Generalized Kalman filter positioning method and system based on distance measurement

By adjusting the measurement equation to retain only the finite order non-zero derivative and adaptive noise mean estimation, combined with the SAGE-HUSA algorithm, the error problem caused by the higher order terms of the generalized Kalman filtering algorithm in nonlinear systems is solved, achieving more accurate and Stable state estimation improves the accuracy of indoor positioning.

CN114994601BActive Publication Date: 2024-12-06HEFEI BLUEX MICRO TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202210621595.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-06-02
Publication Date
2024-12-06
Estimated Expiration
2042-06-02

AI Technical Summary

Technical Problem

The existing generalized Kalman filtering algorithm in nonlinear systems has large estimation errors due to ignoring higher-order terms and cannot effectively control the impact of higher-order derivatives, resulting in reduced positioning accuracy and instability in the system.

Method used

By adjusting the measurement equation, only the finite-order non-zero derivative is retained, combined with the adaptive noise mean estimation and the SAGE-HUSA adaptive algorithm, the nonlinearity of the measurement equation is reduced, ensuring that the Taylor series expansion only contains finite terms and avoiding higher-order terms truncation error.

Benefits of technology

More accurate state estimation is achieved, positioning errors are reduced, and the stable operation of the Kalman filter is ensured. Even if the deviation between the estimated value and the nominal value is large, it will not cause the filter to diverge, and the positioning accuracy is improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114994601B_ABST
    Figure CN114994601B_ABST
Patent Text Reader

Abstract

The present invention provides a generalized Kalman filter positioning system based on distance measurement, including a distance measurement calculation module, characterized in that it also includes: a finite-order non-zero derivative measurement equation generation module, which adjusts and changes the measurement equation so that the Taylor series expansion of the nonlinear measurement equation only contains finite terms; an adaptive noise mean estimation module, which uses an adaptive algorithm to estimate the measurement noise mean in the adjusted measurement equation; a generalized Kalman filter position solution module, which uses the generalized Kalman filter principle to solve the target position based on the system state equation and the adjusted measurement equation in combination with an adaptive algorithm. After the measurement equation is adjusted and changed, all the partial derivatives above the third order are zero, and most of the second-order partial derivatives are zero. Since all the partial derivatives above the third order are zero, it is ensured that all the items above the third order of the Taylor series are zero. Even if the deviation between the estimated value and the nominal value is large, it will not easily cause the Kalman filter to diverge.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of wireless positioning and navigation technology, and in particular to a generalized Kalman filter positioning method and system based on distance measurement. Background Art

[0002] With the continuous development of modern society, the process of urbanization is accelerating, and large buildings are increasing. People spend more than 80% of their time in indoor environments (including underground, mines, tunnels, etc.). Catering, shopping, and subway transportation have become very important parts of people's lives. These changes have caused people's demand for indoor location services to increase rapidly. Public safety, production safety, emergency rescue, public health, the Internet of Things, special population monitoring, large venue management, smart city construction and other fields all require the use of accurate indoor positioning information. Therefore, the indoor navigation and location service industry has a trillion-dollar market space.

[0003] At present, there are many kinds of indoor positioning technologies, which can be divided into two categories according to the size of the signal coverage area: ① Local indoor positioning, such as wireless local area network (WLAN), radio frequency tag (RFID), Zigbee, Bluetooth (Blue Tooth, BT), Ultra Wide Band (UWB), geomagnetic field, infrared positioning, optical tracking positioning, computer vision positioning, ultrasonic positioning, etc. ② Wide-area indoor positioning, such as assisted GPS (A-GPS) based on mobile communication network, pseudo satellite (Pseudolite), terrestrial digital communication and broadcast network positioning system, etc. Bluetooth positioning technology has achieved wide network coverage with its advantages of low cost, low power consumption and high positioning accuracy, and has been widely used in indoor positioning scenarios.

