An industrial mobile target hybrid optimization wireless positioning method under uncertain measurement

Abstract

The application discloses a kind of industrial mobile target mixed optimization wireless positioning method under uncertain measurement, comprising the following steps: the model of wireless signal to industrial mobile target position under uncertain measurement is established, adaptive threshold is designed to enrich particle diversity to carry out one filtering estimation, in order to reduce the cumulative error caused by long-time filtering estimation, the positioning accuracy optimization problem is converted into iterative search optimization problem with one filtering estimation as search initial value, the problem of positioning accuracy decline or even filtering accuracy divergence is avoided.The present application considers the uncertain measurement of wireless ranging noise and anchor node calibration error in industrial environment, combines the advantages of filtering estimation and iterative search to form a hybrid optimization positioning method, which improves the wireless distributed positioning accuracy of industrial mobile target.

Description

Technical Field

[0001] This invention belongs to the technical field of wireless positioning and tracking of industrial mobile targets, specifically involving a hybrid optimization wireless positioning method for industrial mobile targets under uncertain measurement conditions. Background Technology

[0002] With numerous users needing to know their location, wireless sensor network (WSN) positioning has become a crucial driving force for various location-based services (LoCPS) applications. Regardless of the type of data collected and transmitted by sensors, it must correspond to the location of objects to explain "what specific event happened at what location." Examples include assisting smart manufacturing companies in scheduling decisions, improving the safety of workers in coal mines, using electronic fences for workers in hazardous chemical enterprises, and strengthening smart city construction, including smart transportation and smart warehousing. In a wide range of industrial applications, the scope of location-based applications extends beyond our imagination, from single-target tracking to multi-target collaboration. Thanks to the ubiquitous sensing and communication capabilities of wireless nodes, WSNs, composed of many wireless nodes connected via multi-hops, can map the physical world to the network world, providing unique opportunities for information interconnection between things. WSNs can provide ranging methods such as distance measurement, angle measurement, and multi-hop counting, thus forming various ranging-based positioning algorithms. Many uncertainties, such as node deployment strategies, wireless ranging, and environmental noise, lead to instability in wireless positioning performance.

[0003] Many scholars have conducted research on distributed computation, filtering estimation, and intelligent optimization. However, simple distributed computation is easily affected by noise, resulting in large positioning errors at some sampling points. Simple filtering estimation has high accuracy in the short term but inevitably accumulates errors over long flight times. Simple intelligent filtering estimation is prone to getting trapped in local optima or has a slow convergence speed. In summary, existing technologies have the following problems:

[0004] 1) Due to the influence of noise measurement and network deployment, wireless sensor network positioning has inconsistent positioning accuracy, resulting in high accuracy or large error in some areas of interest. It is necessary to incorporate uncertain measurement values ​​into the solution model.

[0005] 2) Particle filters are susceptible to measurement noise, resulting in accumulated errors over long periods. Furthermore, new observation data during the particle filter sampling phase requires the calculation of the important weights of the entire sequence, necessitating the resolution of the error divergence problem in filter estimation.

[0006] 3) In order to reduce the cumulative error of filter estimation over time, it is necessary to combine filter estimation with intelligent optimization algorithm to transform the positioning accuracy optimization problem into an iterative search problem, so as to avoid getting trapped in local optima and improve the wireless positioning performance of industrial mobile targets. Summary of the Invention

[0007] Objective: This invention aims to provide a hybrid optimization wireless positioning method for industrial mobile targets under uncertain measurements. It establishes a model of the wireless signal's position to the industrial mobile target under uncertain measurements, designs an adaptive threshold to enrich particle diversity for primary filtering estimation, and transforms the positioning accuracy optimization problem into an iterative search optimization problem by using the primary filtering estimate as the initial search value. This avoids the problems of decreased positioning accuracy or even divergence in filtering accuracy. This invention considers uncertain measurements, including wireless ranging noise and anchor node calibration errors in industrial environments, and combines the advantages of filtering estimation and iterative search to form a hybrid optimization positioning method. This improves the accuracy of wireless distributed positioning of industrial mobile targets and solves the problems of inconsistency in accuracy due to the uncertainty of wireless measurements and the positioning calculation process.

