Long-distance tunnel in-hole man-machine real-time positioning method based on UWB

Abstract

The invention relates to a UWB-based man-machine real-time positioning method in a long-distance tunnel hole. The method comprises the following steps: S1, collecting original ranging information of integrated signal characteristics; s2, correcting an original distance based on signal quality; s3, carrying out NLOS signal identification based on SVM (Support Vector Machine); s4, NLOS ranging error elimination based on an error distribution fitting model; s5, preliminarily resolving a tag position stake number; s6, label position correction based on Kalman filtering; and S7, determining a man-machine position based on label matching. According to the invention, the UWB positioning system is used for positioning and tracking personnel machinery in a hole, so that real-time transmission and analysis of man-machine position pile number information in the hole are realized, and management personnel can timely and accurately master man-machine operation positions; according to the man-machine real-time positioning method, the influence of the complex construction environment in the tunnel hole is reduced, and the man-machine positioning precision and stability in the tunnel hole are effectively improved.

Description

Technical Field

[0001] The invention belongs to the technical field of tunnel engineering construction, and in particular relates to a UWB-based real-time human-machine positioning method in a long-distance tunnel. Background Art

[0002] During the construction of long-distance tunnels, problems such as tight sealing, inconvenient transportation, and harsh environment seriously threaten the personal safety of construction workers. Once a landslide or other safety accident occurs, due to the limitations of communication conditions, it is difficult for rescue workers outside the tunnel to timely and accurately find the location information of trapped people in the tunnel, making it difficult to carry out rescue. In addition, long-distance tunnel construction requires the cooperation of multiple teams. Timely and accurate grasp of the working position of man and machine has a positive effect on managing their working status, analyzing working efficiency, and reducing the time-consuming handover of teams.

[0003] At present, the technologies commonly used for real-time positioning of indoor personnel include Radio Frequency Identification (RFID) positioning technology, infrared positioning technology, ZigBee positioning technology, Ultra Wide Band (UWB) positioning technology, etc. Among them, UWB positioning technology is widely used for positioning of tunnel construction personnel and mechanical equipment due to its unique advantages such as strong anti-interference ability, high transmission rate and high positioning accuracy.

[0004] However, in the complex environment of tunnel construction, there are often obstructions such as rock walls, large machinery, and workers, which cause the non-line of sight (NLOS) propagation of UWB signals, resulting in an increase in UWB ranging errors, seriously affecting the positioning accuracy of personnel and machinery in the tunnel.

[0005] Therefore, it is necessary to accurately identify line of sight (LOS) and non-line of sight propagation signals, and use corresponding algorithms to eliminate or reduce the identified NLOS signals. Summary of the invention

[0006] The purpose of the present invention is to overcome the shortcomings of the prior art, provide a UWB-based real-time positioning method for humans and machines in long-distance tunnels, improve the UWB tunnel positioning accuracy, and realize the real-time positioning of humans and machines in long-distance tunnels.

[0007] The present invention solves the technical problem by the following technical solutions:

[0008] A method for real-time positioning of humans and machines in a long-distance tunnel based on UWB, the method comprising the following steps:

[0009] S1, collection of original ranging information with integrated signal features;

[0010] S2, original distance correction based on signal quality;

[0011] S3, NLOS signal recognition based on SVM;

[0012] S4, NLOS ranging error removal based on error distribution fitting model;

[0013] S5, preliminary calculation of the tag position stake number;

[0014] S6, tag position correction based on Kalman filtering;

[0015] S7. Determine the human-machine position based on tag matching.

[0016] Moreover, S1 specifically includes: arranging a UWB human-machine positioning system in a long-distance tunnel, wherein the UWB human-machine positioning system includes a UWB positioning base station, a UWB positioning tag, a server and a database and an application terminal, and collecting original ranging information from the UWB positioning base station including the positioning base station identifier Mac, the human-machine positioning tag number ID, the distance Distance between the positioning base station and the positioning tag, the direction Direction of the positioning tag position relative to the base station position, the current time t, the UWB signal quality SQI and the UWB signal channel pulse response waveform CIR.

[0017] Moreover, the S2 is specifically: the larger the SQI value in the original ranging information, the better the UWB signal quality, and the more reliable the corresponding ranging result. The distance value between the positioning base station and the positioning tag is assigned different weights according to the SQI value, and the distance is comprehensively corrected, and the calculation is as follows:

[0018]

[0019] Where: d(t) is the final distance between the positioning base station and the positioning tag at the current time t;

[0020] N is the number of ranging results generated within Δt, N = fΔt, f is the positioning sampling frequency f;

[0021] w t,e is the weight of the e-th ranging result at time t;

[0022] d t,e is the Distance value corresponding to the e-th distance measurement result at time t;

[0023] SQI t,e is the SQI value corresponding to the e-th ranging result at time t.

