An indoor high-precision positioning system and method based on UWB technology

Abstract

The application discloses an indoor high-precision positioning system and method based on UWB technology, and particularly relates to the technical field of positioning, and comprises a region division module, based on the structural layout of a building and the coverage range of a UWB anchor node, indoor space is divided into N dynamic region units; a two-way mapping relationship between regions and UWB tags is established with key building nodes as boundaries; the movement trajectory data of tags associated with each region unit is introduced, and a spatial encoder is introduced to generate a region feature vector to obtain original three-dimensional coordinates to represent the positioning characteristics of the region; through analysis and processing of multi-source data of UWB signal layer data, motion perception layer data and environment perception layer data, the application effectively reduces the superimposed error of environmental reflection and hardware deviation, and can significantly improve the geometric convergence precision of position solution in a complex space.

Description

Technical Field

[0001] This invention relates to the field of positioning technology, and more specifically, to an indoor high-precision positioning system and method based on UWB technology. Background Technology

[0002] With the rapid development of the Internet of Things, smart buildings, and other related fields, the demand for indoor positioning technology is increasing. UWB technology has advantages such as extremely wide signal bandwidth, high time resolution, and strong anti-interference ability, showing great potential in the field of indoor positioning. Therefore, a high-precision indoor positioning system based on UWB technology is needed to meet the growing demand for indoor positioning technology.

[0003] Traditional indoor positioning systems include a signal transmitting module, a signal receiving module, and a positioning module. The signal transmitting module is responsible for transmitting signals for positioning. The signal receiving module is used to receive signals transmitted by the signal transmitting module. The positioning module obtains the location of the object or person being positioned based on the signal parameters obtained by the signal receiving module.

[0004] However, in practical use, it still has some drawbacks, such as low positioning accuracy. Traditional positioning technology is easily affected by multipath effects, signal interference, and other factors, resulting in large fluctuations in signal strength, which leads to increased positioning errors and makes it difficult to achieve high-precision positioning in complex environments. It also has weak anti-interference capabilities. The signal frequency band of traditional positioning technology is relatively narrow, making it susceptible to interference from other devices on the same frequency band, as well as the effects of reflection and refraction from obstacles in the indoor environment, which leads to signal distortion and increased positioning errors. Furthermore, it has poor real-time performance. Traditional positioning systems often require a long time for data acquisition, transmission, and processing, resulting in poor real-time performance of positioning results.

[0005] Therefore, there is an urgent need to provide an indoor high-precision positioning system and method based on UWB technology to solve the problems of low positioning accuracy, weak anti-interference ability and poor real-time performance of existing indoor positioning systems. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an indoor high-precision positioning system and method based on UWB technology, which solves the problems mentioned in the background art through the following solutions.

[0007] To achieve the above objectives, the present invention provides the following technical solution: an indoor high-precision positioning system and method based on UWB technology, comprising:

[0008] The area division module divides the indoor space into N dynamic area units based on the building's structural layout and the coverage of UWB anchor nodes. It establishes a two-way mapping relationship between the area and the UWB tag, using key building nodes as boundaries. Each area unit is associated with the movement trajectory data of the tag, and a spatial encoder is introduced to generate an area feature vector to obtain the original three-dimensional coordinates to characterize the positioning characteristics of the area.

[0009] The multi-source data acquisition module constructs a raw localization dataset for complex scenarios through heterogeneous sensors deployed in UWB anchor base stations, mobile tags, and environmental perception modules.

[0010] The data processing module establishes a mathematical model to calculate indicators by collecting the original positioning dataset from the multi-source data acquisition module, and obtains the UWB signal layer data influence coefficient, motion perception layer data influence coefficient, and environmental perception layer data influence coefficient, thus constructing three sets of core data features.

[0011] The comprehensive analysis module maps multi-source heterogeneous data into a unified positioning reliability index through weighted geometric fusion of the influence coefficients of UWB signal layer data, motion sensing layer data, and environmental sensing layer data.