[0008] Technical solution: A hybrid optimization wireless positioning method for industrial moving targets under uncertain measurement, comprising the following steps:

[0009] S1. Based on uncertain measurements in the industrial environment, establish the measurement equation and state equation of the moving target, and design an adaptive threshold to enrich particle diversity for a first-order filtering estimation;

[0010] S2. Using the first-order filtered estimate as the initial search value, the positioning accuracy optimization problem is transformed into an unconstrained optimization problem, and the unconstrained optimization problem is solved by iterative search.

[0011] S3. Improve the accuracy of wireless distributed positioning for industrial mobile targets by using a hybrid optimization wireless positioning method formed by filtering estimation and iterative search.

[0012] Furthermore, step S1, which establishes the measurement equation and state equation for the moving target based on uncertain measurements in the industrial environment, includes the following steps:

[0013] An industrial wireless sensor network consists of M anchor nodes and N mobile nodes, where the set of M anchor nodes is an. s =[a1,a2,…,a M ] T Where the theoretical coordinates of the anchor node are a i for i∈[1,M]; considering the error in the anchor node calibration process, it is represented as Δan s =[Δa1,Δa2,…,Δa M ] T actual coordinates of anchor nodes It can be modeled as a i +Δa i , where Δa i =[Δxi ,Δy i ,Δz i ] T The set formed by N moving nodes is mn s =[m1,m2,…,m N ] T The nodes are installed on N industrial mobile target fuselages, where the theoretical coordinates of the mobile nodes are m. j For [x j ,y j ,z j ] T , j∈[1,N]; move node m j With the set of anchor nodes an s Geometric distance difference It can be determined by the time difference of arrival. Multiplied by the speed of wireless signal propagation v c Received, among which ||·|| represents the L2 norm; considering the wireless ranging error, it is expressed as... Actual geometric distance difference The model is At time k, establish the measurement equation for the industrial moving target:

[0014] Z j (k)=H(X j (k))+v j (k)

[0015] In the formula: v j (k) represents the ranging noise; at time k, the state equation for the industrial moving target is established as X. j (k)=T j X j (k-1)+Γ j u j (k-1), where T j Γ is the state transition matrix; j The noise matrix; u j (k-1) represents process noise;

[0016] Furthermore, step S1, designing an adaptive threshold to enrich particle diversity for a single filtering estimation, specifically includes the following steps: Filtering estimation is performed based on the established measurement equation and state equation. Due to uncertainties in industrial measurements, including wireless ranging errors and anchor node calibration errors, the target positioning performance based on particle filtering degrades. After multiple resampling cycles, particle samples degenerate, and a few parent particles with larger weights cannot completely cover the posterior probability distribution area. An adaptive threshold η is designed to enrich particle diversity.

[0017]

[0018] In the formula: η min and η max These are the minimum threshold and the maximum threshold, respectively; K η S is the scaling factor; S is the number of particles; For particle set The standard deviation; the target state estimation equation at time k based on low-weight particle adaptive adjustment is:

[0019]

[0020] In the formula: For normalized weights, based on the state estimation equation Coordinates to be estimated for industrial moving target actual coordinates of anchor nodes First-order filtered estimate of the geometric distance between for:

[0021]

[0022] In the formula: Coordinates to be estimated for industrial moving target actual coordinates of anchor node The first-order filtered estimate of the geometric distance between them is

[0023] Furthermore, step S2 transforms the positioning accuracy optimization problem into an unconstrained optimization problem, including the following steps:

[0024] The first filter estimate is Accumulated errors occur over time, necessitating improved positioning accuracy, taking into account the existing wireless ranging errors. And anchor node calibration error Δan s =[Δa1,Δa2,…,Δa M ] T Coordinates of an industrial moving target to be estimated The positioning accuracy optimization problem is transformed into an unconstrained optimization problem.