[0024] Moreover, S3 uses SVM to classify and identify NLOS signals, specifically:

[0025] (1) Constructing a UWB positioning data sample dataset: First, collect the UWB positioning signal data of the UWB positioning tag and the UWB positioning base station at different distances in both unobstructed and obstructed situations, i.e., the corresponding LOS signal and NLOS signal; then assign different labels to the two types of UWB positioning signal data to produce a UWB positioning data sample dataset;

[0026] (2) Selecting signal identification features: In the NLOS propagation environment, the direct path between the UWB positioning base station and the UWB positioning tag is damaged due to reflection or obstacles. The signal will have different degrees of loss when passing through obstacles, resulting in a large difference in the channel impulse response waveform CIR between the NLOS and LOS propagation environments. Combined with the waveform characteristics of the NLOS propagation environment, the position difference dix, saturation S, and maximum amplitude r are extracted from the CIR waveform data. max , total energy ε4 key parameters are used as the identification and classification characteristic indicators of LOS and NLOS signals, and these characteristic indicators are normalized to have similar scales;

[0027] (3) Training the NLOS signal recognition model based on SVM: The UWB positioning data sample dataset is divided into a training set and a test set according to a ratio of 5:1. The Gaussian kernel function is selected as the kernel function. The kernel function parameters and the regularization parameter C are adjusted to optimize the objective function and find the optimal hyperplane to distinguish LOS and NLOS signals.

[0028] Moreover, the S4 is specifically: correcting the distance between the UWB positioning base station and the UWB positioning tag in the NLOS environment, and the specific calculation is as follows:

[0029]

[0030] in: is the correction value of ranging in NLOS environment; d NLOS is the measured value of the distance between the positioning base station and the positioning tag in the NLOS environment; ε NLOS is the ranging error in NLOS environment;

[0031] (1) Constructing an NLOS ranging error database: The ranging error at different distances in the NLOS environment is measured in a long-distance tunnel. The error measurement at each distance is performed multiple times to establish an NLOS ranging error database including actual distance, measured distance, ranging error, average error, maximum error, minimum error, and standard deviation.

[0032] (2) Establish an error distribution fitting model: For the ranging error data of all ranging points, error analysis is performed to observe their distribution patterns. According to the central limit theorem, the ranging error of the i-th ranging point obeys Normal distribution. According to the characteristics of normal distribution, most of the ranging errors are concentrated in Based on the ranging error of the ranging point, the error distribution fitting model is established. According to the upper and lower limits of the error of each ranging point, the fitting curves of the upper and lower limits of the error are established. When the input measurement distance matches the measurement distance of the i-th ranging point in the error database, ε is set. NLOS exist When the input measurement distance does not match the measurement point in the error database, the model performs interpolation of the previous and next data and outputs ε NLOS .

[0033] Furthermore, the S5 is specifically as follows: the stake number gradually increases from the tunnel opening to the tunnel face, and the position stake number of the positioning tag is calculated by the following formula:

[0034]

[0035] in: is the measurement position stake number of the human-machine ID at time t; Mac The installation stake number of the positioning base station is marked as Mac, which has been determined when the UWB positioning base station is installed.

[0036] Moreover, the S6 is specifically: using the Kalman filter KF to measure the position stake number The Kalman filter updates the human-machine position estimate based on the prediction step and the update step. In the prediction step, the Kalman filter uses the dynamic model of the moving target in the tunnel to predict the position of the moving personnel and machinery at the next moment; in the update step, the Kalman filter combines the actual measurement data and corrects the predicted position estimate by weighted averaging, thereby obtaining a more accurate and stable position estimate.

[0037] Furthermore, S7 specifically includes: matching the ID number of the positioning tag with the corresponding input process team and name identity information to obtain the position number x of the operator or mechanical equipment j in process i at time t. i,j (t), that is, x i,j (t) = x ID (t), where i = 1, 2, 3, 4, 5, respectively representing the five processes of drilling, charging and blasting, slag removal, erection of frames, and concrete spraying; j = 1, 2, ..., p i , p i is the number of operators and mechanical equipment in process i.

