Indoor positioning system based on ultra-wideband communication technology

Abstract

The invention discloses an indoor positioning system based on an ultra-wideband communication technology. The indoor positioning system comprises a base station, a label and an upper computer, the upper computer carries out data interaction with the data processing and storage module, the monitoring and data display module and the communication module, the upper computer receives data from a base station, the data processing and storage module extracts positioning information and comprises a preliminary positioning stage and an accurate positioning and error correction stage, and the positioning stage utilizes a trilateration method to realize accurate positioning and error correction. Calculating the initial position of the label according to the distance information between the label and the at least three base stations, and then further optimizing the initial position by utilizing a weighted centroid algorithm and combining the position information and the relative distance of the plurality of base stations to obtain an estimated position; in the accurate positioning and error correction stage, the position obtained in the initial positioning stage is used as the initial position estimation of a Taylor series expansion method, and the Taylor series expansion method is utilized to gradually approach the real position in an iterative approximation mode.

Description

Technical Field

[0001] The present invention relates to the technical field of indoor positioning, and in particular to an indoor positioning system based on ultra-wideband communication technology. Background Art

[0002] The primary objective of an ultra-wideband indoor positioning system is to achieve high-precision positioning. By adopting broadband narrow-pulse communication technology, multi-source data fusion, and time-series signal processing technology, the system can extract the first-arrival path signal, thereby reducing the positioning error and improving the positioning accuracy. The ultra-wideband indoor positioning system can be applied to multiple fields, such as precious item warehousing, mine personnel positioning, robot motion tracking, and car basement parking. These applications are of great significance for improving production efficiency, ensuring personnel safety, and optimizing management processes.

[0003] Underground parking lots usually have a relatively low floor height and dense columns, which will exacerbate the multi-path propagation of UWB signals. In addition, the movement of vehicles and personnel will further interfere with the signal stability, making positioning more difficult. Therefore, an indoor positioning system based on ultra-wideband communication technology is proposed. Summary of the Invention

[0004] The purpose of the present invention is to solve the disadvantages existing in the prior art that underground parking lots usually have a relatively low floor height and dense columns, which will exacerbate the multi-path propagation of UWB signals. In addition, the movement of vehicles and personnel will further interfere with the signal stability, making positioning more difficult, and to propose an indoor positioning system based on ultra-wideband communication technology.

[0005] To achieve the above objective, the present invention adopts the following technical solutions:

[0006] An indoor positioning system based on ultra-wideband communication technology, including a base station, a tag, and a host computer;

[0007] Data interaction occurs between the base station and a signal transmitting and receiving module, a signal processing and optimization module, and a communication module. The signal transmitting and receiving module built in the base station is responsible for generating and transmitting UWB pulse signals, and the UWB pulse signals are received by the tag and used for ranging. The signal transmitting and receiving module of the base station is responsible for receiving the response signals from the tag to complete the two-way ranging process. The base station optimizes the received signals through the signal processing and optimization module. The signal processing and optimization module establishes a machine learning model by collecting environmental characteristic data (such as column positions, wall materials, bend curvatures, etc.) and corresponding positioning parameter data of each area in the underground parking lot. The machine learning model learns and identifies the influence of different environmental characteristics on the positioning accuracy, and dynamically adjusts positioning parameters such as transmission power and receiving sensitivity according to the environmental characteristics of the current position;

[0008] Data is exchanged between the tag, the signal transmitting and receiving module, and the communication module. After receiving the signal from the base station, the signal transmitting and receiving module connected to the tag transmits a response signal.

[0009] Data is exchanged between the host computer, the data processing and storage module, the monitoring and data display module, and the communication module. The host computer receives data from the base station. The data processing and storage module processes, analyzes, and stores the received data, and extracts positioning information. The monitoring and data display module displays the position information of the tag and monitors the system status. The positioning information extracted by the data processing and storage module includes a preliminary positioning stage and an accurate positioning and error correction stage. In the preliminary positioning stage, trilateration is used to calculate the preliminary position of the tag based on the distance information between the tag and at least three base stations. Then, the weighted centroid algorithm is used to further optimize the preliminary position by combining the position information and relative distances of multiple base stations to obtain an estimated position. In the accurate positioning and error correction stage, the position obtained in the preliminary positioning stage is used as the initial position estimate for the Taylor series expansion method. The Taylor series expansion method is used to gradually approximate the true position through iterative approximation.