[0004] The release of Bluetooth 5.0 has reduced power consumption and greatly improved positioning accuracy, and its positioning-IoT devices have made indoor positioning even more popular. There are two main ways of Bluetooth iBeacon positioning: based on received signal strength RSSI (Received Signal Strength Indication) and based on positioning fingerprints, or a combination of the two. The biggest problem with positioning fingerprints is that the labor cost and time cost of obtaining fingerprint data in the early stage are very high, and database maintenance is difficult. Moreover, if a shopping mall adds a new base station or undergoes other modifications, the original fingerprint data may no longer be applicable. At present, the mainstream Bluetooth positioning technology on the market uses a ranging triangulation positioning algorithm combined with a ranging method based on the received signal strength RSSI value, that is, the distance between two Bluetooth devices is calculated by the RSSI value and the indoor electromagnetic wave propagation loss model. This method is simple and easy to implement, but the disadvantage is that the Bluetooth signal RSSI value is greatly affected by the environment. The signal attenuation caused by the moving human body blocking the electromagnetic wave and the multipath effect cause serious abnormal fluctuations in the ranging results, and the positioning error increases significantly. In fact, all indoor positioning technologies that implement triangulation based on received signal strength RSSI ranging (such as wireless LAN (WIFI), ZigBee, radio frequency tags, etc.) are plagued by the same problem.

[0005] Kalman filtering is an algorithm that uses the linear system state equation to optimally estimate the system state through the system input and output observation data. Since the observation data includes the influence of noise and interference in the system, the optimal estimate can also be regarded as a filtering process. Kalman filtering is a linear minimum variance estimation that uses the dynamic equation, i.e., the state equation, to describe the dynamic change law of the estimated quantity. The basic equations of Kalman filtering include the state equation that describes the dynamic change law of the estimated quantity and the measurement equation that reflects the quantitative relationship between the observed quantity and the estimated quantity. In the standard Kalman filtering theory, both the state equation and the measurement equation are linear equations (groups), and the Kalman filter can output the optimal estimate of the system state. Therefore, Kalman filtering is essentially a linear optimal estimate and cannot be directly applied to the estimation of nonlinear systems. The so-called nonlinear system refers to the state equation and the measurement equation, at least one of which is a nonlinear equation (group).

[0006] The state equation of a more general nonlinear continuous system can be expressed as (Qin Yongyuan, Zhang Hongyue, Wang Shuhua, 1998):

[0007]

[0008] The measurement equation is expressed as:

[0009] Z(t)=h[X(t),t]+v(t)

[0010] Among them, X(t) is the state quantity to be estimated, Z(t) is the observed quantity (measured value), w(t) and v(t) are zero-mean white noises that are unrelated to each other, and G(t) is a certain function. f[X(t), t] is the system function, and h[X(t), t] is the observation function; both are certain linear or nonlinear functions, and at least one is a nonlinear function.

[0011] The mathematical models of physical systems encountered in engineering practice are often nonlinear. For example, inertial navigation systems on aircraft and ships, missile guidance systems, Doppler navigation systems, satellite navigation systems, and many other industrial control systems are generally nonlinear systems. In order to use Kalman filtering theory to solve the state estimation problem of nonlinear systems, a generalized Kalman filtering algorithm was proposed. The basic idea is to first estimate the state of a certain nominal state X. n (t) Expand the nonlinear system function and observation function into Taylor series:

[0012]

[0013]

[0014] Then, assuming that the deviation X(t)-X n The absolute value of (t) is small enough, and the higher-order terms of the Taylor series are ignored, and the first approximation is taken:

[0015]

[0016]

[0017] The system after the above approximate processing is a linear system. The state equation and the measurement equation are both linear equations (groups). The standard Kalman filter theory can be applied to obtain the suboptimal estimate of the state, that is, the generalized Kalman filter estimate.

[0018] From the above analysis, we can see that the generalized Kalman filter ignores the higher-order terms above the quadratic term of the Taylor series. The effectiveness and estimation error of the generalized Kalman filter depend on whether the influence of the higher-order terms can really be ignored. The higher-order terms of the Taylor series of the system function and the observation function in the state equation and the measurement equation

[0019] and

[0020] All are high-order derivatives

[0021] and

[0022] The higher power of the deviation [X(t)-X n (t)] mAlthough the generalized Kalman filter algorithm uses the optimal state estimate as the nominal state, that is, A smaller deviation can be obtained. However, in actual engineering applications, it is difficult to ensure that the generalized Kalman filter can always maintain a sufficiently small deviation during operation in any application scenario. The existing generalized Kalman filter algorithm has no control over the size of the higher-order derivatives in the Taylor series, another aspect of the problem. Whether the absolute value of the higher-order derivative is small enough is completely unknown. Therefore, ignoring the truncation error introduced by the higher-order terms brings hidden dangers to the stable operation of the generalized Kalman filter. That is, the generalized Kalman filter requires that the system function and the observation function are smooth enough and weakly nonlinear. If the function has strong nonlinearity, it will lead to a large state estimation error and the generalized Kalman filter will lose its effectiveness.