[0038] The positive effects that the present invention can produce are:

[0039] 1. The present invention uses the UWB positioning system to locate and track personnel and machinery in the tunnel, realizes real-time transmission and analysis of the position and pile number information of personnel and machines in the tunnel, and helps management personnel to grasp the working positions of personnel and machines in a timely and accurate manner.

[0040] 2. The human-machine real-time positioning method proposed in the present invention reduces the impact of the complex construction environment in the tunnel and effectively improves the human-machine positioning accuracy and stability in the tunnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a flow chart of the present invention;

[0042] Figure 2 The present invention is a flow chart of NLOS signal recognition based on SVM;

[0043] Figure 3 a) is a CIR waveform diagram of the present invention under LOS state; Figure 3 b) is a CIR waveform diagram under the NLOS state of the present invention;

[0044] Figure 4 a) is a schematic diagram of a positioning base station in a tunnel of the present invention; Figure 4 b) is a schematic diagram of a positioning label on a worker's helmet of the present invention;

[0045] Figure 5 It is the error distribution fitting model curve diagram of the present invention;

[0046] Figure 6 It is the cumulative probability distribution curve of NLOS error of the present invention;

[0047] Figure 7 It is a schematic diagram of positioning accuracy under LOS and NLOS environments of the present invention;

[0048] Figure 8 This is a display diagram of the real-time human-machine positioning based on tag matching in the present invention. DETAILED DESCRIPTION

[0049] The present invention is further described in detail below through specific examples. The following examples are only illustrative and not restrictive, and the protection scope of the present invention cannot be limited thereto.

[0050] like Figure 1 The present invention provides a real-time human-machine positioning method in a long-distance tunnel based on UWB, and its innovation lies in: the steps of the method are:

[0051] Step 1: Collecting raw ranging information with integrated signal features

[0052] (1) A UWB human-machine positioning system is deployed in a long-distance tunnel, which mainly includes several UWB positioning base stations, several UWB positioning tags, servers and databases, and application terminals.

[0053] The UWB positioning base station installed in the tunnel and the UWB positioning tag on the worker's helmet are as follows: Figure 4 As shown, a UWB positioning base station is arranged at a certain distance in the tunnel, and this embodiment adopts a 300m interval; the positioning tags are placed on the corresponding mechanical equipment and the safety helmets worn by construction workers of different teams, and the tags have the identity information of the bound personnel and equipment. The positioning tags are installed on the safety helmets of the workers and the mechanical equipment, and the positioning tags and the positioning base stations are in a line of sight as much as possible. After the positioning base station obtains the ranging information of the positioning tags, it can realize the positioning and tracking of different personnel and equipment. The signal data is transmitted to the server and database outside the tunnel through the wireless mesh self-organizing network between the base stations.

[0054] The server is arranged at the tunnel construction site command center, and the database is established in the server. They are used to receive, store and analyze UWB ranging information, positioning equipment information and application terminal transmission data, and perform real-time analysis and solution of the human and machine positions in the tunnel.

[0055] The application terminals include PC clients, mobile clients, and large screen terminals on site, which can realize the input of the name, team and other information corresponding to the location stake number of the positioning base station and the positioning tag ID, and transmit them to the server and database through the network. At the same time, the location stake number in the manhole can be monitored in real time, and the location trajectory can be queried.

[0056] (2) The original positioning data of personnel and machinery in the tunnel is collected through the UWB human-machine positioning system. The original ranging information is in the following form:

[0057] D={<Mac>, <id> , <distance> , <direction> , <timestamp> , <sqi> , <cir>}

[0058] D is the distance measurement information of the positioning base station (Mac) to the positioning tag (ID). Among them, Mac is the unique identifier of the positioning base station; ID is the number of the positioning tag worn on the human machine; Distance is the distance between the positioning base station and the positioning tag; Direction is the direction of the positioning tag relative to the base station, such as "1" means that the positioning tag is in the direction of the cave relative to the base station, and "-1" means that the positioning tag is in the direction of the cave entrance relative to the base station; Timestamp is the current measurement time t; SQI is the UWB signal quality, and CIR is the channel impulse response waveform of the UWB signal.

[0059] Step 2: Correction of original distance based on SQI information

[0060] Since the original UWB signal has noise, which leads to abnormal ranging, the multiple ranging results generated within the real-time tracking time interval Δt (if the positioning sampling frequency is f, N=fΔt ranging results are generated within Δt) are corrected to improve the credibility of the original distance.