[0012] The indoor positioning module dynamically allocates the ranging weights of multiple anchor points through the positioning confidence index, and combines them with a pre-trained residual correction model to achieve robust high-precision positioning and obtain the final coordinates.

[0013] The data interaction module transmits the final coordinates, the location credibility index, and related data from the original location dataset to the user information segment.

[0014] Preferably, the original localization dataset includes UWB signal layer data, motion sensing layer data, and environmental sensing layer data.

[0015] Preferably, the UWB signal layer data includes the peak amplitude of the CIR main path, denoted as A. peak The entropy of multipath energy distribution, denoted as H. mp Delay spread, denoted as τ rms The clock offset compensation amount is denoted as Δt. clock The motion sensing layer data includes IMU-predicted displacement, denoted as Δd. IMU UWB ranging displacement, denoted as Δd UWB The heading angle deviation, denoted as Δθ, and the environmental perception layer data, including the radio frequency noise level, denoted as N, are also included. rf Occlusion status code, denoted as O flag Channel occupancy rate, denoted as R ch .

[0016] Preferably, the UWB signal layer data influence coefficient is obtained by multiplying four parameters: multipath suppression weight, main path strength, delay spread control, and clock synchronization error. This normalizes the UWB signal characteristics of different dimensions into a reliability index of 0-1, reflecting the comprehensive attenuation of the signal due to multipath interference, hardware errors, and environmental interference. The ratio of the measured peak amplitude to the calibrated maximum value limits the influence of excessive signal amplitude. An exponential function penalizes the multipath entropy value; the larger the delay spread, the more significant the attenuation of signal quality by the denominator. Delay spread control amplifies the negative impact of clock skew with a large coefficient. Specifically:

[0017]

[0018] Preferably, the motion sensing layer data influence coefficient is obtained by superimposing a cosine penalty term for the heading angle deviation on the difference between the IMU predicted displacement and the UWB ranging displacement, thus fusing kinematic characteristics with wireless ranging results to reflect the consistency of sensor data. The heading deviation is smoothed by displacement difference calculation and a half-angle cosine function, suppressing outliers with the displacement difference term in reciprocal form. The heading deviation is attenuated by a cosine function. Specifically:

[0019]

[0020] Preferably, the environmental perception layer data influence coefficient is obtained by dynamically weighting the radio frequency noise intensity and the occlusion state, combining the channel occupancy rate to determine the noise weight ratio, quantifying the interference level of the environment on the UWB signal, normalizing the radio frequency noise intensity through a linear noise mapping, and then dynamically allocating and superimposing the weights for output, merging the noise and occlusion effects. Specifically:

[0021]

[0022] Preferably, the positioning reliability index first normalizes and truncates the influence coefficients of the UWB signal layer data, the motion sensing layer data, and the environmental sensing layer data, then assigns weights according to the index weights, and performs geometric fusion to achieve cross-level dimension normalization, comprehensively reflecting the dynamic reliability of the positioning system. Specifically:

[0023] LCI = SQI 0.4 ·MCI 0.3 ·(1-EIC) 0.2 .

[0024] Preferably, the specific operation steps of the indoor positioning module are as follows:

[0025] 1. Dynamic weight allocation: Based on the bidirectional mapping relationship between the dynamic region unit and the UWB label, and the region feature vector, the positioning reliability index corresponding to the ranging value of each dynamic region unit anchor point is obtained, and the weight is calculated. Specifically:

[0026]

[0027] LCI i represents the positioning reliability index of the i-th dynamic region unit anchor point, and N represents the total number of dynamic region unit anchor points.

[0028] 2. Based on the weights obtained in step 1 and the original 3D coordinates, a weighted localization solution is calculated. The least squares method is used to model and construct the objective function, yielding the estimated label coordinates. Specifically:

[0029]

[0030] Where the ωth i χ represents the weight of the anchor point of the i-th dynamic region unit, and χ represents the original 3D coordinates (x, y, z). i The known three-dimensional coordinates (x, y) of the anchor point of the i-th dynamic region unit. i y i , z i ), d i This represents the distance measurement between the anchor point and the label of the i-th dynamic region unit.