[0010] The above technical solution further includes:

[0011] Further, the signal transmitting and receiving module includes a signal transmitting unit and a signal receiving unit. The signal transmitting unit is responsible for generating and transmitting UWB pulse signals. These UWB pulse signals have an ultra-narrow pulse width at the sub-nanosecond level, can propagate quickly in the air, and carry the time information required for positioning. The signal receiving unit is responsible for receiving UWB pulse signals from the base station or the tag.

[0012] Further, the signal processing and optimization module includes a data acquisition unit, a preprocessing unit, a feature extraction unit, a model unit, and a parameter adjustment unit. The data acquisition unit is responsible for collecting environmental characteristic data of each area in the underground parking lot, such as column positions, wall materials, and bend curvatures, and also collecting corresponding positioning parameter data, such as received signal strength, phase information, and time of arrival. The preprocessing unit preprocesses the collected data, including data cleaning, denoising, format conversion, etc. The feature extraction unit extracts features that have an important impact on positioning accuracy from the preprocessed data, such as the multipath components of the signal and the quantization indexes of environmental characteristics. The results of feature extraction will be used as the input of the machine learning model. The model unit trains the machine learning model based on the extracted features, learns and identifies the impact of different environmental characteristics on positioning accuracy. After the model training is completed, the positioning parameters are dynamically adjusted according to the environmental characteristics of the current position. The parameter adjustment unit adjusts the parameters in the positioning algorithm, such as transmission power and receiving sensitivity, according to the output results of the machine learning model.

[0013] Furthermore, the feature extraction unit extracts features that have an important impact on the positioning accuracy from the preprocessed data. The features include signal strength, time of arrival, time difference of arrival, multipath components, and quantization indicators of environmental characteristics. The signal strength reflects the attenuation of the signal during propagation and can be used to preliminarily judge the signal quality. The time of arrival (TOA): the transmission time of the signal from the base station to the tag, which is a key parameter for calculating the distance. The time difference of arrival (TDOA): the time difference between different base stations receiving the same tag signal, which can be used to determine the relative position of the tag. The multipath components: the reflection, refraction, etc. phenomena that occur when the signal encounters obstacles (such as walls, columns) during propagation, and these components will affect the positioning accuracy. The quantization indicators of environmental characteristics: such as the layout, materials, and obstacle distribution of the indoor space, and these characteristics can be obtained through pre-measurement or modeling and used as input features of the machine learning model.

[0014] Furthermore, the model unit trains a machine learning model based on the extracted features, learns and identifies the impact of different environmental characteristics on the positioning accuracy. After the model training is completed, the positioning parameters are dynamically adjusted according to the environmental characteristics of the current position. The specific steps are as follows:

[0015] Data preparation and feature extraction: Prepare the training dataset, which includes the original data collected by the base station and the tag, and the corresponding positioning results obtained through geometric algorithms or other methods. Subsequently, use the feature extraction unit to extract useful features from these original data, and these features will be used as the input of the random forest model;

[0016] Random forest model training:

[0017] Random sampling: Randomly draw multiple subsets from the training dataset, and each subset contains a part of the samples in the original dataset;

[0018] Construct decision trees: For each tree in the random forest, use the Bootstrap method to draw a training set with the same size as the original dataset from the training dataset. At each split point, randomly select a part of the features for investigation, and select the optimal feature according to the Gini impurity to split the node. Repeat the above process until the stopping condition is met. The Gini impurity is expressed as where c is the number of classes, and p i is the probability of the i-th class. The value of the Gini impurity ranges from 0 to 1, and the smaller the value, the purer the dataset;

[0019] Integrated prediction: For the new input sample, input it into all decision trees for prediction, and synthesize the prediction results of all trees to obtain the final prediction result;

[0020] Model evaluation and optimization: After training is completed, the random forest model is evaluated to check its performance. The training dataset is divided into multiple parts, and one part is used as the test set in turn, while the remaining parts are used as the training set for model training and testing. According to the evaluation results, the model is optimized, such as adjusting parameters such as the number of decision trees, the maximum depth, and the minimum number of samples for splitting;

[0021] Online positioning and parameter adjustment: When a new UWB signal is received, the feature extraction unit extracts the features of the signal and then inputs them into the random forest model for prediction. The model dynamically adjusts the positioning parameters (such as the weights and thresholds of decision trees) according to the environmental characteristics of the current location to improve the positioning accuracy.

