CN105357753A - Multimode fusion recursion iteration based indoor positioning method - Google Patents

Abstract

The invention discloses a multimode fusion recursion iteration based indoor positioning method. The multimode fusion recursion iteration based indoor positioning method comprises the steps of obtaining distances from a to-be-measured point to multiple interference points based on a relationship model between a RSSI (Received Signal Strength Indicator) and distances by utilizing the RSSI, and obtaining the coordinate of the to-be-measured point by using a least square method; measuring a value of a moving object inertial sensor while measuring the RSSI and obtaining coordinate values of the object by utilizing an acceleration value integral operation and combining with angle information of a gyroscope; filtering the two obtained coordinate values and obtaining relatively smooth measuring data through a local averaging method; obtaining fused coordinates by adopting a weighted fusing method for the filtered data, replacing the original inertial sensor coordinate values by the fused coordinates, performing integral operation again and fusing the obtained coordinates with the RSSI coordinates; and circulating for multiple times to obtain finally recursion and iteration coordinate values. According to the multimode fusion recursion iteration based indoor positioning method, a measured data error is reduced, and influence from too loud environmental noise to the measuring result is avoided.

Description

A kind of indoor orientation method based on multimodality fusion recursive iteration

Technical field

The present invention relates to wireless location system design, particularly a kind of indoor orientation method based on multimodality fusion recursive iteration.

Background technology

Along with the fast development of Internet of Things, the demand of people to position & navigation increases day by day, especially in the indoor environment of relative complex, as hospital, parking lot, fire-fighting district, in the environment such as underground coal mine, usually will understand exact position and the activity situation of personnel in time, guarantee the safety of personnel, this just needs higher positioning precision.Global positioning system (GPS) is navigation system comparatively ripe at present, meet the demand of outdoor positioning and navigation, but it cannot penetrate wall, therefore can not be applied to indoor environment.The indoor positioning technologies proposed at present has WiFi, bluetooth, radio-frequency (RF) identification (RFID), ZigBee etc., and received signal strength (RSSI) value utilizing location node to receive reference node is carried out algorithm for design and positioned.And due to the impact of various barrier in indoor environment, wireless signal can produce serious multi-path jamming in transmitting procedure, thus cause the accuracy of DATA REASONING to reduce, locating effect is also not satisfactory.

Summary of the invention

The present invention is directed to the deficiencies in the prior art, propose a kind of indoor orientation method based on multimodality fusion recursive iteration.

The present invention is a kind of indoor orientation method based on multimodality fusion recursive iteration, specifically comprises the following steps:

Step 1, utilizes the received signal strength RSSI value received to calculate the distance d of tested point to reference node _m,

$d_m={10}^{\frac{-{\mathrm{RSSI}}_m-A}{10\·n}}$

Wherein, A represents the signal strength signal intensity that signal receives at distance 1m place, and n represents propagation factor;

Step 2, obtains least square solution according to least square method, that is:

$\widehat{\mathrm{\θ}}={\left(Q^{T}Q\right)}^{-1}{Q}^{T}b$

In formula, $Q=\left[\begin{array}{cccc}-2a_1& -2b_1& -2c_1& 1\\ -2a_2& -2b_2& -2c_1& 1\\ ...& ...& ...& ...\\ -2a_m& -2b_m& -2c_m& 1\end{array}\right],\mathrm{\θ}=\left[\begin{array}{c}x\\ y\\ z\\ {R_m}^2\end{array}\right],b=\left[\begin{array}{c}{r_1}^2\\ {r_2}^2\\ ...\\ {r_m}^2\end{array}\right],$

r _m ²＝d _m ²-(a _m ²+b _m ²+c _m ²)

Wherein, (a _m, b _m, c _m) be the coordinate of reference node, the coordinate that (x, y, z) is tested point, R _m ²represent the quadratic sum of tested point coordinate, r _m ²represent the quadratic sum of tested point to reference point distance and the difference of reference point coordinate quadratic sum, m represents reference node number;

Step 3, measures inertial sensor accelerometer and gyrostatic data, carries out integral operation, that is: while measurement received signal strength

s'＝s·cosθ

In formula, s=∫ vdt, v=∫ adt

Wherein, a is accelerometer accekeration in one direction on object under test, and θ is gyroscope angle value in one direction on object under test, and s represents the displacement of object under test, and v represents the speed of object under test;

