least squares problem.

[0109] Specifically, the distance calculation value is calculated by the position information p of the moving target k And the base station coordinates are calculated, and the calculation process is as follows:

[0110]

[0111] For the moving distance of the moving target on the moving trajectory, firstly, based on the position set of the moving target { p k|k=1,2,…,N} is fitted using the B-spline interpolation method to obtain the motion trajectory L(x,y) of the moving target. Then, based on the second-class curve integration method, the motion trajectory is calculated to obtain the motion trajectory. The calculation process is as follows:

[0112]

[0113] At the same time, based on the motion trajectory after the collection, the derivative is taken at the position set of the moving target, and the moving target at p can be obtained. k The movement speed The calculation process is as follows:

[0114]

[0115] Then the movement speed is calculated to obtain the relative movement speed calculation value. The calculation process is as follows:

[0116]

[0117] In an optional embodiment, a nonlinear least squares problem is constructed by combining the movement distance, the relative movement speed calculation value and the measurement information set, including the following steps:

[0118] Calculate the distance between the moving target and the base station to obtain a distance calculation value;

[0119] Extracting distance measurement values ​​and relative motion speed measurement values ​​from the measurement information set;

[0120] A nonlinear least squares problem is constructed using the difference between the distance calculation value and the distance measurement value, the relative motion speed calculation value and the relative motion speed measurement value, and the motion distance between the two base stations.

[0121] In an optional embodiment, an improved particle swarm optimization algorithm is used to solve the nonlinear least squares problem, comprising the following steps:

[0122] The positioning space spanned by the base station coordinates is set as the search space, and a number of particle clusters are randomly scattered in the search space in a uniform distribution manner;

[0123] Using the nonlinear least squares problem as a fitness function, performing particle exploration on a plurality of particle clusters based on the fitness function to obtain a local optimum and a global optimum;

[0124] Particle swarm development is performed based on the local optimal value and the global optimal value to obtain a position estimation value.

[0125] It should be noted that the solution to the nonlinear least squares problem of formula (5) is currently solved by numerical solution methods such as Newton gradient descent method and Levenberg-Marquardt (LM) algorithm. However, this type of algorithm is highly sensitive to the selection of initial values, which directly affects the final calculation accuracy.

[0126] Therefore, in the present invention, an improved particle swarm optimization algorithm is used to solve the nonlinear least squares problem. The improved particle swarm optimization algorithm is designed based on a star topology structure with variable distance, and is divided into two stages: a particle exploration stage and a particle swarm development stage.

[0127] For the particle exploration phase, the goal is to find the local and global optimal values ​​of a single particle cluster;

[0128] For the particle swarm development phase, after the first phase of particle exploration in which each particle is fully explored, the purpose of this phase is to converge as quickly as possible to obtain the final calculation results.

[0129] Specifically, the positioning space spanned by the base station coordinates is set as the search space, and a number of particle clusters are randomly scattered in the search space in a uniform distribution manner. The implementation process is as follows:

[0130] First, n clusters C are randomly scattered in the search space in a uniform distribution. i and the cluster center particle is denoted as {P i |i=1,2,…,n}. Randomly set the distance of particle clusters And make the m particles in each particle cluster The distance from the central particle is And the distances between adjacent numbered particles are equal.

[0131] Among them, when the particle performs the qth iteration, the process is expressed as follows:

[0132]

[0133] In an optional embodiment, performing particle exploration on a plurality of particle clusters based on the fitness function comprises the following steps:

[0134] Randomly select a particle cluster as an exploration particle cluster, and calculate the fitness value of each particle in the exploration particle cluster based on the fitness function;

[0135] Performing local optimization in the exploration particle cluster based on the fitness value, and taking the position corresponding to the minimum fitness value in the exploration particle cluster as the local optimal value of the exploration particle cluster;

[0136] Randomly assigning the local optimal value to the global optimal value pointed to by the exploration particle cluster;

[0137] Based on the global optimal value, each particle in the exploration particle cluster is iteratively updated until the number of iterations exceeds the exploration iteration threshold or the change of the local optimal value and the global optimal value is less than the exploration change threshold.

[0138] It should be noted that, as mentioned above, the core point of the present invention is the construction and application of high-dimensional nonlinear least squares problems. The exploration of particle clusters corresponds to the application of nonlinear least squares problems. In this embodiment, the nonlinear least squares problem is used as a fitness function, and each particle calculates its fitness value based on the fitness function:

[0139] First, the distance calculation value of the particle is calculated based on formula (6); secondly, the motion trajectory of the moving target is obtained based on the B-spline interpolation method, and the motion distance of the moving target is obtained based on formula (7); then, the relative motion speed calculation value of the moving target is calculated based on formulas (8) and (9); finally, the fitness value of the particle is calculated based on the nonlinear least squares problem of formula (5).