[0023] Taking the Bluetooth indoor two-dimensional positioning scenario as an example, assume that Bluetooth beacons are deployed at N known locations to provide indoor positioning navigation for people or assets (robots) carrying Bluetooth devices. The Bluetooth device with unknown location coordinates (x, y) and the location coordinates (x i ,y i ) is:

[0024]

[0025] Measurement equation:

[0026]

[0027] because

[0028]

[0029] is a nonlinear (vector) function, so the measurement equation is a nonlinear equation (group). According to the existing generalized Kalman filter theory, it is necessary to perform Taylor series expansion on the measurement equation and take a first-order approximation. Obviously, The higher-order partial derivatives of X(t) and y(t) exist and are not zero. Therefore, the higher-order partial derivatives of h[X(t), t] with respect to X(t) are

[0030]

[0031] It also exists and is not zero. The size of the higher-order partial derivative is not controlled in any way, and whether its absolute value is small enough is completely unknown. Summary of the invention

[0032] The present invention provides a generalized Kalman filter positioning method and system based on distance measurement, which significantly reduces the nonlinearity of the measurement equation by changing the measurement equation.

[0033] The technical problem to be solved by the present invention is achieved through the following technical solutions:

[0034] The present invention provides a generalized Kalman filter positioning system based on distance measurement, comprising a distance measurement calculation module, characterized in that it also includes:

[0035] Finite order non-zero derivative measurement equation generation module, adjusts and changes the measurement equation so that the Taylor series expansion of the nonlinear measurement equation contains only finite terms;

[0036] An adaptive noise mean estimation module uses an adaptive algorithm to estimate and adjust the measurement noise mean in the changed measurement equation;

[0037] The generalized Kalman filter position solution module is based on the system state equation, the adjusted measurement equation, and the adaptive algorithm, and uses the generalized Kalman filter principle to solve the target position.

[0038] Preferably, the adjustment to change the measurement equation is to perform a square operation on both sides of the original measurement equation simultaneously.

[0039] Preferably, the adaptive algorithm is the SAGE-HUSA adaptive algorithm.

[0040] Preferably, the distance measurement calculation module performs Bluetooth RSSI ranging.

[0041] A generalized Kalman filter positioning method based on distance measurement, comprising calculating and measuring the distance between a target to be positioned and each beacon, characterized in that it also includes:

[0042] Adjust and change the measurement equation so that the Taylor series expansion of the nonlinear measurement equation contains only finite terms;

[0043] Using an adaptive algorithm to estimate the mean value of the measurement noise in the measurement equation after the adjustment;

[0044] Based on the system state equation, the changed measurement equation is adjusted and combined with the adaptive algorithm, the generalized Kalman filter principle is used to calculate the target position.

[0045] Preferably, the adjustment to change the measurement equation is to perform a square operation on both sides of the original measurement equation simultaneously.

[0046] Preferably, the adaptive algorithm is the SAGE-HUSA adaptive algorithm.

[0047] Preferably, the distance measurement method is Bluetooth RSSI ranging.

[0048] The beneficial effects of the present invention are: after the measurement equation is adjusted and changed, all the partial derivatives above the third order are zero, most of the second-order partial derivatives are zero, and a small number (2) of the second-order partial derivatives are equal to 2 (smaller). Therefore, the measurement equation (group) is expanded into a Taylor series and a first-order approximation is performed, which only ignores a small number of quadratic terms of the Taylor series. Under the premise that the generalized Kalman filter algorithm uses the optimal state estimate as the nominal state and can obtain a smaller deviation, the Taylor series first-order approximation of the linear measurement equation and the nonlinear state equation of the minimum high-order derivative generalized Kalman filter technical solution can make a fairly accurate description of the system. Since all the partial derivatives above the third order are zero, it ensures that all the terms above the third order of the Taylor series are zero, even if the deviation between the estimated value and the nominal value It is large and will not easily cause the Kalman filter to diverge. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is the finite order non-zero derivative adaptive generalized Kalman filtering process of the present invention;

