An indoor positioning method based on double fault detection of UWB

Abstract

The application discloses a UWB indoor positioning method based on double fault detection, and relates to the technical field of indoor positioning, comprising the following steps: building an indoor positioning network comprising a repeater and multi-stage UWB positioning anchor points, and synchronizing all anchor points by using wireless cascade clock synchronization based on UWB; obtaining observation values of each anchor point and a target to be positioned; after distance residual consistency preliminary detection, calculating the dispersion of the target to be positioned from fault observation values and combining random forest classification; using dynamic weighted least squares to solve normal data, and after abnormal data are compensated by biased Kalman filtering, the abnormal data are solved again to obtain final positioning coordinates. By using wireless cascade clock synchronization based on UWB, combining double fault detection and type positioning solving strategy, the clock synchronization precision and fault data processing capacity of the UWB indoor positioning system are improved, and the positioning precision in a complex indoor environment is significantly improved.

Description

Technical Field

[0001] This application relates to the field of indoor positioning technology, and in particular to a UWB indoor positioning method based on dual fault detection. Background Technology

[0002] Ultra-wideband (UWB) technology is widely used in indoor positioning due to its low power consumption, strong resistance to multipath interference, high security, and excellent ranging accuracy, theoretically achieving centimeter-level or even millimeter-level positioning accuracy. However, in real-world complex indoor environments, UWB communication signals are susceptible to interference from factors such as obstruction, multipath effects, and clock drift, resulting in ranging deviations such as non-line-of-sight errors and multipath errors, leading to a significant decrease in positioning accuracy.

[0003] Existing UWB indoor positioning methods mostly employ a single solution algorithm, lacking effective detection and processing of fault data in the observations. The direct involvement of fault signals in the solution can easily lead to distortion of positioning results. Although some methods introduce fault detection mechanisms, their error compensation effect for abnormal data is limited. In addition, clock synchronization between positioning nodes often relies on wired methods, resulting in complex network deployment, high maintenance costs, and large fluctuations in synchronization accuracy, further restricting the overall positioning performance of the system.

[0004] In the prior art, Chinese patent CN107835491A discloses a signal synchronization method and an indoor positioning system based on UWB, including: a repeater forwarding the positioning signal of an outdoor base station to the first-level positioning node of each link in the indoor positioning network corresponding to the indoor positioning system; the first-level positioning node of each link performs preset processing on the positioning signal to obtain the corresponding synchronization positioning signal; from the first-level to the penultimate-level positioning node of each link, each level positioning node of each link sends its corresponding synchronization positioning signal and signal time of flight to the next-level positioning node of that link; the next-level positioning node performs clock correction on the received synchronization positioning signal according to the received signal time of flight to obtain the corresponding synchronization positioning signal.

[0005] However, the aforementioned existing technologies have not studied or solved the above problems, and the clock synchronization accuracy, fault data processing capability, and positioning accuracy of UWB indoor positioning systems in complex indoor environments need to be improved. Summary of the Invention

[0006] This application provides a UWB indoor positioning method based on dual fault detection to address the issues of insufficient clock synchronization accuracy, fault data processing capability, and positioning accuracy in complex indoor environments in existing UWB indoor positioning systems.

[0007] On the one hand, this application provides a UWB indoor positioning method based on dual fault detection, including the following steps: Step 1: Build an indoor positioning network that includes repeaters and multi-level UWB positioning anchors, and use a UWB-based wireless cascaded clock synchronization method to synchronize the clocks of all anchors.

[0008] Step 2: In an indoor positioning network that has completed clock synchronization, acquire the observations between the target to be located and each UWB positioning anchor point.

[0009] Step 3: Based on the least squares positioning principle, perform an initial check on the distance residual consistency of the observations to determine whether there are any faults in the observations.

[0010] Step four: For the observations that were initially determined to have faults, calculate the dispersion of the target to be located and use the random forest algorithm to classify the fault data, distinguishing between normal data and abnormal data.

[0011] Step 5: The normal data is solved using a dynamic weighted least squares algorithm, and the abnormal data is first compensated for by a biased Kalman filter, and then substituted into the weighted least squares algorithm to obtain the final positioning coordinates of the target to be located.