[0031] 3. Input the localization residual into the pre-trained LSTM network, output the coordinate compensation amount Δχ, and correct the final coordinates based on the coordinate compensation amount Δχ and the coordinates estimated by the labels obtained in S2.

[0032] The positioning residual includes the positioning residual vector and the environmental feature vector.

[0033] The specific location residual vector is as follows:

[0034] r = [r1, r2, ..., r N ],

[0035] in

[0036]

[0037] The environmental feature vector is specifically:

[0038] e = [SQI, EIC, MCI],

[0039] The coordinate compensation amount Δx is obtained by importing the positioning residual vector and the environmental feature vector into the residual input model, specifically:

[0040] Δx = LSTM(r, e),

[0041] The corrected final coordinates are obtained by adding the coordinate compensation amount to the estimated coordinates of the label, specifically:

[0042]

[0043] Preferably, an indoor high-precision positioning method based on UWB technology includes the following steps:

[0044] S1. Area division: Based on the building's structural layout and the coverage of UWB anchor nodes, the indoor space is divided into N dynamic area units; a two-way mapping relationship between the area and the UWB tag is established with key building nodes as boundaries; the movement trajectory data of the tag is associated with each area unit, and a spatial encoder is introduced to generate an area feature vector to obtain the original three-dimensional coordinates to characterize the positioning characteristics of the area.

[0045] S2. Multi-source data acquisition: Through heterogeneous sensors deployed in UWB anchor base stations, mobile tags, and environmental perception modules, a raw localization dataset for complex scenarios is constructed.

[0046] S3. Data processing: By establishing a mathematical model and calculating indicators based on the original positioning dataset collected in S2, the influence coefficients of the UWB signal layer data, the motion perception layer data, and the environmental perception layer data are obtained, and three sets of core features of the kernel data are constructed.

[0047] S4. Comprehensive analysis: Through weighted geometric fusion of the UWB signal layer data influence coefficient, motion sensing layer data influence coefficient, and environmental sensing layer data influence coefficient, multi-source heterogeneous data is mapped into a unified positioning reliability index.

[0048] S5. Indoor positioning: By dynamically allocating the ranging weights of multiple anchor points through the positioning reliability index and combining them with a pre-trained residual correction model, robust high-precision positioning is achieved to obtain the final coordinates.

[0049] S6. Data interaction: The final coordinates, positioning confidence index, and related data from the original positioning dataset are transmitted to the user information segment.

[0050] The technical effects and advantages of this invention are as follows:

[0051] 1. This invention analyzes and processes multi-source data, including UWB signal layer data, motion sensing layer data, and environmental sensing layer data, effectively reducing the superposition error caused by environmental reflection and hardware deviation, and significantly improving the accuracy of position calculation in complex spaces.

[0052] 2. This invention analyzes radio frequency noise and physical obstruction status in real time through a built-in environmental perception module, and uses a multi-sensor data cross-verification mechanism to automatically isolate abnormal signals in strong electromagnetic interference or sudden obstruction scenarios, ensuring positioning continuity and demonstrating anti-interference capabilities.

[0053] 3. This invention analyzes and judges the positioning reliability index by real-time detection and dynamic adjustment of multi-source data, and ensures high-frequency location updates through the real-time positioning. Attached Figure Description

[0054] Figure 1 This is a schematic diagram of the overall structure of the present invention.

[0055] Figure 2 This is a schematic diagram of the method structure of the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] As attached Figure 1 The indoor high-precision positioning system based on UWB technology shown includes:

[0058] The area division module divides the indoor space into N dynamic area units based on the building's structural layout and the coverage of UWB anchor nodes. It establishes a two-way mapping relationship between the area and the UWB tag, using key building nodes as boundaries. Each area unit is associated with the movement trajectory data of the tag, and a spatial encoder is introduced to generate an area feature vector to obtain the original three-dimensional coordinates to characterize the positioning characteristics of the area.