[0140] Specifically, calculate each particle The fitness value is denoted as The position with the minimum fitness value in the particle cluster is recorded as the local optimal value of the cluster Right now:

[0141]

[0142] In order to increase the activity of particles at this stage, the local optimal value of each particle cluster is randomly assigned to the global optimal value pointed to by the particle cluster. Right now:

[0143]

[0144] Among them, randomsamplint{·} is a random sampling operator, which randomly extracts a sample from the data set.

[0145] Then, the state of each particle in the particle cluster is updated, and the update formula is as follows:

[0146]

[0147] Among them, w q-1 is the inertia weight of the particle; c 1 and c 2 They are convergence factor and social factor; and are two random numbers and follow a uniform distribution.

[0148] Iterative calculation is performed based on formula (12) until the number of iterations q exceeds the exploration iteration threshold or the change of the local optimal value or the global optimal value is less than the exploration change threshold.

[0149] It should be noted that the exploration iteration threshold and the exploration change threshold can be set by those skilled in the art according to actual projects and are not further limited in this embodiment.

[0150] In an optional embodiment, performing particle swarm development based on the local optimal value and the global optimal value comprises the following steps:

[0151] Performing local optimization among the local optimal values ​​of several particle clusters and the global optimal value to obtain a local optimal value of the particle cluster;

[0152] Randomly assign the local optimal value of the particle swarm to the global optimal value of the particle swarm pointed to by several particle clusters;

[0153] Iteratively update a plurality of particle clusters based on the global optimal value of the particle swarm until the number of iterations exceeds a development iteration threshold or the changes of the local optimal value of the particle swarm and the global optimal value of the particle swarm are less than a development change threshold.

[0154] It should be noted that in the particle exploration stage, the local optimal value and the global optimal value of each particle cluster are calculated respectively. On this basis, the local optimal value and the global optimal value of all particle clusters are integrated. Based on formula (11), the local optimization of n particle clusters is performed. The process is as follows:

[0155]

[0156] After completing the local optimization, the local optimization result is iteratively calculated based on formula (12). The process is as follows:

[0157]

[0158] Until the number of iterations q exceeds the development iteration threshold or the change of the local optimal value and the global optimal value is less than the development change threshold, the overall optimal result g is finally obtained. best .

[0159] When the iteration is complete, the optimal result is used as the position estimate.

[0160] Example 2

[0161] Embodiment 2 of the present invention is based on Embodiment 1 and takes a three-base station positioning scenario as an example to execute a high-precision positioning method for a sound indoor TDMA positioning architecture as proposed in Embodiment 1.

[0162] It should be noted that, in the embodiments of the present invention, unless otherwise specified, the units of the data described subsequently are all meters (m).

[0163] For Figure 3 The three base station positioning scenarios shown in the figure have base station coordinates of a 1 =[0,0] T 、a 1 =[20,0] T and a 1 =[20, 20] T To highlight the advantages of the algorithm of the present invention, the playback interval of the base station signal is set to 1 second, and the moving target is along Figure 3 The dotted line in the figure moves at a speed of 2 m / s. When the moving target broadcasts the measurement sound signal at the 3 base stations, its position coordinates are p 1 =[4.58, 12.51] T 、p 2 =[6.33, 12.60] T and p 3 =[7.49, 10.95] T Correspondingly, the target's moving speed at each position is υ 1 =[1.39, 1.44] T , 2 =[1.46-1.37] T and 3 =[0.98, -1.75] T .

[0164] The base station broadcast signal uses a multi-component hyperbolic frequency modulation (HFM) modulation signal with a signal frequency between 17kHz and 22kHz. Based on the existing distance and speed simultaneous estimation method, the distance and relative speed measurement values ​​are estimated. When subjected to noise interference, the distance measurement values ​​and relative speed measurement values ​​are:

[0165]

[0166] Based on the above distance measurement values, the existing maximum likelihood method based on grid division is used to estimate the position, and the position estimation result is obtained as follows:

[0167] The algorithm of Example 1 of the present invention is used, the number of particle clusters is set to 40, the number of particles in each particle cluster is set to 5, the maximum number of iterations is set to 1000, and the first stage accounts for 60% of the total iteration process. 1 =2, c 2 =2, the maximum speed of the particle is 1 and the minimum speed is 0.001.

[0168] In the positioning space spanned by the base station coordinates, 40 position combinations are randomly selected and used as the centers of 40 particle clusters. The distances of the particle clusters are randomly set, and the particles at the kth iteration are calculated based on formula (10).