[0022] Furthermore, the data processing and storage module includes a data receiving unit, a preprocessing unit, a preliminary positioning calculation unit, a precise positioning and error correction unit, a data storage unit, and a data interaction interface unit. The data receiving unit is responsible for receiving data from the base stations and tags, including the location information of the base stations, the distance information between the tags and the base stations, the signal strength, etc. The preprocessing unit preprocesses the received data, including data cleaning (removing noise, outliers, etc.), format conversion, etc., to ensure the quality of the data for subsequent processing. The preliminary positioning calculation unit uses the trilateration method to calculate the preliminary position of the tag based on the distance information between the tag and at least three base stations. Subsequently, using the weighted centroid algorithm, combined with the location information and relative distances (or signal strengths) of multiple base stations, the preliminary position is further optimized to obtain the estimated position. The precise positioning and error correction unit uses the position obtained by the preliminary positioning calculation unit as the initial position estimate of the Taylor series expansion method, and uses the Taylor series expansion method to gradually approach the true position through iterative approximation to achieve positioning. By introducing a neural network, the positioning error is corrected to further improve the positioning accuracy. The data storage unit is responsible for storing the processed positioning data, base station information, tag information, etc. for subsequent query and analysis. The data interaction interface unit provides a data interaction interface with other modules to achieve data sharing and synchronization.

[0023] Furthermore, the preliminary positioning calculation unit uses the trilateration method to calculate the preliminary position of the tag based on the distance information between the tag and at least three base stations. The specific steps are as follows:

[0024] Suppose there are three base stations A, B, and C, and their position coordinates are (x A , y A ), (x B , y B ), (x C , y C ) respectively;

[0025] The distances between the tag and base stations A, B, and C are d A , d B , d C respectively. The distance is calculated based on the transmission time of the UWB signal and the signal propagation speed. The distance formula is expressed as d = c × TOA, where d is the distance between the tag and the base station, c is the propagation speed of the UWB signal in the air (about the speed of light), and TOA is the transmission time of the signal from the base station to the tag;

[0026] Taking base station A as the center, draw a circle with radius d A ; taking base station B as the center, draw a circle with radius d B ; taking base station C as the center, draw a circle with radius d C . The intersection of these three circles is the preliminary position of the tag. The geometric equation of the trilateration method is

[0027] Furthermore, using the weighted centroid algorithm, combining the position information and relative distances (or signal strengths) of multiple base stations, further optimize the preliminary position to obtain the estimated position. The specific steps are as follows:

[0028] For each base station, calculate a weighting factor according to its position coordinates and the relative distance (or signal strength) from the tag;

[0029] Perform a weighted average of the weighting factors of all base stations and their position coordinates to obtain the optimized tag position. The weighted centroid formula is expressed as where (x', y') are the coordinates of the optimized tag position, (x i , y i ) are the position coordinates of the i-th base station, and w i is the weighting factor of the i-th base station. The weighting factor can be calculated based on the relative distance or signal strength.

[0030] Furthermore, the precise positioning and error correction unit uses the position obtained by the preliminary positioning calculation unit as the initial position estimate of the Taylor series expansion method. Using the Taylor series expansion method, gradually approach the true position through iterative approximation to achieve positioning. The specific steps are as follows:

[0031] Step 1: Set the initial position estimate value (x0, y0), which is obtained by the preliminary positioning calculation unit;

[0032] Step 2: According to the UWB signal transmission time and base station position information, construct a non-linear equation system to represent the distance relationship between the true position of the tag and the base stations;

[0033] Step 3: Perform a Taylor series expansion on the non-linear equation system to obtain a linearized equation system;

[0034] Step 4: Solve the linearized equations to obtain the position correction amounts Δx and Δy;

[0035] Step 5: Update the position estimate: x1 = x0 + Δx, y1 = y0 + Δy;

[0036] Step 6: Repeat Step 3, Step 4, and Step 5 until the iteration termination condition is satisfied (such as the position correction amount is less than a preset threshold).