[0059] The multi-source data acquisition module constructs a raw localization dataset for complex scenarios through heterogeneous sensors deployed in UWB anchor base stations, mobile tags, and environmental perception modules.

[0060] In this embodiment, it should be specifically noted that the original localization dataset includes UWB signal layer data, motion sensing layer data, and environment sensing layer data.

[0061] In this embodiment, it should be specifically noted that the UWB signal layer data includes the peak amplitude of the CIR main path, denoted as A. peak The entropy of multipath energy distribution, denoted as H. mp Delay spread, denoted as τ rms The clock offset compensation amount is denoted as Δt. clock The motion sensing layer data includes IMU-predicted displacement, denoted as Δd. IMU UWB ranging displacement, denoted as Δd UWB The heading angle deviation, denoted as Δθ, and the environmental perception layer data, including the radio frequency noise level, denoted as N, are also included. rfOcclusion status code, denoted as O flag Channel occupancy rate, denoted as R ch .

[0062] In this embodiment, it is specifically explained that the peak amplitude of the CIR main path is obtained by receiving the UWB pulse signal transmitted by the tag through the UWB anchor base station in high-speed ADC sampling mode, capturing the complete channel impulse response waveform within the time window, recording the time-domain signal sequence, applying the sliding window peak detection algorithm, traversing the CIR time-domain signal sequence and identifying the main path signal point with the largest amplitude, and obtaining the peak amplitude of the CIR main path.

[0063] A peak =max(|s(t)|),

[0064] Where s(t) represents the original CIR time-domain signal sequence.

[0065] In this embodiment, it should be specifically explained that the multipath energy distribution entropy is obtained by performing wavelet decomposition on the CIR time-domain signal sequence to separate the direct path and multipath components, calculating the energy proportion of each multipath component, and then quantifying the randomness of the multipath energy distribution based on the Shannon entropy formula, specifically as follows:

[0066]

[0067] Where N1 represents the number of detected multipath components, p i This represents the energy percentage of the i-th path.

[0068] In this embodiment, it should be specifically explained that the delay spread is achieved by recording the time offset of each detected multipath component relative to the main path, and calculating the root mean square value of the delay using the energy of each path as a weight. Specifically:

[0069]

[0070] Where τ i Let N1 represent the delay of the i-th path, N1 represent the number of detected multipath components, τ represent the average delay, and E represent the delay of the i-th path. i Let i represent the path energy of the i-th path.

[0071] In this embodiment, it should be specifically explained that the clock offset compensation amount is calculated based on the signal propagation time using a two-way ranging protocol. The tag and the anchor base station perform a three-way handshake. The tag sends a Poll message and records the sending time. After receiving the message, the anchor base replies with a Response message, obtaining the recorded receiving time and recorded reply time. The tag receives the Response message and records the time. Frequency drift is eliminated through linear regression, and the relative clock offset is estimated, specifically as follows:

[0072]

[0073] Where t1 represents the recording sending time, t2 represents the recording receiving time, t3 represents the recording reply time, t4 represents the recording time, and f tag The local clock frequency of the tag, f anchor This indicates the local clock frequency of the base station.

[0074] In this embodiment, it should be specifically explained that the IMU predicted displacement by integrating an IMU sensor at the mobile tag end, using an SPI interface to collect the triaxial acceleration output by the IMU in real time, then performing double integration, accumulating the displacement components, and calculating the total displacement, specifically:

[0075]

[0076] a x [k] represents the x-axis acceleration at the k-th sampling point, a y [k] represents the y-axis acceleration at the k-th sampling point, and Δt represents the sampling time interval.

[0077] In this embodiment, it should be specifically explained that the UWB ranging displacement is acquired in real time by integrating an IMU sensor at the mobile tag end and using an SPI interface. The distance change of each anchor point is calculated by recording the UWB ranging values ​​at adjacent times. Specifically:

[0078]

[0079] Where d i t d represents the distance measured at the i-th anchor point at time t. i t-1 N represents the distance measurement value of the i-th anchor point at time t-1, and N2 represents the number of anchor points involved in the calculation.