[0012] In one possible implementation, in step one, the indoor positioning network is divided into several links, each link containing multiple levels of UWB positioning anchors connected in sequence; the repeater forwards the positioning signal from the outdoor base station to the first-level anchor of each link, the first-level anchor performs preset processing on the positioning signal to generate a synchronous positioning signal; each level of anchor sequentially sends its own synchronous positioning signal and the signal flight time to the next level anchor to the next level anchor, and the next level anchor performs clock correction on the received synchronous positioning signal according to the signal flight time.

[0013] In one possible implementation, in step one, different links correspond to different UWB channels, and when there is signal interference between the two links, the transmission of the synchronous positioning signal is completed in a time-division manner.

[0014] In one possible implementation, in step two, the acquired observations include the three-dimensional coordinates of each UWB positioning anchor point and the distance between the target to be positioned and each UWB positioning anchor point.

[0015] In one possible implementation, in step three, the observation is linearized into a matrix equation based on the least squares positioning principle, the ranging residual vector and its root mean square value are calculated, and the root mean square value is compared with a preset residual threshold to determine whether the observation has a fault.

[0016] In one possible implementation, in step four, the dispersion of the target to be located is calculated based on the anchor point distance values; the classification thresholds for normal and abnormal data are obtained through the random forest algorithm, and normal and abnormal data are distinguished based on the comparison results between the dispersion of the target to be located and the classification thresholds.

[0017] In one possible implementation, in step five, for normal observation data, a weighted least squares localization algorithm with dynamically assigned weights is used for calculation; for abnormal observation data, the ranging error is first estimated by a biased Kalman filter that introduces a gross error discrimination factor, and then the error-compensated observations are substituted into the weighted least squares algorithm for calculation.

[0018] The UWB indoor positioning method based on dual fault detection in this application has the following advantages: By employing UWB-based wireless cascaded clock synchronization, combined with dual fault detection and categorized positioning solutions, the clock synchronization accuracy and fault data processing capabilities of the UWB indoor positioning system are improved, significantly enhancing positioning accuracy in complex indoor environments. This method, implemented based on existing UWB hardware modules, can be widely applied to complex indoor scenarios, demonstrating good engineering practicality and promotional value.

[0019] By dividing the indoor positioning network into multiple multi-level anchor links, the repeater forwards the signal to the first anchor to generate a synchronization signal, and each level of anchor sequentially transmits the synchronization signal and the time-of-flight correction clock, which improves the wireless synchronization accuracy of UWB anchors and simplifies the system deployment and maintenance process.

[0020] By allocating different UWB channels to different links and using time-division multiplexing to transmit synchronous positioning signals when there is signal interference between links, the reliability of synchronous signal transmission in multi-link UWB positioning networks is improved, and synchronization errors caused by signal interference between links are avoided.

[0021] By calculating the dispersion of the target to be located, and combining it with the random forest algorithm to obtain the classification thresholds for normal and abnormal data, and distinguishing the data types accordingly, the detection accuracy and classification accuracy of fault data in UWB positioning observations are improved, and fault data is avoided from directly participating in the solution.

[0022] By using dynamic weighted least squares to solve normal observations and then applying biased Kalman filtering compensation with gross error discrimination factor to abnormal observations before solving, the utilization efficiency of UWB positioning data and the error compensation effect of abnormal data are improved, further enhancing positioning accuracy. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1A flowchart illustrating a UWB indoor positioning method based on dual fault detection, provided for an embodiment of this application; Figure 2 This is a schematic diagram of the indoor positioning network architecture and clock synchronization link provided in the embodiments of this application; Figure 3 A schematic diagram of the UWB positioning anchor hardware module and signal processing flow provided in the embodiments of this application; Figure 4 This is a schematic diagram of the dual fault detection process provided in the embodiments of this application; Figure 5 The flowchart for the classification and positioning solution provided in the embodiments of this application is shown. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] like Figure 1 As shown, this application provides a UWB indoor positioning method based on dual fault detection, including the following steps: Step 1: Build an indoor positioning network that includes repeaters and multi-level UWB positioning anchors, and use a UWB-based wireless cascaded clock synchronization method to synchronize the clocks of all anchors.