[0037] The present invention has the following beneficial effects:

[0038] 1. In the present invention, using the trilateration method, based on the distance information between the tag and at least three base stations, the preliminary position of the tag is calculated. The trilateration method can provide a relatively accurate preliminary positioning result, laying a foundation for the subsequent steps. Taking the position obtained in the preliminary positioning stage as the initial position estimate of the Taylor series expansion method, and using the Taylor series expansion method, gradually approaching the true position through iterative approximation can effectively reduce the errors that may exist in the preliminary positioning stage, improve the accuracy of the final positioning result, and achieve more accurate and stable positioning in multi-layer and complex-structured underground parking lots.

[0039] 2. In the present invention, the base station collects environmental characteristic data of each area in the underground parking lot, such as column positions, wall materials, bend curvatures, etc., and corresponding positioning parameter data through the signal processing and optimization module, and uses a machine learning model to learn and identify the influence of different environmental characteristics on the positioning accuracy, so as to dynamically adjust positioning parameters, such as transmission power, reception sensitivity, etc., according to the environmental characteristics of the current position. This helps to reduce positioning errors caused by environmental differences and thus improve the positioning accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a system block diagram of an indoor positioning system based on ultra-wideband communication technology proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0042] Please refer to Figure 1 As shown, the present invention is an indoor positioning system based on ultra-wideband communication technology, including a base station, a tag, and a host computer;

[0043] Data interaction occurs between the base station, the signal transmitting and receiving module, the signal processing and optimization module, and the communication module. The signal transmitting and receiving module built into the base station is responsible for generating and transmitting UWB pulse signals. The UWB pulse signals are received by the tag and used for ranging. The signal transmitting and receiving module of the base station is responsible for receiving the response signals from the tag to complete the two-way ranging process. The base station optimizes the received signals through the signal processing and optimization module. The signal processing and optimization module establishes a machine learning model by collecting environmental characteristic data (such as column positions, wall materials, bend curvatures, etc.) and corresponding positioning parameter data in each area of the underground parking lot. The machine learning model learns and identifies the impact of different environmental characteristics on the positioning accuracy, and dynamically adjusts positioning parameters such as transmission power and reception sensitivity according to the environmental characteristics of the current location;

[0044] Data interaction occurs between the tag, the signal transmitting and receiving module, and the communication module. After receiving the signal from the base station, the signal transmitting and receiving module connected to the tag transmits a response signal;

[0045] Data interaction occurs between the host computer, the data processing and storage module, the monitoring and data display module, and the communication module. The host computer receives data from the base station. The data processing and storage module processes, analyzes, and stores the received data, and extracts positioning information. The monitoring and data display module displays the position information of the tag and monitors the system status. The positioning information extracted by the data processing and storage module includes the preliminary positioning stage and the precise positioning and error correction stage. In the preliminary positioning stage, the trilateration method is used to calculate the preliminary position of the tag based on the distance information between the tag and at least three base stations. Then, the weighted centroid algorithm is used to further optimize the preliminary position by combining the position information and relative distances of multiple base stations to obtain an estimated position. In the precise positioning and error correction stage, the position obtained in the preliminary positioning stage is used as the initial position estimate of the Taylor series expansion method. The Taylor series expansion method is used to gradually approximate the true position through an iterative approximation method.

[0046] In one embodiment, the signal transmitting and receiving module includes a signal transmitting unit and a signal receiving unit. The signal transmitting unit is responsible for generating and transmitting UWB pulse signals. These UWB pulse signals have an ultra-narrow pulse width of sub-nanosecond level, can propagate quickly in the air, and carry the time information required for positioning. The signal receiving unit is responsible for receiving UWB pulse signals from the base station or the tag.