[0080] In this embodiment, it should be specifically explained that the heading angle deviation is achieved by integrating an IMU sensor into the mobile tag, using an SPI interface to collect and fuse magnetometer and accelerometer data in real time to calculate the IMU heading angle, calculating the UWB heading angle based on continuous position coordinates, and taking the absolute difference between the two. Specifically:

[0081] Δθ=|θ mag -θ UWB |,

[0082] Where θmag represents the IMU heading angle, θ UWB This indicates the heading angle based on UWB.

[0083] In this embodiment, it should be specifically noted that the radio frequency noise level is measured by the background noise power spectral density through the spectrum analyzer module built into the anchor base station, and the noise energy is integrated over a UWB signal bandwidth of 500MHz. Specifically:

[0084]

[0085] Where f c Let f(f) represent the UWB center frequency, B represent the signal bandwidth, and P(f) represent the background noise power spectral density.

[0086] In this embodiment, it should be specifically explained that the occlusion status code is obtained by scanning the direction of the tag with radar mounted on the base station to acquire reflection point cloud data and count the number of reflection points per unit area. Specifically:

[0087]

[0088] in

[0089]

[0090] Where N points A represents the number of reflection points detected by the radar. scan ρ represents the area of ​​the radar scanned region. thresh This represents the occlusion threshold of the reflected point cloud density.

[0091] In this embodiment, it is specifically necessary to explain that the channel occupancy rate is calculated within a period when the channel power exceeds the threshold. Specifically:

[0092]

[0093] Where T busy T represents the cumulative time the channel has been occupied. total This indicates the total statistical time.

[0094] The data processing module establishes a mathematical model to calculate indicators by collecting the original positioning dataset from the multi-source data acquisition module, and obtains the UWB signal layer data influence coefficient, motion perception layer data influence coefficient, and environmental perception layer data influence coefficient, thus constructing three sets of core data features.

[0095] In this embodiment, it should be specifically explained that the UWB signal layer data influence coefficient is obtained by multiplying four parameters: multipath suppression weight, main path strength, delay spread control, and clock synchronization error. This normalizes the UWB signal characteristics of different dimensions into a reliability index of 0-1, reflecting the comprehensive attenuation of the signal due to multipath interference, hardware errors, and environmental interference. The ratio of the measured peak amplitude to the calibrated maximum value limits the influence of excessive signals. The exponential function penalizes the multipath entropy value. The larger the delay spread, the more significant the attenuation of signal quality by the denominator. Delay spread control amplifies the negative impact of clock offset by a large coefficient. Specifically:

[0096]

[0097] In this embodiment, it is specifically noted that the motion sensing layer data influence coefficient is obtained by using the difference between the IMU-predicted displacement and the UWB ranging displacement, superimposed with a cosine penalty term for the heading angle deviation, to fuse kinematic features with wireless ranging results, reflecting the consistency of sensor data. The heading deviation is smoothed using displacement difference calculation and a half-angle cosine function, suppressing outliers with the reciprocal form of the displacement difference term. The heading deviation is attenuated using a cosine function, specifically as follows:

[0098]

[0099] In this embodiment, it is specifically explained that the environmental perception layer data influence coefficient is obtained by dynamically weighting the radio frequency noise intensity and the occlusion state, combining the channel occupancy rate to determine the noise weight ratio, quantifying the interference level of the environment on the UWB signal, normalizing the radio frequency noise intensity through a linear noise mapping, and then performing dynamic weight allocation and superposition output to combine the noise and occlusion effects. Specifically:

[0100]

[0101] The comprehensive analysis module maps multi-source heterogeneous data into a unified positioning reliability index through weighted geometric fusion of the influence coefficients of UWB signal layer data, motion sensing layer data, and environmental sensing layer data.