[0027] Step 2: In an indoor positioning network that has completed clock synchronization, acquire the observations between the target to be located and each UWB positioning anchor point.

[0028] Step 3: Based on the least squares positioning principle, perform an initial check on the distance residual consistency of the observations to determine whether there are any faults in the observations.

[0029] Step four: For the observations that were initially determined to have faults, calculate the dispersion of the target to be located and use the random forest algorithm to classify the fault data, distinguishing between normal data and abnormal data.

[0030] Step 5: The normal data is solved using a dynamic weighted least squares algorithm, and the abnormal data is first compensated for by a biased Kalman filter, and then substituted into the weighted least squares algorithm to obtain the final positioning coordinates of the target to be located.

[0031] For example, in step one, the indoor positioning network is divided into several links, each link containing multiple levels of UWB positioning anchors connected in sequence; the repeater forwards the positioning signal from the outdoor base station to the first-level anchor of each link, the first-level anchor performs preset processing on the positioning signal to generate a synchronous positioning signal; each level of anchor sequentially sends its own synchronous positioning signal and the signal flight time to the next level anchor to the next level anchor, and the next level anchor performs clock correction on the received synchronous positioning signal according to the signal flight time.

[0032] Specifically, in this embodiment, in step one, the repeater forwards the positioning signal of the outdoor base station to the first-level UWB positioning anchor point of each link via wired communication. The preset processing includes down-conversion of the positioning signal, A / D sampling, signal acquisition, loop tracking, disciplining and frequency calibration, and synchronous broadcasting.

[0033] For example, in step one, different links correspond to different UWB channels. When there is signal interference between two links, the transmission of the synchronous positioning signal is completed in a time-division manner.

[0034] Specifically, in this embodiment, in step one, the UWB positioning anchor point measures the signal flight time in a two-way bidirectional manner using the UWB module and updates the time at preset time intervals. The signal transmission time measurement accuracy of the UWB module is within 0.5ns.

[0035] For example, in step two, the acquired observations include the three-dimensional coordinates of each UWB positioning anchor point and the distance between the target to be positioned and each UWB positioning anchor point.

[0036] For example, in step three, the observation is linearized into a matrix equation based on the least squares positioning principle, the ranging residual vector and its root mean square value are calculated, and the root mean square value is compared with a preset residual threshold to determine whether there is a fault in the observation.

[0037] Specifically, in this embodiment, in step three, based on the least squares positioning principle, the observations are linearized into matrix equations. ,in The coefficient matrix, Let b be the least squares iteration vector and b be the distance residual vector; calculate the distance residual vector. ,in , for The identity matrix, Number of UWB positioning anchor points and ; through formula Calculate the root mean square value of the distance measurement residual. ;like If the residual exceeds the preset threshold, the observed measurement is determined to be faulty and an alarm flag is generated.

[0038] Matrix equations middle, The elements are composed of the direction cosines of each UWB positioning anchor point. , , The components are composed of direction cosine components. , , ,in( , , ) is the first The coordinates of the anchor points, ( , , ) is the first The coordinates of the target to be located are obtained in the next iteration. For the first The result of the second iteration and the first The straight-line distance between the anchor points.

[0039] For example, in step four, the dispersion of the target to be located is calculated based on the anchor point distance measurement value; the classification threshold of normal and abnormal data is obtained through the random forest algorithm, and normal data and abnormal data are distinguished according to the comparison result of the dispersion of the target to be located and the classification threshold.

[0040] Specifically, in this embodiment, in step four, the distance error is modeled and classified using a formula. Calculate the dispersion of the target to be located ,in The number of combinations of any three UWB positioning anchor points, ( , , ( ) represents the coordinates of the target to be located, calculated based on the distance measurements from all anchor points. , , The coordinates of the target to be located are calculated based on the distance measurements from any three anchor points; the classification thresholds for normal and abnormal data are obtained using the random forest algorithm. and according to and The comparison results divided the observations into normal data ( ) and abnormal data ( ).