Step 4, adopt moving average filter algorithm to carry out filtering to the RSSI received and accelerometer, gyrostatic measurement data, Filtering Formula is as follows:

${\stackrel{\‾}{y}}_j=\frac{1}{M}\underset{i=j-M+1}{\overset{k}{\Σ}}{y}_i$

In formula, M is sliding window length, y _ibe i-th and measure numerical value; Wherein k>=M, k are integer;

Step 5, calculates coordinate figure and inertial sensor coordinate figure to RSSI and is weighted fusion and obtains merging coordinate, that is:

d _τ＝βR _τ+(1-β)D _τ

In formula, $\mathrm{\β}=\frac{{\mathrm{\β}}_2}{{\mathrm{\β}}_1+{\mathrm{\β}}_2},{\mathrm{\β}}_1=\frac{1}{\mathrm{\α}F}\underset{i=1}{\overset{\mathrm{\α}F}{\Σ}}({\left(x_{{\mathrm{\α}}_i}-{\stackrel{\‾}{x}}_{\mathrm{\α}}\right)}^2+{\left(y_{{\mathrm{\α}}_i}-{\stackrel{\‾}{y}}_{\mathrm{\α}}\right)}^2+{\left(z_{{\mathrm{\α}}_i}-{\stackrel{\‾}{z}}_{\mathrm{\α}}\right)}^2),$

$\begin{array}{c}{\mathrm{\β}}_2=\frac{1}{\mathrm{\α}F}\underset{i=1}{\overset{\mathrm{\α}F}{\Σ}}({\left(x_{{\mathrm{\β}}_i}-{\stackrel{\‾}{x}}_{\mathrm{\β}}\right)}^2+{\left(y_{{\mathrm{\β}}_i}-{\stackrel{\‾}{y}}_{\mathrm{\β}}\right)}^2+{\left(z_{{\mathrm{\β}}_i}-{\stackrel{\‾}{z}}_{\mathrm{\β}}\right)}^2),{\stackrel{\‾}{x}}_{\mathrm{\α}}=\frac{1}{\mathrm{\α}F}\underset{i=1}{\overset{\mathrm{\α}F}{\Σ}}{x}_{{\mathrm{\α}}_i},{\stackrel{\‾}{y}}_{\mathrm{\α}}=\frac{1}{\mathrm{\α}F}\underset{i=1}{\overset{\mathrm{\α}F}{\Σ}}{y}_{{\mathrm{\α}}_i},\\ {\stackrel{\‾}{z}}_{\mathrm{\α}}=\frac{1}{\mathrm{\α}F}\underset{i=1}{\overset{\mathrm{\α}F}{\Σ}}{z}_{{\mathrm{\α}}_i},{\stackrel{\‾}{x}}_{\mathrm{\β}}=\frac{1}{\mathrm{\α}F}\underset{i=1}{\overset{\mathrm{\α}F}{\Σ}}{x}_{{\mathrm{\β}}_i},{\stackrel{\‾}{y}}_{\mathrm{\β}}=\frac{1}{\mathrm{\α}F}\underset{i=1}{\overset{\mathrm{\α}F}{\Σ}}{y}_{{\mathrm{\β}}_i},{\stackrel{\‾}{z}}_{\mathrm{\β}}=\frac{1}{\mathrm{\α}F}\underset{i=1}{\overset{\mathrm{\α}F}{\Σ}}{z}_{{\mathrm{\β}}_i}\end{array}$

Wherein β represents weights, and 0 < α < 1, F represents sample rate, and the coordinate figure that RSSI calculates is R _τ(x _α, y _α, z _α), for its average; The coordinate figure of inertial sensor is D _τ(x _β, y _β, z _β), for its average;

Step 6, uses d _τreplace original inertial sensor coordinate, adopt recursive iteration method to calculate new fusion coordinate figure, that is:

$D_{\mathrm{\&tau;}}=d_{\mathrm{\&tau;}-1}+{\&Integral;}_{\mathrm{\&tau;}-1}^{\mathrm{\&tau;}}v\left(\mathrm{\&tau;}\right)d_{\mathrm{\&tau;}}$

In formula, the instantaneous velocity that v (τ) is τ moment object; And repeat calculation procedure 5,6 reach 50 times after obtain final recursive iteration coordinate.