[0102] In this embodiment, it should be specifically explained that the positioning reliability index first normalizes and truncates the influence coefficients of the UWB signal layer data, the motion sensing layer data, and the environmental sensing layer data, and then performs geometric fusion after weight allocation according to the index weights to achieve cross-level dimension normalization and comprehensively reflect the dynamic reliability of the positioning system. Specifically:

[0103] LCI = SQI 0.4 ·MCI 0.3 ·(1-EIC) 0.2 .

[0104] The indoor positioning module dynamically allocates the ranging weights of multiple anchor points through the positioning confidence index, and combines them with a pre-trained residual correction model to achieve robust high-precision positioning and obtain the final coordinates.

[0105] In this embodiment, the specific operation steps of the indoor positioning module are as follows:

[0106] 1. Dynamic weight allocation: Based on the bidirectional mapping relationship between the dynamic region unit and the UWB label, and the region feature vector, the positioning reliability index corresponding to the ranging value of each dynamic region unit anchor point is obtained, and the weight is calculated. Specifically:

[0107]

[0108] LCI i represents the positioning reliability index of the i-th dynamic region unit anchor point, and N represents the total number of dynamic region unit anchor points.

[0109] 2. Based on the weights obtained in step 1 and the original 3D coordinates, a weighted localization solution is calculated. The least squares method is used to model and construct the objective function, yielding the estimated label coordinates. Specifically:

[0110]

[0111] Where the ωth i χ represents the weight of the anchor point of the i-th dynamic region unit, and χ represents the original 3D coordinates (x, y, z). i The known three-dimensional coordinates (x, y) of the anchor point of the i-th dynamic region unit. i y i , z i ), d i This represents the distance measurement between the anchor point and the label of the i-th dynamic region unit.

[0112] 3. Input the localization residual into the pre-trained LSTM network, output the coordinate compensation amount Δχ, and correct the final coordinates based on the coordinate compensation amount Δχ and the coordinates estimated by the labels obtained in S2.

[0113] The positioning residual includes the positioning residual vector and the environmental feature vector.

[0114] The specific location residual vector is as follows:

[0115] r = [r1, r2, ..., r N ],

[0116] in

[0117]

[0118] The environmental feature vector is specifically:

[0119] e = [SQI, EIC, MCI],

[0120] The coordinate compensation amount Δx is obtained by importing the positioning residual vector and the environmental feature vector into the residual input model, specifically:

[0121] Δx = LSTm(r, e),

[0122] The corrected final coordinates are obtained by adding the coordinate compensation amount to the estimated coordinates of the label, specifically:

[0123]

[0124] The data interaction module transmits the final coordinates, the location credibility index, and related data from the original location dataset to the user information segment.

[0125] refer to Figure 2 In this embodiment, it should be specifically noted that the present invention provides an indoor high-precision positioning method based on UWB technology, which includes the following steps:

[0126] S1. Area division: Based on the building's structural layout and the coverage of UWB anchor nodes, the indoor space is divided into N dynamic area units; a two-way mapping relationship between the area and the UWB tag is established with key building nodes as boundaries; the movement trajectory data of the tag is associated with each area unit, and a spatial encoder is introduced to generate an area feature vector to obtain the original three-dimensional coordinates to characterize the positioning characteristics of the area.

[0127] S2. Multi-source data acquisition: Through heterogeneous sensors deployed in UWB anchor base stations, mobile tags, and environmental perception modules, a raw localization dataset for complex scenarios is constructed.

[0128] S3. Data processing: By establishing a mathematical model and calculating indicators based on the original positioning dataset collected in S2, the influence coefficients of the UWB signal layer data, the motion perception layer data, and the environmental perception layer data are obtained, and three sets of core features of the kernel data are constructed.

[0129] S4. Comprehensive analysis: Through weighted geometric fusion of the UWB signal layer data influence coefficient, motion sensing layer data influence coefficient, and environmental sensing layer data influence coefficient, multi-source heterogeneous data is mapped into a unified positioning reliability index.