[0041] For example, in step five, for normal observation data, a weighted least squares positioning algorithm with dynamically allocated weights is used for calculation; for abnormal observation data, the ranging error is first estimated by a biased Kalman filter that introduces a gross error discrimination factor, and then the error-compensated observations are substituted into the weighted least squares algorithm for calculation.

[0042] Specifically, in this embodiment, in step five, for normal observation data, a weighted least squares positioning algorithm is used to perform positioning calculation. Obtain the least squares iterative values, where the weighting matrix is... , This is a diagonal matrix of weight coefficients, where the weight coefficients are... , ( For pseudo-distance residuals, The distance measurement residual vector (Maximum value of component); For abnormal observation data, the ranging error is first estimated by biased Kalman filtering, and then the error-compensated observation data is substituted into the weighted least squares positioning algorithm for calculation to obtain the final positioning coordinates of the target to be located.

[0043] In step five, the state equation for the partial Kalman filter is: + The observation equation is , where the state vector , for Distance value at any time for Time-based ranging error, state transition matrix , The sampling interval is the sampling interval after linear interpolation of equal sampling intervals, and the observation matrix is... Gross error discriminant factor New information vector , for The covariance matrix is ​​obtained by weighting the covariance matrix of the observed noise vector. This enables accurate estimation of ranging errors.

[0044] In one embodiment, the indoor positioning network deployment is synchronized with the clock as follows: like Figure 2 As shown, an indoor positioning network is constructed, including repeaters and multiple levels of UWB positioning anchors. The network is divided into multiple links, each link containing multiple levels of UWB positioning anchors connected sequentially. Repeaters receive positioning signals from outdoor base stations via wired connections and forward them to the first-level UWB positioning anchor on each link. Figure 3 As shown, the first-level anchor point performs down-conversion, A / D sampling, signal acquisition, loop tracking, discipline frequency calibration, and synchronous broadcasting on the signal to generate a synchronous positioning signal.

[0045] Each anchor point transmits its synchronization positioning signal and the time of flight (measured bidirectionally via a UWB module with an accuracy ≤0.5ns) to the next anchor point. The next anchor point then performs clock correction on the received synchronization positioning signal based on the time of flight, thus completing clock synchronization for all anchor points sequentially. Different links use different UWB channels; if interference exists, time-division transmission is employed to ensure synchronization reliability.

[0046] In one embodiment, dual fault detection is as follows: like Figure 4 As shown, after clock synchronization is completed, the observations (anchor point 3D coordinates, distance values) between the point to be located and each anchor point are acquired. First, a distance residual consistency check is performed: a matrix equation is constructed. Calculate the distance measurement residual vector and its root mean square value .like If the preset threshold is exceeded, the observation is determined to be faulty.

[0047] Then, the dispersion of the point to be located is calculated: the coordinates of the point to be located are calculated based on all anchor points. , , The coordinates of the point to be located are calculated based on all three-point combinations. , , Substitute into the discreteness formula to calculate The classification threshold is obtained using the random forest algorithm. ,like If it is normal, then it is normal data; otherwise, it is abnormal data.

[0048] In one embodiment, the categorized localization solution is calculated as follows: For normal observation data, a weighted least squares localization algorithm is used, with weighting coefficients... ,in The contribution of each anchor point is dynamically adjusted, and the coordinates of the point to be located are calculated.

[0049] For anomalous observation data, the ranging error is first estimated using a biased Kalman filter. For example... Figure 5 As shown, the UWB positioning anchor point internally includes modules such as a signal processing unit, a clock correction unit, and a transmission time measurement unit. In the biased Kalman filter, a gross error discrimination factor is introduced. The observation noise covariance matrix is ​​obtained by weighting. This allows for accurate estimation of ranging errors. The error-compensated observations are then substituted into a weighted least squares algorithm to obtain the final positioning coordinates.

[0050] This application's embodiments improve the clock synchronization accuracy and fault data processing capabilities of the UWB indoor positioning system by employing UWB-based wireless cascaded clock synchronization, combined with dual fault detection and categorized positioning solutions, significantly enhancing positioning accuracy in complex indoor environments. The method in this application is implemented based on existing UWB hardware modules and can be widely applied to complex indoor scenarios, demonstrating good engineering practicality and promotional value.