[0025]

[0026] In the formula:

[0027]

[0028]

[0029] Furthermore, step S2 employs an iterative search method to solve the unconstrained optimization problem, with the industrial movement target set mn being used when the number of iterations is 0. s First-order filtered estimate As the initial value X0 for the iterative search, the entire search space is divided into a global search in the early stage of iteration and a local search in the later stage of iteration. The update equation for the industrial moving target is:

[0030]

[0031] In the formula: X best The objective function is f(X) t The variable corresponding to the initial global search minimum is given by f(X) = (AX - b). T (HO -1 H T ) -1 (AX-b); t max The maximum number of iterations; C1 and C2 are random values ​​between 0 and 1; X max and X min Let be the lower and upper bounds of the search space, respectively. Then the iterative search update matrix is:

[0032] D t+1 ΔG t =X t+1 -X t

[0033] In the formula: ΔG t =G t+1 -G t G t The objective function f(X) represents t The first derivative at point D, since D t+1 =D t +ΔD t and X t+1 =X t +ΔX t The iterative search update matrix can be transformed into

[0034] ΔD t ΔG t =ΔX t -D t ΔG t

[0035] In the formula: Unconstrained optimization problems can be solved through iterative search.

[0036] X t+1 =X t +τ t S t

[0037] In the formula: S t The search direction is defined as -D t G t , τ t To accelerate convergence by adaptive step size, if |G t If the error threshold is met, the iterative search stops; otherwise, the iterative search continues, t = t + 1.

[0038] Beneficial effects:

[0039] 1. Based on uncertain measurements in the industrial environment, establish measurement equations and state equations for estimating the position of industrial moving targets, design adaptive thresholds to enrich particle diversity, and perform a first-pass filtering estimation of industrial moving targets.

[0040] 2. Considering the cumulative error caused by long-term filtering estimation, the first filtering estimation is used as the initial value for the search. Wireless ranging error and anchor node calibration error are introduced to transform the wireless positioning accuracy optimization problem into an iterative search optimization problem, thereby further improving the wireless positioning performance of industrial mobile targets. Attached Figure Description

[0041] Figure 1 This is a deployment diagram of a hybrid optimization wireless positioning method for industrial mobile targets under uncertain measurement conditions, according to the present invention.

[0042] Figure 2 This is a schematic diagram of a hybrid optimization wireless positioning method for industrial mobile targets under uncertain measurement conditions, according to the present invention. Detailed Implementation

[0043] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0044] A hybrid optimization wireless positioning method for industrial moving targets under uncertain measurement conditions includes the following steps:

[0045] S1. Based on uncertain measurements in the industrial environment, establish the measurement equation and state equation of the moving target, and design an adaptive threshold to enrich particle diversity for a first-order filtering estimation;

[0046] S2. Using the first-order filtered estimate as the initial search value, the positioning accuracy optimization problem is transformed into an unconstrained optimization problem, and the unconstrained optimization problem is solved by iterative search.

[0047] S3. Improve the accuracy of wireless distributed positioning for industrial mobile targets by using a hybrid optimization wireless positioning method formed by filtering estimation and iterative search.