[0169] For each particle, firstly, the distance calculation value is calculated based on formula (6); secondly, the motion trajectory of the moving target is obtained based on the B-spline interpolation method, and then the motion distance of the moving target is obtained based on formula (7); then, the relative motion speed calculation value of the moving target is calculated based on formulas (8) and (9); finally, the fitness value of the particle is calculated based on the nonlinear least squares problem of formula (5).

[0170] In the particle exploration phase, q≤0.6·q max , calculate the local optimal value of each particle cluster at the qth iteration according to formula (11), and then assign the global optimal value of the particle cluster according to formula (12). Then update the speed and state of each particle according to formula (13).

[0171] When q>0.6·q max , then we enter the particle swarm development phase. Compared with the particle exploration phase, the difference in the calculation process is that the global optimal assignment form is changed to formula (14). max When , the entire iteration process ends. According to formula (15), the calculation result is output as and

[0172] Compared with the true position, the total positioning error of the method proposed in the present invention is 0.158 (m), while the total positioning error of the traditional maximum likelihood estimation is 3.96 (m). It can be seen that compared with the traditional positioning method, the method proposed in the present invention can simultaneously estimate the three positions that the moving target has passed during the entire positioning cycle, which can not only improve the positioning accuracy of the target, but also effectively express the motion attributes and motion trajectory of the target, and has good practical application value.

[0173] The comparison diagram of the cumulative probability distribution of positioning errors between the method proposed in this invention and the traditional positioning method is shown in the figure below: Figure 4 As shown, from Figure 4 It can be seen that the positioning accuracy of the moving target is significantly improved by using the method proposed in the present invention.

[0174] Example 3

[0175] Embodiment 3 of the present invention provides a high-precision positioning system for a sound indoor TDMA positioning architecture, including:

[0176] A data acquisition module, used for a moving target to receive acoustic signals broadcasted by a base station in a fixed time sequence in a round-robin manner;

[0177] An estimation module, used to estimate the acoustic signal using a time delay estimation method and a speed estimation method to obtain a measurement information set;

[0178] A construction module, for performing motion calculation on a set of positions of a moving target using a B-spline interpolation method, and combining the result of the motion calculation with the measurement information set to construct a nonlinear least squares problem;

[0179] The solution module is used to solve the nonlinear least squares problem using an improved particle swarm optimization algorithm to obtain the position of the moving target at each moment in a fixed time series.

[0180] In an optional embodiment, the building block includes:

[0181] A fitting unit, used to obtain a position set of a moving target, and fit the position set using a B-spline interpolation method to obtain a motion trajectory of the moving target;

[0182] A motion distance calculation unit, used for calculating the motion trajectory of the motion target based on a second type curve integration method to obtain the motion distance of the motion target;

[0183] The relative motion speed calculation unit is used to calculate the motion speed of the motion target at the position based on the motion trajectory, and calculate the motion speed of the motion target at the position to obtain a relative motion speed calculation value.

[0184] In an optional embodiment, the building block further includes:

[0185] A distance calculation unit is used to calculate the distance between the moving target and the base station to obtain a distance calculation value;

[0186] An extraction unit, configured to extract a distance estimation value and a relative motion speed estimation value from the measurement information set;

[0187] The difference calculation unit is used to construct a nonlinear least squares problem using the difference between the distance calculation value and the distance estimation value, the relative motion speed calculation value and the relative motion speed estimation value, and the motion distance between two base stations.

[0188] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A high-precision positioning method for sound indoor TDMA positioning architecture, characterized in that: The steps include: The moving target receives the acoustic signal broadcasted by the base station in a fixed time sequence in a round-robin manner; The acoustic signal is estimated using a time delay estimation method and a speed estimation method to obtain a measurement information set; Using the B-spline interpolation method to perform motion calculation on the position set of the moving target, and combining the result of the motion calculation with the measurement information set to construct a nonlinear least squares problem; The improved particle swarm optimization algorithm is used to solve the nonlinear least squares problem and obtain the position of the moving target at each moment in a fixed time series. The B-spline interpolation method is used to calculate the motion of the position set of the moving target, including the following steps: Acquire a position set of a moving target, and fit the position set using a B-spline interpolation method to obtain a motion trajectory of the moving target; Calculating the motion trajectory of the moving target based on the second type curve integration method to obtain the moving distance of the moving target; Calculating the movement speed of the moving target at the position based on the movement trajectory, calculating the movement speed of the moving target at the position, and obtaining a relative movement speed calculation value; Wherein, a nonlinear least squares problem is constructed by combining the movement distance, the calculated value of the relative movement speed and the measurement information set.