[0047] In one embodiment, the signal processing and optimization module includes a data acquisition unit, a preprocessing unit, a feature extraction unit, a model unit, and a parameter adjustment unit. The data acquisition unit is responsible for collecting environmental characteristic data of various areas in the underground parking lot, such as column positions, wall materials, bend curvatures, etc., and also collects corresponding positioning parameter data, such as received signal strength, phase information, time of arrival, etc. The preprocessing unit preprocesses the collected data, including data cleaning, denoising, format conversion, etc. The feature extraction unit extracts features that have an important impact on positioning accuracy from the preprocessed data, such as multipath components of signals, quantization indexes of environmental characteristics, etc. The result of feature extraction will be used as the input of the machine learning model. The model unit trains the machine learning model based on the extracted features, learns and identifies the impact of different environmental characteristics on positioning accuracy. After the model training is completed, the positioning parameters are dynamically adjusted according to the environmental characteristics of the current position. The parameter adjustment unit adjusts the parameters in the positioning algorithm, such as transmission power, reception sensitivity, etc., according to the output result of the machine learning model.

[0048] In one embodiment, the feature extraction unit extracts features that have an important impact on positioning accuracy from the preprocessed data. The features include signal strength, time of arrival, time difference of arrival, multipath components, and quantization indexes of environmental characteristics. The signal strength reflects the attenuation of the signal during propagation and can be used to preliminarily judge the quality of the signal. The time of arrival (TOA): the transmission time of the signal from the base station to the tag, which is a key parameter for calculating the distance. The time difference of arrival (TDOA): the time difference between different base stations receiving the same tag signal, which can be used to determine the relative position of the tag. The multipath components: the reflection, refraction, etc. phenomena that occur when the signal encounters obstacles (such as walls, columns) during propagation. These components will affect the accuracy of positioning. The quantization indexes of environmental characteristics: such as the layout, materials, and obstacle distribution of the indoor space. These characteristics can be obtained through pre-measurement or modeling and used as input features of the machine learning model.

[0049] In one embodiment, the model unit trains the machine learning model based on the extracted features, learns and identifies the impact of different environmental characteristics on positioning accuracy. After the model training is completed, the positioning parameters are dynamically adjusted according to the environmental characteristics of the current position. The specific steps are as follows:

[0050] Data preparation and feature extraction: Prepare the training data set, which includes the original data collected by the base station and the tag, and the corresponding positioning results obtained through geometric algorithms or other methods. Subsequently, use the feature extraction unit to extract useful features from these original data. These features will be used as the input of the random forest model;

[0051] Random forest model training:

[0052] Random sampling: Randomly extract multiple subsets from the training dataset, and each subset contains a part of the samples in the original dataset;

[0053] Constructing a decision tree: For each tree in the random forest, extract a training set with the same size as the original dataset from the training dataset using the Bootstrap method. At each split point, randomly select a part of the features for examination, and select the optimal feature according to the Gini impurity to split the node. Repeat the above process until the stopping condition is met. The Gini impurity is expressed as where c is the number of classes, and p i is the probability of the i-th class. The value of the Gini impurity ranges from 0 to 1, and the smaller the value, the purer the dataset;

[0054] Ensemble prediction: For a new input sample, input it into all decision trees for prediction, and synthesize the prediction results of all trees to obtain the final prediction result;

[0055] Model evaluation and optimization: After training, evaluate the random forest model to check its performance. Divide the training dataset into multiple parts, and take one part as the test set in turn, and the rest as the training set for model training and testing. According to the evaluation results, optimize the model, such as adjusting parameters such as the number of decision trees, the maximum depth, and the minimum number of samples for splitting;

[0056] Online positioning and parameter adjustment: When a new UWB signal is received, use the feature extraction unit to extract the features of the signal, and then input it into the random forest model for prediction. The model dynamically adjusts the positioning parameters (such as the weights and thresholds of decision trees) according to the environmental characteristics of the current location to improve the accuracy of positioning.