[0048] Furthermore, step S1, which establishes the measurement equation and state equation for the moving target based on uncertain measurements in the industrial environment, includes the following steps:

[0049] like Figure 1 As shown, the industrial environment wireless sensor network consists of M anchor nodes and N mobile nodes, with the set of M anchor nodes being an. s =[a1,a2,…,a M ] T Where the theoretical coordinates of the anchor node are a i for i∈[1,M]; considering the error in the anchor node calibration process, it is represented as Δan s =[Δa1,Δa2,…,Δa M ] T actual coordinates of anchor nodes It can be modeled as a i +Δa i , where Δa i =[Δx i ,Δy i ,Δz i ] T The set formed by N moving nodes is mn s =[m1,m2,…,m N ] T The nodes are installed on N industrial mobile target fuselages, where the theoretical coordinates of the mobile nodes are m. j For [x j ,y j ,z j ] T , j∈[1,N]; move node m j With the set of anchor nodes an s Geometric distance difference It can be determined by the time difference of arrival. Multiplied by the speed of wireless signal propagation v c Received, among which ||·|| represents the L2 norm; considering the wireless ranging error, it is expressed as... Actual geometric distance difference The model is At time k, establish the measurement equation for the industrial moving target:

[0050] Z j (k)=H(X j (k))+v j (k)

[0051] In the formula: v j (k) represents the ranging noise; at time k, the state equation for the industrial moving target is established as X. j (k)=T j X j (k-1)+Γ j uj (k-1), where T j Γ is the state transition matrix; j The noise matrix; u j (k-1) represents process noise;

[0052] Furthermore, step S1, designing an adaptive threshold to enrich particle diversity for a single filtering estimation, specifically includes the following steps: Filtering estimation is performed based on the established measurement equation and state equation. Due to uncertainties in industrial measurements, including wireless ranging errors and anchor node calibration errors, the target positioning performance based on particle filtering degrades. After multiple resampling cycles, particle samples degenerate, and a few parent particles with larger weights cannot completely cover the posterior probability distribution area. An adaptive threshold η is designed to enrich particle diversity.

[0053]

[0054] In the formula: η min and η max These are the minimum threshold and the maximum threshold, respectively; K η S is the scaling factor; S is the number of particles; For particle set The standard deviation; the target state estimation equation at time k based on low-weight particle adaptive adjustment is:

[0055]

[0056] In the formula: For normalized weights, based on the state estimation equation Coordinates to be estimated for industrial moving target actual coordinates of anchor node First-order filtered estimate of the geometric distance between for:

[0057]

[0058] In the formula: Coordinates to be estimated for industrial moving target actual coordinates of anchor node The first-order filtered estimate of the geometric distance between them is

[0059] Furthermore, step S2 transforms the positioning accuracy optimization problem into an unconstrained optimization problem, including the following steps:

[0060] The first filter estimate is Accumulated errors occur over time, necessitating improved positioning accuracy, taking into account the existing wireless ranging errors. And anchor node calibration error Δan s=[Δa1,Δa2,…,Δa M ] T Coordinates of an industrial moving target to be estimated The positioning accuracy optimization problem is transformed into an unconstrained optimization problem.

[0061]

[0062] In the formula:

[0063]

[0064]

[0065] Furthermore, step S2 employs an iterative search method to solve the unconstrained optimization problem, with the industrial movement target set mn being used when the number of iterations is 0. s First-order filtered estimate As the initial value X0 for the iterative search, the entire search space is divided into a global search in the early stage of iteration and a local search in the later stage of iteration. The update equation for the industrial moving target is:

[0066]

[0067] In the formula: X best The objective function is f(X) t The variable corresponding to the initial global search minimum is given by f(X) = (AX - b). T (HO -1 H T ) -1 (AX-b); t max The maximum number of iterations; C1 and C2 are random values ​​between 0 and 1; X max and X min Let be the lower and upper bounds of the search space, respectively. Then the iterative search update matrix is:

[0068] D t+1 ΔG t =X t+1 -X t

[0069] In the formula: ΔG t =G t+1 -G t G t The objective function f(X) represents t The first derivative at point D, since D t+1 =D t +ΔD t and X t+1 =X t +ΔX tThe iterative search update matrix can be transformed into

[0070] ΔD t ΔG t =ΔX t -D t ΔG t

[0071] In the formula: Unconstrained optimization problems can be solved through iterative search.