[0050] Figure 2 It is a computer simulation verification scheme for the finite-order non-zero derivative adaptive generalized Kalman filter algorithm of the present invention;

[0051] Figure 3 It is the target position solution error obtained by computer simulation of the finite-order non-zero derivative adaptive generalized Kalman filtering algorithm of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0053] In order to avoid the above-mentioned problem of reduced accuracy of Kalman estimation due to excessive truncation error introduced by ignoring high-order terms, it is considered to adjust the measurement equation; reduce the nonlinearity of the measurement equation / observation function to ensure that the first-order Taylor series approximation can still accurately describe the system.

[0054] The reason why the high-order derivatives of the aforementioned Bluetooth indoor two-dimensional positioning scenario measurement equation are always not zero is that its h[X(t), t] function is a radical, that is, a power function with an exponent of 0.5:

[0055]

[0056] If we remove the 0.5 power by squaring it, the result is a polynomial containing only positive integer exponents, whose higher-order derivatives must be zero. That is, squaring both sides of the distance equation gives

[0057] di_tr 2 =(xx i ) 2 +(yy i ) 2

[0058] Distance Observation

[0059]

[0060] n i (t) is zero-mean white noise. The distance measurement between the target device and different beacons is also an independent process, so no matter t j With t k Is it a different time, n i (t j ) and n1(t k ) are all independent random variables.

[0061] Since the square of the distance observed

[0062]

[0063] Knowable

[0064] D i (t) = d i 2 (t) = d i_tr 2 (t)+v i (t)

[0065] in

[0066]

[0067] The mean of the above noise

[0068] E[v i (t)]=E[n i 2 (t)]

[0069] Covariance

[0070]

[0071] variance

[0072]

[0073] From this we can see that v i (t) is white noise with non-zero mean.

[0074] After squaring both sides of the distance equation and transposing the terms, we can get:

[0075] d i_tr 2 -(x i 2 +y i 2 )=-2xx i -2yy i +(x 2 +y 2 )

[0076] make

[0077] v i (t) = m i (t)+vv i (t) m i (t) = E[v i (t)]

[0078] vv i (t) is zero-mean white noise, and its variance

[0079]

[0080] Take the mean value m i (t) and its derivative as the state quantity to be estimated, the new measurement equation is:

[0081] z i (t) = D i (t)-(x i 2 +y i 2 ) = d i_tr 2 (t)+v i (t)-(x i 2 +y i 2 )

[0082] =-2xx i -2yy i +(x 2 +y 2 )+m i (t)+vv i (t)

[0083] The new measurement equation is

[0084]

[0085] Right now

[0086] Z(t)=h[X(t)]+m(t)+vv(t)

[0087]

[0088] m(t)=[m1(t),m2(t),…,m N (t)] T

[0089] vv(t)=[vv1(t),vv2(t),…,vv N (t)] T

[0090] The superscript T represents the matrix transpose.

[0091]

[0092] but

[0093]

[0094] in

[0095]

[0096] w(t)=[0,0,w x (t), w y (t)] T

[0097] The state equation is a linear equation (group), and its Taylor series first approximation is the state equation itself. The measurement equation (group) consists of N equations, and the determination function part

[0098] h[X(t),t]=[h1[X(t),t],h2[X(t),t],…,h N [X(t), t]] T

[0099] h i [X(t), t] = x 2 (t)+y 2 (t)-2x(t)x i -2y(t)y i

[0100] According to the generalized Kalman filter theory, the measurement equation (group) needs to be expanded into a Taylor series and approximated. The Taylor series expansion of the measurement equation (group) is the Taylor series expansion of the observation function h[X(t), t]. In essence, each member function h1[X(t), t] to h N [X(t), t] are each expanded by Taylor series, and the member function h i [X(t), t] is a scalar multivariate function.