[0057] In one embodiment, the data processing and storage module includes a data receiving unit, a preprocessing unit, a preliminary positioning calculation unit, an accurate positioning and error correction unit, a data storage unit, and a data interaction interface unit. The data receiving unit is responsible for receiving data from base stations and tags, including the position information of base stations, the distance information between tags and base stations, signal strength, etc. The preprocessing unit preprocesses the received data, including data cleaning (removing noise, outliers, etc.), format conversion, etc., to ensure the quality of the data for subsequent processing. The preliminary positioning calculation unit uses the trilateration method to calculate the preliminary position of the tag based on the distance information between the tag and at least three base stations. Subsequently, using the weighted centroid algorithm, combining the position information and relative distances (or signal strengths) of multiple base stations, the preliminary position is further optimized to obtain an estimated position. The accurate positioning and error correction unit takes the position obtained by the preliminary positioning calculation unit as the initial position estimate of the Taylor series expansion method, and uses the Taylor series expansion method to gradually approximate the true position through iterative approximation to achieve positioning. By introducing a neural network, the positioning error is corrected to further improve the positioning accuracy. The data storage unit is responsible for storing processed positioning data, base station information, tag information, and other data for subsequent query and analysis. The data interaction interface unit provides a data interaction interface with other modules to achieve data sharing and synchronization.

[0058] In one embodiment, the preliminary positioning calculation unit uses the trilateration method to calculate the preliminary position of the tag based on the distance information between the tag and at least three base stations. The specific steps are as follows:

[0059] Suppose there are three base stations A, B, and C, and their position coordinates are (x A , y A ), (x B , y B ), (x C , y C );

[0060] The distances between the tag and base stations A, B, and C are d A , d B , d C respectively. The distance is calculated through the transmission time of the UWB signal and the signal propagation speed. The distance formula is expressed as d = c × TOA, where d is the distance between the tag and the base station, c is the propagation speed of the UWB signal in the air (about the speed of light), and TOA is the transmission time of the signal from the base station to the tag;

[0061] Draw a circle with base station A as the center and d A as the radius; draw a circle with base station B as the center and d B as the radius; draw a circle with base station C as the center and d CDraw circles with this radius. The intersection points of these three circles are the preliminary positions of the tags. The geometric equation of the trilateration method is

[0062] In one embodiment, using the weighted centroid algorithm, combining the position information and relative distances (or signal strengths) of multiple base stations, the preliminary position is further optimized to obtain the estimated position. The specific steps are as follows:

[0063] For each base station, calculate a weighted factor according to its position coordinates and the relative distance (or signal strength) from the tag;

[0064] Perform a weighted average of the weighted factors of all base stations and their position coordinates to obtain the optimized tag position. The weighted centroid formula is expressed as where (x', y') are the coordinates of the optimized tag position, (x i , y i ) are the coordinates of the i-th base station, and w i is the weighted factor of the i-th base station. The weighted factor can be calculated based on the relative distance or signal strength.

[0065] In one embodiment, the precise positioning and error correction unit uses the position obtained by the preliminary positioning calculation unit as the initial position estimate for the Taylor series expansion method. Using the Taylor series expansion method, it gradually approaches the true position through iterative approximation to achieve positioning. The specific steps are as follows:

[0066] Step 1: Set the initial position estimate value (x0, y0), which is obtained by the preliminary positioning calculation unit;

[0067] Step 2: According to the UWB signal transmission time and the base station position information, construct a system of nonlinear equations to represent the distance relationship between the true position of the tag and the base stations;

[0068] Step 3: Perform a Taylor series expansion on the system of nonlinear equations to obtain a linearized system of equations;

[0069] Step 4: Solve the linearized system of equations to obtain the position correction amounts Δx and Δy;

[0070] Step 5: Update the position estimate value: x1 = x0 + Δx, y1 = y0 + Δy;

[0071] Step 6: Repeat Step 3, Step 4, and Step 5 until the iteration termination condition is satisfied (such as the position correction amount is less than a preset threshold).