[0072] X t+1 =X t +τ t S t

[0073] In the formula: S t The search direction is defined as -D t G t , τ t To accelerate convergence by adaptive step size, if |G t If the error threshold is met, the iterative search stops; otherwise, the iterative search continues, t = t + 1.

[0074] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A hybrid optimization wireless positioning method for industrial moving targets under uncertain measurement, characterized in that, Includes the following steps: S1. Based on uncertain measurements in the industrial environment, establish the measurement equation and state equation of the moving target, and design an adaptive threshold to enrich particle diversity for a first-order filtering estimation; S2. Using the first filtering estimate as the initial search value, the positioning accuracy optimization problem is transformed into an unconstrained optimization problem, and the unconstrained optimization problem is solved by iterative search method; S3. Improve the accuracy of wireless distributed positioning for industrial mobile targets by using a hybrid optimization wireless positioning method formed by filtering estimation and iterative search; Step S1, establishing the measurement equation and state equation for the moving target based on uncertain measurements in the industrial environment, includes the following steps: Industrial environment wireless sensor network Anchor nodes and It consists of mobile nodes. The set formed by the anchor nodes is Theoretical coordinates of anchor nodes for , Considering the errors in the anchor node calibration process, it is expressed as follows: actual coordinates of anchor nodes It can be modeled as ,in ; The set formed by the mobile nodes is Installed separately An industrial mobile target fuselage, where the theoretical coordinates of the moving nodes are... for , Mobile node With anchor node set Geometric distance difference Due to the time difference of arrival Multiplied by the speed of wireless signal propagation Received, among which , The L2 norm is used; considering the wireless ranging error, it is expressed as... The actual geometric distance difference The model is ;exist At any given moment, establish the measurement equation for industrial moving targets: In the formula: ; For ranging noise; in At time t, the state equation for the industrial moving target is established as follows: ,in This is the state transition matrix; This is the noise matrix; This is process noise; Step S1, designing an adaptive threshold to enrich particle diversity for a single filtering estimation, specifically includes the following steps: designing an adaptive threshold... To enrich particle diversity: In the formula: and These are the minimum threshold and the maximum threshold, respectively. It is a scaling factor; The number of particles; For particle set Standard deviation; adaptive adjustment based on low-weight particles The target state estimation equation at time t is: In the formula: For normalized weights, based on the state estimation equation Coordinates of an industrial moving target to be estimated actual coordinates of anchor node First-order filtered estimate of the geometric distance between for: In the formula: Coordinates of an industrial moving target to be estimated actual coordinates of anchor node The first-order filtered estimate of the geometric distance between them is ; Step S2 transforms the positioning accuracy optimization problem into an unconstrained optimization problem, including the following steps: The first filter estimate is Accumulated errors occur over time, necessitating improved positioning accuracy, taking into account the existing wireless ranging errors. and anchor node calibration error Coordinates of an industrial moving target to be estimated The positioning accuracy optimization problem is transformed into an unconstrained optimization problem. In the formula: ; ; ； ； ； ； Step S2 uses an iterative search method to solve the unconstrained optimization problem. When the number of iterations is 0, the industrial moving target set is used. First-order filtered estimate As initial values ​​for iterative search The entire search space is divided into a global search in the early stage of iteration and a local search in the later stage of iteration. The update equation for the industrial moving target is: In the formula: For the objective function The variable corresponding to the minimum value in the initial global search, where ; This represents the maximum number of iterations. and A random value between 0 and 1; and Let be the lower and upper bounds of the search space, respectively. Then the iterative search update matrix is: In the formula: , Describe the objective function The first derivative at that point, since and The iterative search update matrix can be transformed into In the formula: Then, the unconstrained optimization problem can be solved by iterative search. In the formula: For the search direction, defined as , To accelerate convergence by adaptive step size, if If the error threshold is met, the iterative search stops; otherwise, the iterative search continues. .

Images