[0130] S5. Indoor positioning: By dynamically allocating the ranging weights of multiple anchor points through the positioning reliability index and combining them with a pre-trained residual correction model, robust high-precision positioning is achieved to obtain the final coordinates.

[0131] S6. Data interaction: The final coordinates, positioning confidence index, and related data from the original positioning dataset are transmitted to the user information segment.

[0132] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0133] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An indoor high-precision positioning system based on UWB technology, characterized in that, include: The zone division module divides the interior space into N dynamic zone units based on the building's structural layout and the coverage of UWB anchor nodes. Using key building nodes as boundaries, a two-way mapping relationship between regions and UWB tags is established; the movement trajectory data of each region unit is associated with the tag, and a spatial encoder is introduced to generate region feature vectors to obtain the original three-dimensional coordinates to characterize the positioning characteristics of the region. The multi-source data acquisition module constructs a raw localization dataset for complex scenarios through heterogeneous sensors deployed in UWB anchor base stations, mobile tags, and environmental perception modules. The original localization dataset includes UWB signal layer data, motion sensing layer data, and environmental sensing layer data. The UWB signal layer data includes the peak amplitude of the CIR main path, denoted as A. peak The entropy of multipath energy distribution, denoted as H. mp Delay spread, denoted as τ rms The clock offset compensation amount is denoted as Δt. clock The motion sensing layer data includes IMU-predicted displacement, denoted as Δd. IMU UWB ranging displacement, denoted as Δd UWB The heading angle deviation is denoted as Δθ; the environmental perception layer data includes the radio frequency noise level, denoted as N. rf Occlusion status code, denoted as O flag Channel occupancy rate, denoted as R ch ; The data processing module establishes a mathematical model to calculate indicators by collecting the original positioning dataset from the multi-source data acquisition module, and obtains the UWB signal layer data influence coefficient, motion perception layer data influence coefficient and environmental perception layer data influence coefficient, and constructs three sets of core data features. The UWB signal layer data influence coefficient is SQI, the motion sensing layer data influence coefficient is MCI, and the environment sensing layer data influence coefficient is EIC. The comprehensive analysis module maps multi-source heterogeneous data into a unified positioning reliability index through weighted geometric fusion of the influence coefficients of UWB signal layer data, motion sensing layer data, and environmental sensing layer data. The indoor positioning module dynamically allocates the ranging weights of multiple anchor points through the positioning confidence index, and combines it with a pre-trained residual correction model to achieve robust high-precision positioning and obtain the final coordinates. The data interaction module transmits the final coordinates, positioning confidence index, and related data from the original positioning dataset to the user information terminal.