2. The high-precision positioning method for sound indoor TDMA positioning architecture according to claim 1, characterized in that: The acoustic signal is estimated using a time delay estimation method and a speed estimation method, comprising the following steps: Extracting the flight time and relative motion speed in the time slice from the acoustic signal; The flight time and the relative movement speed are estimated by using a time delay estimation method and a speed estimation method to obtain a distance measurement value and a relative movement speed measurement value; The time slice, the distance measurement value and the relative motion speed measurement value are combined to obtain a measurement information set.

3. The high-precision positioning method for sound indoor TDMA positioning architecture according to claim 1, characterized in that: Combining the movement distance, the relative movement speed calculation value and the measurement information set to construct a nonlinear least squares problem includes the following steps: Calculate the distance between the moving target and the base station to obtain a distance calculation value; Extracting distance measurement values ​​and relative motion speed measurement values ​​from the measurement information set; A nonlinear least squares problem is constructed using the difference between the distance calculation value and the distance measurement value, the relative motion speed calculation value and the relative motion speed measurement value, and the motion distance between the two base stations.

4. The high-precision positioning method for sound indoor TDMA positioning architecture according to claim 1, characterized in that: The improved particle swarm optimization algorithm is used to solve the nonlinear least squares problem, which includes the following steps: The positioning space spanned by the base station coordinates is set as the search space, and a number of particle clusters are randomly scattered in the search space in a uniform distribution manner; Using the nonlinear least squares problem as a fitness function, performing particle exploration on a plurality of particle clusters based on the fitness function to obtain a local optimum and a global optimum; Particle swarm development is performed based on the local optimal value and the global optimal value to obtain a position estimation value.

5. The high-precision positioning method for sound indoor TDMA positioning architecture according to claim 4, characterized in that: Based on the fitness function, particle exploration is performed on a plurality of particle clusters, including the following steps: Randomly select a particle cluster as an exploration particle cluster, and calculate the fitness value of each particle in the exploration particle cluster based on the fitness function; Performing local optimization in the exploration particle cluster based on the fitness value, and taking the position corresponding to the minimum fitness value in the exploration particle cluster as the local optimal value of the exploration particle cluster; Randomly assigning the local optimal value to the global optimal value pointed to by the exploration particle cluster; Based on the global optimal value, each particle in the exploration particle cluster is iteratively updated until the number of iterations exceeds the exploration iteration threshold or the change of the local optimal value and the global optimal value is less than the exploration change threshold.

6. The high-precision positioning method for sound indoor TDMA positioning architecture according to claim 4, characterized in that: The particle swarm development is performed based on the local optimal value and the global optimal value, comprising the following steps: Performing local optimization among the local optimal values ​​of several particle clusters and the global optimal value to obtain a local optimal value of the particle cluster; Randomly assign the local optimal value of the particle swarm to the global optimal value of the particle swarm pointed to by several particle clusters; Iteratively update a plurality of particle clusters based on the global optimal value of the particle swarm until the number of iterations exceeds a development iteration threshold or the changes of the local optimal value of the particle swarm and the global optimal value of the particle swarm are less than a development change threshold.

7. A high-precision positioning system for sound indoor TDMA positioning architecture, characterized in that: include: A data acquisition module, used for a moving target to receive acoustic signals broadcasted by a base station in a fixed time sequence in a round-robin manner; An estimation module, used to estimate the acoustic signal using a time delay estimation method and a speed estimation method to obtain a measurement information set; A construction module, for performing motion calculation on a set of positions of a moving target using a B-spline interpolation method, and combining the result of the motion calculation with the measurement information set to construct a nonlinear least squares problem; A solution module is used to solve the nonlinear least squares problem using an improved particle swarm optimization algorithm to obtain the position of the moving target at each moment in a fixed time series; The building blocks include: A fitting unit, used to obtain a position set of a moving target, and fit the position set using a B-spline interpolation method to obtain a motion trajectory of the moving target; A motion distance calculation unit, used for calculating the motion trajectory of the motion target based on a second type curve integration method to obtain the motion distance of the motion target; The relative motion speed calculation unit is used to calculate the motion speed of the motion target at the position based on the motion trajectory, and calculate the motion speed of the motion target at the position to obtain a relative motion speed calculation value.

8. The high-precision positioning system for sound indoor TDMA positioning architecture according to claim 7, characterized in that: The building blocks also include: A distance calculation unit is used to calculate the distance between the moving target and the base station to obtain a distance calculation value; An extraction unit, configured to extract a distance estimation value and a relative motion speed estimation value from the measurement information set; The difference calculation unit is used to construct a nonlinear least squares problem using the difference between the distance calculation value and the distance estimation value, the relative motion speed calculation value and the relative motion speed estimation value, and the motion distance between two base stations.

Images