[0061] The larger the SQI value in the original ranging information, the better the UWB signal quality is, and the corresponding ranging result is more reliable. Therefore, the Distance value is assigned different weights according to the SQI value, and the distance is comprehensively corrected. The calculation is as follows:

[0062]

[0063] Where d(t) is the final distance between the positioning base station and the positioning tag at the current time t; N is the number of ranging results generated within Δt; w t,e is the weight of the e-th distance measurement result at time t; d t,e is the Distance value corresponding to the e-th distance measurement result at time t; SQI t,e is the SQI value corresponding to the e-th ranging result at time t.

[0064] Step 3: NLOS signal recognition based on SVM, such as Figure 2 As shown;

[0065] (1) Construct a UWB positioning data sample dataset. First, collect the UWB positioning signal data of the positioning tag and the base station at different distances without obstruction and with obstruction (metal obstruction, human body obstruction, rock obstruction, etc.), namely the corresponding LOS signal and NLOS signal; then assign different labels to the two types of UWB positioning signal data, that is, label the LOS signal as 0 and the NLOS signal as 1, to produce a UWB positioning data sample dataset.

[0066] (2) Select signal identification features. In the LOS propagation environment, the first identification path is the direct path, which is also the peak path. In the NLOS propagation environment, the direct path between the positioning base station and the tag is damaged due to reflection or obstacles. The signal will have different degrees of loss when passing through obstacles, causing the first path to be severely attenuated, even below the noise threshold. Figure 3 The following is a waveform diagram of the channel impulse response (CIR) under the two propagation states of LOS and NLOS. It can be seen from the figure that in the NLOS state, the first path is not the peak path. This is because the signal encounters obstacles during the propagation process, and the signal is weakened or blocked, resulting in an increase in the signal propagation distance and a significant delay in the arrival time.

[0067] In addition, in NLOS channel propagation, signal reflection or passing through obstacles will cause signal attenuation, which will cause a decrease in the maximum amplitude and total energy. Therefore, the position difference dix, saturation S, and maximum amplitude r are extracted from the CIR waveform data. max , total energy ε4 parameters are used as the identification and classification feature indicators of LOS and NLOS signals, and these features are normalized to have similar scales. The four feature indicators are described as follows:

[0068] 1) Position difference dix

[0069] The position difference dix represents the position difference between the peak path and the first path in the CIR waveform. The expression is as follows:

[0070] dix=|p first -p peak | (2)

[0071] Among them, p first is the position of the sample index point corresponding to the first identification path in CIR; p peak is the position of the sample index point corresponding to the peak path in CIR.

[0072] 2) Saturation S

[0073] In the process of signal propagation in the NLOS environment, the first path has different degrees of attenuation. In order to describe the attenuation degree of the first path, the characteristic index saturation S is proposed, and the expression is:

[0074]

[0075] Among them, r 1max is the maximum value among the three amplitudes near the first path; r max is the maximum amplitude, that is, the amplitude corresponding to the peak path. The value of S ranges from 0 to 1. Usually, the S value in the LOS environment is larger than that in the NLOS environment, indicating that the attenuation of the first path in the LOS environment is smaller than that in the NLOS environment.

[0076] 3) Maximum amplitude r max and the total energy ε

[0077] During signal propagation in the NLOS environment, the signal passes through obstacles and the signal attenuation causes the maximum amplitude r max and the total energy ε is reduced, so the maximum amplitude r max And the total energy ε reflects to a certain extent whether the signal is affected by the NLOS environment. The maximum amplitude and total energy expressions are:

[0078] r max =max{r(t i )} (4)

[0079]

[0080] Where N is the CIR waveform r(t i ), r(t i ) is the amplitude of the ith sample. i is r(t i ) is the sampling time of the i-th sample.

[0081] (3) Training the NLOS signal recognition model based on SVM. The UWB positioning data sample dataset is divided into a training set and a test set at a ratio of 5:1. The Gaussian kernel function is selected as the kernel function. By adjusting the kernel function parameters and the regularization parameter C, the objective function is continuously optimized to find the optimal hyperplane to distinguish LOS and NLOS signals as much as possible.

[0082] (4) Evaluate the performance of the NLOS signal recognition model based on SVM. Since the model determines the LOS or NLOS category of the input signal by calculating the distance between the feature vector of the input signal and the hyperplane, the NLOS signal recognition model is evaluated using a test set, and key indicators such as precision, recall, and F1 value are selected to evaluate the performance of the model. When the key indicators of the model reach a certain value, it can be considered that the model has achieved high accuracy and prediction performance and can be used in actual engineering.