2. The indoor high-precision positioning system based on UWB technology according to claim 1, characterized in that: The UWB signal layer data influence coefficient is obtained by multiplying four parameters: multipath suppression weight, main path strength, delay spread control, and clock synchronization error. This normalizes the UWB signal characteristics of different dimensions into a 0-1 reliability index, reflecting the combined attenuation of the signal due to multipath interference, hardware errors, and environmental disturbances. The ratio of the measured peak amplitude to the calibrated maximum value limits the influence of excessive signal amplitude. An exponential function penalizes the multipath entropy value; the larger the delay spread, the more significant the attenuation of signal quality by the denominator. Delay spread control amplifies the negative impact of clock skew with a large coefficient. Specifically: 。 3. The indoor high-precision positioning system based on UWB technology according to claim 1, characterized in that: The motion sensing layer data influence coefficient is obtained by using the difference between the IMU predicted displacement and the UWB ranging displacement, and superimposing a cosine penalty term for the heading angle deviation. This fuses kinematic characteristics with wireless ranging results, reflecting the consistency of sensor data. The heading deviation is smoothed by displacement difference calculation and a half-angle cosine function. The displacement difference term is used to suppress outliers in the reciprocal form, and the heading deviation is attenuated by a cosine function. Specifically: 。 4. The indoor high-precision positioning system based on UWB technology according to claim 1, characterized in that: The environmental perception layer data influence coefficient is obtained by dynamically weighting the radio frequency noise intensity and the occlusion status, combining the channel occupancy rate to determine the noise weight ratio, quantifying the interference level of the environment on the UWB signal, normalizing the radio frequency noise intensity through a linear noise mapping, and then dynamically allocating and superimposing the weights for output, merging the noise and occlusion effects. Specifically: 。 5. The indoor high-precision positioning system based on UWB technology according to claim 1, characterized in that: The positioning reliability index first normalizes and truncates the influence coefficients of the UWB signal layer data, motion sensing layer data, and environmental sensing layer data. Then, it performs geometric fusion after weighting according to the index weights to achieve cross-level dimension normalization and comprehensively reflect the dynamic reliability of the positioning system. Specifically: 。 6. The indoor high-precision positioning system based on UWB technology according to claim 1, characterized in that: The specific operation steps of the indoor positioning module are as follows: S1. Dynamic weight allocation: Based on the bidirectional mapping relationship between the dynamic region unit and the UWB label, and the region feature vector, the positioning reliability index corresponding to the ranging value of each dynamic region unit anchor point is obtained, and the weight is calculated. Specifically: ， LCI i represents the positioning reliability index of the i-th dynamic region unit anchor point, and N represents the total number of dynamic region unit anchor points; S2. Based on the weighted localization solution obtained in step 1 and the original 3D coordinates, the objective function is constructed using the least squares method to obtain the estimated label coordinates, specifically: ， Where the ωth i χ represents the weight of the anchor point of the i-th dynamic region unit, and χ represents the original 3D coordinates (x, y, z). i Represents the known three-dimensional coordinates (x, y) of the anchor point of the i-th dynamic region unit. i y i , z i ), d i This represents the distance measurement between the anchor point and the label of the i-th dynamic region unit; S3. Input the positioning residual into the pre-trained LSTM network, output the coordinate compensation amount Δχ, and correct the final coordinates based on the coordinate compensation amount Δχ and the coordinates estimated by the label obtained in S2. The positioning residual includes the positioning residual vector and the environmental feature vector; The specific location residual vector is as follows: ， in ， The environmental feature vector is specifically: ， The coordinate compensation amount Δx is obtained by importing the positioning residual vector and the environmental feature vector into the residual input model, specifically: ， The corrected final coordinates are obtained by adding the coordinate compensation amount to the estimated coordinates of the label, specifically: 。 7. An indoor high-precision positioning method based on UWB technology, used in conjunction with the indoor high-precision positioning system based on UWB technology as described in any one of claims 1-6, characterized in that, Includes the following steps: S1. Area division: Based on the building's structural layout and the coverage of UWB anchor nodes, the indoor space is divided into N dynamic area units; a two-way mapping relationship between the area and the UWB tag is established with key building nodes as boundaries; the movement trajectory data of the tag is associated with each area unit, and a spatial encoder is introduced to generate an area feature vector to obtain the original three-dimensional coordinates to characterize the positioning characteristics of the area. S2. Multi-source data acquisition: Through heterogeneous sensors deployed in UWB anchor base stations, mobile tags, and environmental perception modules, a raw localization dataset for complex scenarios is constructed. S3. Data processing: By establishing a mathematical model and calculating indicators based on the original positioning dataset collected in S2, the influence coefficients of the UWB signal layer data, the motion perception layer data, and the environmental perception layer data are obtained, and three sets of core features of the kernel data are constructed. S4. Comprehensive analysis: Through weighted geometric fusion of the UWB signal layer data influence coefficient, motion sensing layer data influence coefficient, and environmental sensing layer data influence coefficient, multi-source heterogeneous data is mapped into a unified positioning reliability index. S5. Indoor positioning: By dynamically allocating the ranging weights of multiple anchor points through the positioning reliability index and combining them with a pre-trained residual correction model, robust high-precision positioning is achieved to obtain the final coordinates. S6. Data interaction: The final coordinates, positioning confidence index, and related data from the original positioning dataset are transmitted to the user information terminal.

Images