[0072] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art will appreciate that various changes, modifications, substitutions and permutations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An indoor positioning system based on ultra-wideband communication technology, characterized in that: Including base station, tag and host computer; The base station interacts with the signal transmitting and receiving module, the signal processing and optimization module, and the communication module. The built-in signal transmitting and receiving module of the base station is responsible for generating and transmitting UWB pulse signals. The UWB pulse signals are received by the tags and used for ranging. The signal transmitting and receiving module of the base station is responsible for receiving response signals from the tags to complete the two-way ranging process. The base station optimizes the received signals through the signal processing and optimization module. The signal processing and optimization module establishes a machine learning model by collecting environmental characteristic data and corresponding positioning parameter data of each area of ​​the underground parking lot. The machine learning model learns and identifies the influence of different environmental characteristics on positioning accuracy, and dynamically adjusts the positioning parameters according to the environmental characteristics of the current location; The tag exchanges data with the signal transmitting and receiving module and the communication module. After receiving the signal from the base station, the signal transmitting and receiving module connected to the tag transmits a response signal. Data is exchanged between the host computer and the data processing and storage module, the monitoring and data display module and the communication module. The host computer receives data from the base station. The data processing and storage module processes, analyzes and stores the received data to extract positioning information. The monitoring and data display module displays the location information of the tag and monitors the system status. The data processing and storage module extracts positioning information including a preliminary positioning stage and a precise positioning and error correction stage. The preliminary positioning stage uses the trilateral measurement method to calculate the preliminary position of the tag according to the distance information between the tag and at least three base stations. Then, the weighted centroid algorithm is used to further optimize the preliminary position in combination with the location information and relative distances of multiple base stations to obtain an estimated position. The precise positioning and error correction stage uses the position obtained in the preliminary positioning stage as the initial position estimate of the Taylor series expansion method, and uses the Taylor series expansion method to gradually approach the true position through iterative approximation.

2. The indoor positioning system based on ultra-wideband communication technology according to claim 1, characterized in that: The signal transmitting and receiving module includes a signal transmitting unit and a signal receiving unit. The signal transmitting unit is responsible for generating and transmitting a UWB pulse signal, and the signal receiving unit is responsible for receiving a UWB pulse signal from a base station or a tag.

3. The indoor positioning system based on ultra-wideband communication technology according to claim 1, characterized in that: The signal processing and optimization module includes a data acquisition unit, a preprocessing unit, a feature extraction unit, a model unit and a parameter adjustment unit. The data acquisition unit is responsible for collecting environmental characteristic data of each area of ​​the underground parking lot, and also collects corresponding positioning parameter data. The preprocessing unit preprocesses the collected data. The feature extraction unit extracts features that have an important impact on positioning accuracy from the preprocessed data. The results of the feature extraction will be used as input to the machine learning model. The model unit trains the machine learning model based on the extracted features to learn and identify the impact of different environmental characteristics on positioning accuracy. After the model training is completed, the positioning parameters are dynamically adjusted according to the environmental characteristics of the current location. The parameter adjustment unit adjusts the parameters in the positioning algorithm according to the output results of the machine learning model.

4. The indoor positioning system based on ultra-wideband communication technology according to claim 3, characterized in that: The feature extraction unit extracts features that have an important impact on positioning accuracy from the preprocessed data, and the features include signal strength, arrival time, arrival time difference, multipath components and quantitative indicators of environmental characteristics.

5. The indoor positioning system based on ultra-wideband communication technology according to claim 3, characterized in that: The model unit trains a machine learning model based on the extracted features to learn and identify the impact of different environmental characteristics on positioning accuracy. After the model training is completed, the positioning parameters are dynamically adjusted according to the environmental characteristics of the current location. The specific steps are as follows: Data preparation and feature extraction: prepare a training data set, which includes the raw data collected by the base station and the tag, and the corresponding positioning results obtained by geometric algorithms or other methods. Then, use the feature extraction unit to extract useful features from these raw data, which will serve as the input of the random forest model. Random Forest Model Training: Random sampling: randomly extract multiple subsets from the training data set, each subset contains a part of the samples of the original data set; Constructing decision trees: For each tree in the random forest, a training set of the same size as the original data set is extracted from the training data set using the Bootstrap method. At each split point, a portion of features is randomly selected for inspection, and the optimal feature is selected to split the node based on the Gini impurity. The above process is repeated until the stopping condition is met. The Gini impurity is expressed as Where c is the number of categories, p i is the probability of the i-th category. The value of Gini impurity is between 0 and 1. The smaller the value, the purer the data set. Integrated prediction: For new input samples, they are input into all decision trees for prediction, and the prediction results of all trees are combined to obtain the final prediction result; Model evaluation and optimization: After training, the random forest model is evaluated to check its performance. The training dataset is divided into multiple parts, and one part is used as the test set and the rest as the training set for model training and testing. Based on the evaluation results, the model is optimized. Online positioning and parameter adjustment: When a new UWB signal is received, the feature extraction unit is used to extract the signal features, which are then input into the random forest model for prediction. The model dynamically adjusts the positioning parameters according to the environmental characteristics of the current location.