Utilize the received signal strength value received, according to the relational model between RSSI and distance, distance RSSI value far away is less, namely signal strength signal intensity decays along with the increase of distance, obtain the distance of tested point to multiple reference point, and form range equation, the coordinate of tested point is obtained by least square method.Similarly, when measuring RSSI, measure the numerical value of moving object inertial sensor, utilize accekeration integral operation and the coordinate figure of object can be obtained in conjunction with gyrostatic angle information.The two kinds of coordinate figures obtained are carried out filtering process, by the method for local average, carries out continuous local average in measurement data one by one minizone, obtain comparatively level and smooth measurement data.To the method for filtered data acquisition Weighted Fusion, using error as weighted factor, obtain merging coordinate, and fusion coordinate is replaced original inertial sensor coordinate figure, again carry out integral operation, the coordinate obtained merges with RSSI coordinate again.Circulation repeatedly obtains final recursive iteration coordinate figure.

Beneficial effect of the present invention: (1) reduces error of measured data, avoids and affects measurement result because ambient noise is excessive.(2) after calculating two kinds of positioning results, utilize recursive iteration method when decision level fusion, constantly reduce the accumulated error because acceleration analysis causes with the coordinate figure under fusion value replacement single-mode, restrained effectively the impact of noise.

Accompanying drawing explanation

Fig. 1 is schematic flow sheet of the present invention.

Embodiment

As shown in Figure 1, a kind of indoor orientation method based on multimodality fusion recursive iteration of the present invention, specifically comprises the following steps:

Step 1, utilizes the received signal strength RSSI value received to calculate the distance d of tested point to reference node _m,

$d_m={10}^{\frac{-{\mathrm{RSSI}}_m-A}{10\&CenterDot;n}}$

Wherein, A represents the signal strength signal intensity that signal receives at distance 1m place, and n represents propagation factor;

Step 2, obtains least square solution according to least square method, that is:

$\widehat{\mathrm{\&theta;}}={\left(Q^{T}Q\right)}^{-1}{Q}^{T}b$

In formula, $Q=\left[\begin{array}{cccc}-2a_1& -2b_1& -2c_1& 1\\ -2a_2& -2b_2& -2c_1& 1\\ ...& ...& ...& ...\\ -2a_m& -2b_m& -2c_m& 1\end{array}\right],\mathrm{\&theta;}=\left[\begin{array}{c}x\\ y\\ z\\ {R_m}^2\end{array}\right],b=\left[\begin{array}{c}{r_1}^2\\ {r_2}^2\\ ...\\ {r_m}^2\end{array}\right],$

r _m ²＝d _m ²-(a _m ²+b _m ²+c _m ²)

Wherein, (a _m, b _m, c _m) be the coordinate of reference node, the coordinate that (x, y, z) is tested point, R _m ²represent the quadratic sum of tested point coordinate, r _m ²represent the quadratic sum of tested point to reference point distance and the difference of reference point coordinate quadratic sum, m represents reference node number;

Step 3, measures inertial sensor accelerometer and gyrostatic data, carries out integral operation, that is: while measurement received signal strength

s'＝s·cosθ

In formula, s=∫ vdt, v=∫ adt

Wherein, a is accelerometer accekeration in one direction on object under test, and θ is gyroscope angle value in one direction on object under test, and s represents the displacement of object under test, and v represents the speed of object under test;

Step 4, adopt moving average filter algorithm to carry out filtering to the RSSI received and accelerometer, gyrostatic measurement data, Filtering Formula is as follows:

${\stackrel{\&OverBar;}{y}}_j=\frac{1}{M}\underset{i=j-M+1}{\overset{k}{\&Sigma;}}{y}_i$

In formula, M is sliding window length, y _ibe i-th and measure numerical value; Wherein k>=M, k are integer;