[0101] For the convenience of mathematical derivation,

[0102]

[0103] Abbreviated as

[0104]

[0105] Its non-zero first-order partial derivative

[0106]

[0107] Its non-zero second-order partial derivative

[0108]

[0109] Because h i All partial derivatives of [X(t), t] above the third order are zero, that is,

[0110]

[0111] The basic equation of the continuous system after the first-order approximation of the Taylor series is (Qin Yongyuan, Zhang Hongyue, Wang Shuhua, 1998):

[0112]

[0113] δZ(t)=H(t)δX(t)+m(t)+vv(t)

[0114] in,

[0115]

[0116]

[0117]

[0118] w(t)=[0,0,w x (t), w y (t)] T

[0119] vv(t)=[vv1(t),vv2(t),…,vv N (t)] T

[0120] w x (t), w y (t), wm(t) and v i (t)1≤i≤N are all uncorrelated zero-mean white noise.

[0121] E[w(t)w T (τ)] = qδ(t-τ)

[0122] E[vv(t)vv T (τ)] = rδ(t-τ)

[0123] where δ(·) is the impulse function, q is the variance intensity matrix of w(t), and r is the variance intensity matrix of v(t).

[0124]

[0125]

[0126] Depend on

[0127] h[X(t),t]=[h1[X(t),t],h2[X(t),t],…,h N [X(t), t]] T

[0128] h i [X(t), t] = x 2 (t)+y 2 (t)-2x(t)Xi-2y(t)y i

[0129] Knowable

[0130]

[0131] and

[0132]

[0133] Therefore

[0134]

[0135] According to linear system theory, the discretized form of the system equation is:

[0136]

[0137] in

[0138]

[0139] Assume that the filtering period T = t k -t k-1 is short enough, F(t) can be approximately regarded as a constant matrix,

[0140] F(t)≈F(t k-1 ) k-1 ≤t<t k

[0141] Get a one-step transfer array

[0142]

[0143] Right now

[0144]

[0145] F k =F(t k )

[0146] approximation

[0147]

[0148] get

[0149]

[0150] Where I is the identity matrix.

[0151] Measurement array

[0152]

[0153] From this, we can get the basic equation of Kalman filter after discretization.

[0154] δX k =Φ k,k-1 δX k-1 +W k-1

[0155] δZ k =H k δX k +m k +V k

[0156] It can be seen that W k It also has zero mean and

[0157]

[0158] in

[0159]

[0160] The numerical algorithm for its value is as follows:

[0161] M1=q

[0162] M i+1 =F k M i +(F k M i ) T i≥1 F k =F(t k )

[0163]

[0164] Measurement equation

[0165] δZ(t k )=H(t k )δX(t k )+m(t k )+vv(t k )

[0166] And its corresponding discrete form

[0167] δZ k =H k δX k +m k +V k

[0168] Knowable

[0169] E[V k V j T ]=E[vv(t k )vv T (t j )]=rδ(t k -t j )=R k δ k,j

[0170] R k =r

[0171] because

[0172]

[0173] and

[0174]

[0175] If n i (t) satisfies Gaussian distribution, that is, n i (t k ) is zero-mean Gaussian white noise, so

[0176]

[0177] therefore

[0178]

[0179] Kalman filter equation with deviation

[0180]

[0181]

[0182] Adaptive filter is a filtering method that can suppress the divergence of Kalman filter. In filtering calculation, it uses measurement to continuously correct the predicted value, and also estimates or corrects the unknown or inaccurate system model parameters and noise statistical parameters. According to the SAGE-HUSA adaptive filtering principle (Li Ning, Zhu Ruihui, Zhang Yonggang, 2014), the estimation of the statistical characteristics of measurement noise

[0183]

[0184] d = (1-b) / (1-b K )

[0185] Where K is the total number of filter sampling points, 0<b<1.

[0186] Filter Gain Array

[0187]

[0188] One-step transfer variance matrix

[0189]

[0190] Estimated mean square error matrix

[0191]

[0192] Discrete nonlinear generalized Kalman filter equation

[0193]

[0194]

[0195]

[0196]

[0197]

[0198]

[0199] Figure 1This is the finite-order non-zero derivative adaptive generalized Kalman filtering process of this embodiment. First, the distance between the device to be located and the beacon with a known position is estimated using RSSI or signal arrival time to generate the original observation. The non-positive integer power exponent in the original observation equation is removed by exponentiation, so that the determination function part in the new observation equation is a polynomial. The SAGE-HUSA adaptive algorithm is used to estimate the noise mean in the new measurement equation. Under the condition that the state equation is linear, the Taylor series expansion of the nonlinear measurement equation has no terms above the third order, and the Taylor series first-order approximation accurately describes the system characteristics. Finally, the generalized Kalman filter is implemented based on the new measurement equation and the state equation.