[0083] The NLOS signal recognition model based on SVM runs on the GTX1080ti graphics card, and the Python machine learning library scikit-learn is used to train and test the SVM model. After parameter debugging, the classification effect is best when the "rbf" kernel, C=5 and gamma=1 are used, with a recognition accuracy of 91.40%, of which the recognition accuracy of LOS and NLOS signals reaches 91.99% and 90.83%, respectively, the recall rates are 90.70% and 92.10%, and the F1 values ​​are 0.913 and 0.915, respectively.

[0084] Step 4: NLOS ranging error elimination based on error distribution fitting model

[0085] In the NLOS environment, due to the multipath effect in the signal propagation process or the increase of non-direct signal paths and signal strength attenuation caused by obstacles, the positioning measurement distance of the ranging point will be greater than the actual distance.

[0086] Therefore, the distance measurement between the positioning base station and the positioning tag in the NLOS environment is corrected. The specific calculation is as follows:

[0087]

[0088] in, is the correction value of ranging in NLOS environment; d NLOS is the measured value of the distance between the positioning base station and the positioning tag in the NLOS environment; ε NLOS is the ranging error in the NLOS environment.

[0089] (1) Constructing an NLOS ranging error database. Covering the positioning tag with a metal plate simulates the NLOS environment. The ranging error at different distances in the NLOS environment is measured in a long-distance tunnel. The error measurement at each distance is performed multiple times to establish an NLOS ranging error database including actual distance, measured distance, ranging error, average error, maximum error, minimum error, and standard deviation.

[0090] (2) Establish an error distribution fitting model. For the ranging error data of all ranging points, error analysis is performed to observe their distribution patterns. According to the central limit theorem, the ranging error of the i-th ranging point obeys Normal distribution. According to the characteristics of normal distribution, most of the ranging errors are concentrated in Based on the ranging error of the ranging point, an error distribution fitting model is established, and according to the upper and lower limits of the error of each ranging point, the fitting curves of the upper and lower limits of the error are established, such as Figure 5 When the input measurement distance matches the measurement distance of the i-th distance measurement point in the error database, set ε NLOS exist When the input measurement distance does not match the measurement point in the error database, the model performs interpolation of the previous and next data and outputs ε NLOS .

[0091] (3) Evaluate the effect of eliminating NLOS ranging errors based on the error distribution fitting model. According to formula (6), the positioning error in the NLOS environment is eliminated. The cumulative probability distribution curve before and after the error elimination is as follows: Figure 6 As shown in the figure, after the NLOS error is eliminated, the proportion of absolute ranging error within 40 cm increases from 45.42% to 94.94%, and the absolute ranging error is significantly reduced, which verifies the effectiveness of the proposed NLOS error elimination method.

[0092] Step 5: Preliminary calculation of the tag position stake number

[0093] The stake numbers gradually increase from the tunnel entrance to the tunnel face. The position stake number of the positioning tag is calculated by the following formula:

[0094]

[0095] in, is the measurement position stake number of the human-machine ID at time t; Mac To locate the installation stake number of the base station identified as Mac, it has been determined when the base station was installed.

[0096] Step 6: Tag position correction based on Kalman filtering

[0097] In order to further reduce the impact of NLOS error and multipath propagation and improve the positioning accuracy of tags, it is proposed to use Kalman Filter (KF) to measure the position stake number. The Kalman filter principle is to use observation data and prediction models to obtain more accurate state estimates through recursive calculations.

[0098] Assuming that the man and machine in the tunnel move at a uniform speed within Δt time, the Kalman filter state equation used for state prediction is:

[0099] u t+1 =Fu t +q t (8)

[0100] The target observation equation is:

[0101] z t =Hu t +r t (9)

[0102] Among them, u t is the state variable at time t, u t =[x ID (t),v ID (t)] T , x ID (t) is the corrected position number of the human-machine ID at time t, v ID (t) is the speed of the human-machine ID at time t, z t is the observed variable at time t, i.e. the measurement position stake number F and H are the state transfer matrix and observation matrix respectively, H = [1 0]; q t and r t They are the process noise and observation noise at time t, which can be adjusted in real time according to the tunnel environment and system positioning performance. The covariance matrices corresponding to the process noise and observation noise at time t are Q t and R t .

[0103] The iterative process of Kalman filtering is as follows:

[0104]

[0105]

[0106]

[0107]

[0108]

[0109] in, is the predicted state variable at time t; and P t are the prediction error covariance matrix and error covariance matrix at time t respectively; K t is the Kalman gain at time t.