[0051] By dividing the indoor positioning network into multiple multi-level anchor links, the repeater forwards the signal to the first anchor to generate a synchronization signal, and each level of anchor sequentially transmits the synchronization signal and the time-of-flight correction clock, which improves the wireless synchronization accuracy of UWB anchors and simplifies the system deployment and maintenance process.

[0052] By allocating different UWB channels to different links and using time-division multiplexing to transmit synchronous positioning signals when there is signal interference between links, the reliability of synchronous signal transmission in multi-link UWB positioning networks is improved, and synchronization errors caused by signal interference between links are avoided.

[0053] By calculating the dispersion of the target to be located, and combining it with the random forest algorithm to obtain the classification thresholds for normal and abnormal data, and distinguishing the data types accordingly, the detection accuracy and classification accuracy of fault data in UWB positioning observations are improved, and fault data is avoided from directly participating in the solution.

[0054] By using dynamic weighted least squares to solve normal observations and then applying biased Kalman filtering compensation with gross error discrimination factor to abnormal observations before solving, the utilization efficiency of UWB positioning data and the error compensation effect of abnormal data are improved, further enhancing positioning accuracy.

[0055] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0056] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A UWB indoor positioning method based on dual fault detection, characterized in that, Includes the following steps: Step 1: Build an indoor positioning network that includes repeaters and multi-level UWB positioning anchors, and use a UWB-based wireless cascaded clock synchronization method to synchronize the clocks of all anchors. Step 2: In an indoor positioning network that has completed clock synchronization, acquire the observations between the target to be located and each UWB positioning anchor point. Step 3: Based on the least squares positioning principle, perform an initial check on the distance residual consistency of the observations to determine whether there are any faults in the observations; Step 4: For the observations that were initially determined to have faults, calculate the dispersion of the target to be located and combine it with the random forest algorithm to classify the fault data and distinguish between normal data and abnormal data. Step 5: The normal data is solved using a dynamic weighted least squares algorithm, and the abnormal data is first compensated for by a biased Kalman filter, and then substituted into the weighted least squares algorithm to obtain the final positioning coordinates of the target to be located.

2. The UWB indoor positioning method based on dual fault detection according to claim 1, characterized in that, In step one, the indoor positioning network is divided into several links, each link containing multiple levels of UWB positioning anchors connected in sequence. The repeater forwards the positioning signal from the outdoor base station to the first-level anchor of each link. The first-level anchor performs preset processing on the positioning signal to generate a synchronous positioning signal. Each level of anchor sequentially sends its own synchronous positioning signal and the signal flight time to the next level anchor. The next level anchor performs clock correction on the received synchronous positioning signal according to the signal flight time.

3. The UWB indoor positioning method based on dual fault detection according to claim 1, characterized in that, In step one, different links correspond to different UWB channels. When there is signal interference between two links, the transmission of the synchronous positioning signal is completed in a time-division manner.

4. The UWB indoor positioning method based on dual fault detection according to claim 1, characterized in that, In step two, the acquired observations include the three-dimensional coordinates of each UWB positioning anchor point and the distance between the target to be positioned and each UWB positioning anchor point.

5. The UWB indoor positioning method based on dual fault detection according to claim 1, characterized in that, In step three, the observations are linearized into matrix equations based on the least squares positioning principle, the ranging residual vector and its root mean square value are calculated, and the root mean square value is compared with a preset residual threshold to determine whether there is a fault in the observations.

6. The UWB indoor positioning method based on dual fault detection according to claim 1, characterized in that, In step four, the dispersion of the target to be located is calculated based on the anchor point distance values; the classification thresholds for normal and abnormal data are obtained through the random forest algorithm, and normal and abnormal data are distinguished based on the comparison results between the dispersion of the target to be located and the classification thresholds.

7. The UWB indoor positioning method based on dual fault detection according to claim 1, characterized in that, In step five, for normal observation data, a weighted least squares positioning algorithm with dynamically allocated weights is used for calculation; for abnormal observation data, the ranging error is first estimated by a biased Kalman filter that introduces a gross error discrimination factor, and then the error-compensated observations are substituted into the weighted least squares algorithm for calculation.

Images