[0200] Figure 2 This is a computer simulation verification scheme for the finite-order non-zero derivative adaptive generalized Kalman filter algorithm of this embodiment. According to the finite-order non-zero derivative adaptive generalized Kalman filter equation derived in the previous section, a MATLAB program is written to implement the algorithm scheme, and a computer simulation experiment is performed. Figure 2 As shown in the figure, the test setting is to evenly distribute 6 Bluetooth beacons along a circle with a radius of 1000 meters (the position of the asterisk * in the figure). The route of the moving target to be located is shown in the red track in the figure. The two-dimensional coordinates of the initial position of the moving target are (500, -500), the initial velocity and initial acceleration are both zero, and the subsequent acceleration is a random quantity.

[0201] Figure 3 It is the target position solution error obtained by computer simulation of the finite order non-zero derivative adaptive generalized Kalman filter algorithm of this embodiment. The MATLAB program implements two-dimensional position estimation for the moving target. Figure 3 From top to bottom, the estimated error of the two-dimensional position estimation in the X-axis direction, the estimated error in the Y-axis direction, and the two-dimensional positioning error are shown. The initial values ​​of the Kalman filter (position, velocity, acceleration, etc.) are all set to zero. Figure 3 It can be seen that the finite-order non-zero derivative adaptive generalized Kalman filter algorithm converges quickly, runs smoothly and works stably, proving that the theoretical design of the filter is basically correct and effective.

[0202] The finite-order non-zero derivative adaptive generalized Kalman filter positioning technology solution protected by the present invention can be widely used in any indoor and outdoor positioning systems based on ranging and triangulation positioning, such as wireless local area networks, radio frequency tags, Zigbee, ultra-wideband radio, etc., as well as Beidou, GPS (A-GPS), pseudo-satellite, terrestrial digital communication and broadcast network positioning systems, etc., and is not limited to Bluetooth positioning devices with RSSI ranging.

Claims

1. A generalized Kalman filter positioning system based on distance measurement, comprising a distance measurement calculation module, characterized in that: Also includes: Finite order non-zero derivative measurement equation generation module, adjusts and changes the measurement equation so that the Taylor series expansion of the nonlinear measurement equation contains only finite terms; An adaptive noise mean estimation module uses an adaptive algorithm to estimate and adjust the measurement noise mean in the changed measurement equation; The generalized Kalman filter position calculation module uses the generalized Kalman filter principle to calculate the target position based on the system state equation, the adjusted measurement equation, and the adaptive algorithm; The adjustment to change the measurement equation is to perform a square operation on both sides of the original measurement equation.

2. The generalized Kalman filter positioning system based on distance measurement according to claim 1, characterized in that: The adaptive algorithm is the SAGE-HUSA adaptive algorithm.

3. The generalized Kalman filter positioning system based on distance measurement according to claim 1, characterized in that: The distance measurement calculation module performs Bluetooth RSSI distance measurement.

4. A generalized Kalman filter positioning method based on distance measurement running on the generalized Kalman filter positioning system based on distance measurement as claimed in any one of claims 1 to 3, comprising calculating and measuring the distance between the target to be positioned and each beacon, characterized in that: Also includes: Adjust and change the measurement equation so that the Taylor series expansion of the nonlinear measurement equation contains only finite terms; Using an adaptive algorithm to estimate the mean value of the measurement noise in the measurement equation after the adjustment; Based on the system state equation, the changed measurement equation is adjusted and combined with the adaptive algorithm, the generalized Kalman filter principle is used to calculate the target position.

5. The generalized Kalman filter positioning method based on distance measurement according to claim 4, characterized in that: The adjustment to change the measurement equation is to perform a square operation on both sides of the original measurement equation.

6. The generalized Kalman filter positioning method based on distance measurement according to claim 4, characterized in that: The adaptive algorithm is the SAGE-HUSA adaptive algorithm.

7. The generalized Kalman filter positioning method based on distance measurement according to claim 4, characterized in that: The distance measurement method is Bluetooth RSSI ranging.

Citation Information

Patent Citations

  • Nonlinear system state estimation method based on Kalman filtering positioning

    CN106646356A

  • Combined navigation method based on Taylor high order remainder term fitting

    CN106931966A