6. The indoor positioning system based on ultra-wideband communication technology according to claim 1, characterized in that: The data processing and storage module includes a data receiving unit, a preprocessing unit, a preliminary positioning calculation unit, a precise positioning and error correction unit, a data storage unit and a data interaction interface unit. The data receiving unit is responsible for receiving data from base stations and tags. The preprocessing unit preprocesses the received data. The preliminary positioning calculation unit uses trilateral measurement to calculate the preliminary position of the tag according to the distance information between the tag and at least three base stations. Subsequently, the weighted centroid algorithm is used to further optimize the preliminary position in combination with the position information and relative distances of multiple base stations to obtain an estimated position. The precise positioning and error correction unit uses the position obtained by the preliminary positioning calculation unit as the initial position estimate of the Taylor series expansion method, and uses the Taylor series expansion method to gradually approach the true position through iterative approximation to achieve positioning. The positioning error is corrected by introducing a neural network. The data storage unit is responsible for storing the processed data. The data interaction interface unit provides a data interaction interface with other modules.

7. The indoor positioning system based on ultra-wideband communication technology according to claim 5, characterized in that: The preliminary positioning calculation unit calculates the preliminary position of the tag using the trilateral measurement method according to the distance information between the tag and at least three base stations. The specific steps are: Assume that there are three base stations A, B, and C, and their location coordinates are (x A ,y A )、(x B ,y B )、(x C ,y C ); The distances between the tag and base stations A, B, and C are d A d B d C , the distance is calculated by the transmission time of the UWB signal and the signal propagation speed, and the distance formula is expressed as d=c×TOA, where d is the distance between the tag and the base station, c is the propagation speed of the UWB signal in the air, and TOA is the transmission time of the signal from the base station to the tag; With base station A as the center, d A Draw a circle with radius d B Draw a circle with base station C as the center, d C Draw a circle with a radius of . The intersection of these three circles is the initial position of the tag. The geometric equation of the trilateral measurement method is:

8. The indoor positioning system based on ultra-wideband communication technology according to claim 5, characterized in that: Using the weighted centroid algorithm, combined with the location information and relative distances of multiple base stations, the preliminary location is further optimized to obtain the estimated location. The specific steps are: For each base station, a weighting factor is calculated based on its location coordinates and relative distance to the tag; The weighted average of the weighted factors of all base stations and their position coordinates is used to obtain the optimized tag position. The weighted centroid formula is expressed as Among them, (x', y') is the optimized label position coordinates, (x i ,y i ) is the location coordinate of the i-th base station, w i is the weighting factor of the ith base station.

9. The indoor positioning system based on ultra-wideband communication technology according to claim 5, characterized in that: The precise positioning and error correction unit uses the position obtained by the preliminary positioning calculation unit as the initial position estimation of the Taylor series expansion method, and uses the Taylor series expansion method to gradually approach the real position through iterative approximation to achieve positioning. The specific steps are: Step 1: Set the initial position estimate (x0, y0), which is obtained by the preliminary positioning calculation unit; Step 2: Based on the UWB signal transmission time and base station location information, a nonlinear equation group is constructed to represent the distance relationship between the actual location of the tag and the base station; Step 3: Perform Taylor series expansion on the nonlinear equations to obtain a linearized equation system; Step 4: Solve the linearized equations to obtain position corrections Δx and Δy; Step 5: Update the estimated position: x1=x0+Δx, y1=y0+Δy; Step 6: Repeat steps 3, 4, and 5 until the iteration termination condition is met.

Images