Step 5, calculates coordinate figure and inertial sensor coordinate figure to RSSI and is weighted fusion and obtains merging coordinate, that is:

d _τ＝βR _τ+(1-β)D _τ

In formula, $\mathrm{\&beta;}=\frac{{\mathrm{\&beta;}}_2}{{\mathrm{\&beta;}}_1+{\mathrm{\&beta;}}_2},{\mathrm{\&beta;}}_1=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}({\left(x_{{\mathrm{\&alpha;}}_i}-{\stackrel{\&OverBar;}{x}}_{\mathrm{\&alpha;}}\right)}^2+{\left(y_{{\mathrm{\&alpha;}}_i}-{\stackrel{\&OverBar;}{y}}_{\mathrm{\&alpha;}}\right)}^2+{\left(z_{{\mathrm{\&alpha;}}_i}-{\stackrel{\&OverBar;}{z}}_{\mathrm{\&alpha;}}\right)}^2),$

$\begin{array}{c}{\mathrm{\&beta;}}_2=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}({\left(x_{{\mathrm{\&beta;}}_i}-{\stackrel{\&OverBar;}{x}}_{\mathrm{\&beta;}}\right)}^2+{\left(y_{{\mathrm{\&beta;}}_i}-{\stackrel{\&OverBar;}{y}}_{\mathrm{\&beta;}}\right)}^2+{\left(z_{{\mathrm{\&beta;}}_i}-{\stackrel{\&OverBar;}{z}}_{\mathrm{\&beta;}}\right)}^2),{\stackrel{\&OverBar;}{x}}_{\mathrm{\&alpha;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{x}_{{\mathrm{\&alpha;}}_i},{\stackrel{\&OverBar;}{y}}_{\mathrm{\&alpha;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{y}_{{\mathrm{\&alpha;}}_i},\\ {\stackrel{\&OverBar;}{z}}_{\mathrm{\&alpha;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{z}_{{\mathrm{\&alpha;}}_i},{\stackrel{\&OverBar;}{x}}_{\mathrm{\&beta;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{x}_{{\mathrm{\&beta;}}_i},{\stackrel{\&OverBar;}{y}}_{\mathrm{\&beta;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{y}_{{\mathrm{\&beta;}}_i},{\stackrel{\&OverBar;}{z}}_{\mathrm{\&beta;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{z}_{{\mathrm{\&beta;}}_i}\end{array}$

Wherein β represents weights, and 0 < α < 1, F represents sample rate, and the coordinate figure that RSSI calculates is R _τ(x _α, y _α, z _α), for its average; The coordinate figure of inertial sensor is D _τ(x _β, y _β, z _β), for its average;

Step 6, uses d _τreplace original inertial sensor coordinate, adopt recursive iteration method to calculate new fusion coordinate figure, that is:

$D_{\mathrm{\&tau;}}=d_{\mathrm{\&tau;}-1}+{\&Integral;}_{\mathrm{\&tau;}-1}^{\mathrm{\&tau;}}v\left(\mathrm{\&tau;}\right)d_{\mathrm{\&tau;}}$

In formula, the instantaneous velocity that v (τ) is τ moment object; And repeat calculation procedure 5,6 reach 50 times after obtain final recursive iteration coordinate.

Claims (1)

1. based on an indoor orientation method for multimodality fusion recursive iteration, it is characterized in that: specifically comprise the following steps:

Step 1, utilizes the received signal strength RSSI value received to calculate the distance d of tested point to reference node _m,

$d_m={10}^{\frac{-{\mathrm{RSSI}}_m-A}{10\&CenterDot;n}}$

Wherein, A represents the signal strength signal intensity that signal receives at distance 1m place, and n represents propagation factor;

Step 2, obtains least square solution according to least square method, that is:

$\widehat{\mathrm{\&theta;}}={\left(Q^{T}Q\right)}^{-1}{Q}^{T}b$

In formula, $Q=\left[\begin{array}{cccc}-2a_1& -2b_1& -2c_1& 1\\ -2a_2& -2b_2& -2c_1& 1\\ ...& ...& ...& ...\\ -2a_m& -2b_m& -2c_m& 1\end{array}\right],\mathrm{\&theta;}=\left[\begin{array}{c}x\\ y\\ z\\ {R_m}^2\end{array}\right],b=\left[\begin{array}{c}{r_1}^2\\ {r_2}^2\\ ...\\ {r_m}^2\end{array}\right],$

r _m ²＝d _m ²-(a _m ²+b _m ²+c _m ²)

Wherein, (a _m, b _m, c _m) be the coordinate of reference node, the coordinate that (x, y, z) is tested point, R _m ²represent the quadratic sum of tested point coordinate, r _m ²represent the quadratic sum of tested point to reference point distance and the difference of reference point coordinate quadratic sum, m represents reference node number;

Step 3, measures inertial sensor accelerometer and gyrostatic data, carries out integral operation, that is: while measurement received signal strength

s'＝s·cosθ

In formula, s=∫ vdt, v=∫ adt

Wherein, a is accelerometer accekeration in one direction on object under test, and θ is gyroscope angle value in one direction on object under test, and s represents the displacement of object under test, and v represents the speed of object under test;

Step 4, adopt moving average filter algorithm to carry out filtering to the RSSI received and accelerometer, gyrostatic measurement data, Filtering Formula is as follows:

${\stackrel{\&OverBar;}{y}}_j=\frac{1}{M}\underset{i=j-M+1}{\overset{k}{\&Sigma;}}{y}_i$

In formula, M is sliding window length, y _ibe i-th and measure numerical value; Wherein k>=M, k are integer;

Step 5, calculates coordinate figure and inertial sensor coordinate figure to RSSI and is weighted fusion and obtains merging coordinate, that is:

d _τ＝βR _τ+(1-β)D _τ

In formula, $\mathrm{\&beta;}=\frac{{\mathrm{\&beta;}}_2}{{\mathrm{\&beta;}}_1+{\mathrm{\&beta;}}_2},$ ${\mathrm{\&beta;}}_1=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}({\left(x_{{\mathrm{\&alpha;}}_i}-{\stackrel{\&OverBar;}{x}}_{\mathrm{\&alpha;}}\right)}^2+{\left(y_{{\mathrm{\&alpha;}}_i}-{\stackrel{\&OverBar;}{y}}_{\mathrm{\&alpha;}}\right)}^2+{\left(z_{{\mathrm{\&alpha;}}_i}-{\stackrel{\&OverBar;}{z}}_{\mathrm{\&alpha;}}\right)}^2),$ ${\mathrm{\&beta;}}_2=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}({\left(x_{{\mathrm{\&beta;}}_i}-{\stackrel{\&OverBar;}{x}}_{\mathrm{\&beta;}}\right)}^2+{\left(y_{{\mathrm{\&beta;}}_i}-{\stackrel{\&OverBar;}{y}}_{\mathrm{\&beta;}}\right)}^2+{\left(z_{{\mathrm{\&beta;}}_i}-{\stackrel{\&OverBar;}{z}}_{\mathrm{\&beta;}}\right)}^2),$ ${\stackrel{\&OverBar;}{x}}_{\mathrm{\&alpha;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{x}_{{\mathrm{\&alpha;}}_i},$ ${\stackrel{\&OverBar;}{y}}_{\mathrm{\&alpha;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{y}_{{\mathrm{\&alpha;}}_i},$ ${\stackrel{\&OverBar;}{z}}_{\mathrm{\&alpha;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{z}_{{\mathrm{\&alpha;}}_i},$ ${\stackrel{\&OverBar;}{x}}_{\mathrm{\&beta;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{x}_{{\mathrm{\&beta;}}_i},$ ${\stackrel{\&OverBar;}{y}}_{\mathrm{\&beta;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{y}_{{\mathrm{\&beta;}}_i},$ ${\stackrel{\&OverBar;}{z}}_{\mathrm{\&beta;}}=\frac{1}{\mathrm{\&alpha;}F}\underset{i=1}{\overset{\mathrm{\&alpha;}F}{\&Sigma;}}{z}_{{\mathrm{\&beta;}}_i}$

Wherein β represents weights, and 0 < α < 1, F represents sample rate, and the coordinate figure that RSSI calculates is R _τ(x _α, y _α, z _α), for its average; The coordinate figure of inertial sensor is D _τ(x _β, y _β, z _β), for its average;

Step 6, uses d _τreplace original inertial sensor coordinate, adopt recursive iteration method to calculate new fusion coordinate figure, that is:

$D_{\mathrm{\&tau;}}=d_{\mathrm{\&tau;}-1}+{\&Integral;}_{\mathrm{\&tau;}-1}^{\mathrm{\&tau;}}v\left(\mathrm{\&tau;}\right)d_{\mathrm{\&tau;}}$

In formula, the instantaneous velocity that v (τ) is τ moment object; And repeat calculation procedure 5,6 reach 50 times after obtain final recursive iteration coordinate.