[0110] In order to discard the measured values ​​with excessive deviation during the iteration process, Equation 13 is modified as follows:

[0111]

[0112] Among them, E z is the threshold value, which is determined according to the measurement error.

[0113] 15 ranging reference points were selected in the tunnel to analyze the UWB positioning accuracy. The distance between the positioning tag and the positioning base station was within the range of 50m to 300m. The positioning tags were placed at the ranging reference point positions for 1 minute in LOS and NLOS environments respectively, and positioning data was continuously collected. The NLOS environment was simulated by covering the positioning tag with a metal plate. The positioning tag position was solved by the method of the present invention, and the root mean square error RMSE of each reference point in the LOS and NLOS environments was calculated. The calculation results are shown in the figure. Figure 7 As shown in the figure, in the LOS environment, the maximum RMSE is 0.57m, the minimum is 0.30m, and the average is 0.38m; in the NLOS environment, the maximum RMSE is 0.92m, the minimum is 0.58m, and the average is 0.75m. Under both working conditions, the positioning accuracy can meet the project's positioning requirements for construction personnel and equipment.

[0114] Step 7: Determine the human-machine location based on tag matching

[0115] In order to realize the association between real-time positioning and human-machine identity, further ensure human-machine safety and the next step of analysis, the ID number of the positioning tag is matched with the corresponding input process team and name identity information to obtain the position stake number x of the operator or mechanical equipment j in process i at time t i,j (t), that is, x i,j (t) = x ID (t), where i = 1, 2, 3, 4, 5, respectively representing the five main processes of drilling, charging and blasting, slag removal, erection of frames, and concrete spraying; j = 1, 2, ..., p i , p i is the number of operators and mechanical equipment in process i.

[0116] The method of the present invention is applied to actual engineering for real-time display, and the real-time positioning display of human and machine based on tag matching is as follows: Figure 8 shown.

[0117] Although the embodiments and drawings of the present invention are disclosed for illustrative purposes, those skilled in the art will appreciate that various substitutions, changes and modifications are possible without departing from the spirit and scope of the present invention and the appended claims. Therefore, the scope of the present invention is not limited to the contents disclosed in the embodiments and drawings.< / cir> < / sqi> < / timestamp> < / direction> < / distance> < / id>

Claims

1. A UWB-based real-time positioning method for humans and machines in long-distance tunnels, characterized by: The steps of the method are: S1, collection of original ranging information with integrated signal characteristics; S2, original distance correction based on signal quality; S3, NLOS signal recognition based on SVM; S4, NLOS ranging error removal based on error distribution fitting model; S5, preliminary calculation of the tag position stake number; S6, tag position correction based on Kalman filtering; S7. Determine the human-machine position based on tag matching.

2. The UWB-based real-time positioning method for humans and machines in long-distance tunnels according to claim 1 is characterized by: Specifically, S1 is as follows: a UWB human-machine positioning system is arranged in a long-distance tunnel, wherein the UWB human-machine positioning system includes a UWB positioning base station, a UWB positioning tag, a server and a database, and an application terminal, and the original ranging information including the positioning base station identifier Mac, the human-machine positioning tag number ID, the distance Distance between the positioning base station and the positioning tag, the direction Direction of the positioning tag relative to the base station position, the current time t, the UWB signal quality SQI, and the UWB signal channel pulse response waveform CIR is collected from the UWB positioning base station.

3. The UWB-based real-time positioning method for humans and machines in long-distance tunnels according to claim 2 is characterized in that: Specifically, S2 is: the larger the SQI value in the original ranging information, the better the UWB signal quality, and the more reliable the corresponding ranging result. The distance value between the positioning base station and the positioning tag is assigned different weights according to the SQI value, and the distance is comprehensively corrected, which is calculated as follows: Where: d(t) is the final distance between the positioning base station and the positioning tag at the current time t; N is the number of ranging results generated within Δt, N = fΔt, f is the positioning sampling frequency f; w t,e is the weight of the e-th ranging result at time t; d t,e is the Distance value corresponding to the e-th distance measurement result at time t; SQI t,e is the SQI value corresponding to the e-th ranging result at time t.

4. The UWB-based real-time positioning method for humans and machines in long-distance tunnels according to claim 1 is characterized in that: S3 uses SVM to classify and identify NLOS signals, specifically: (1) Constructing a UWB positioning data sample dataset: First, collect the UWB positioning signal data of the UWB positioning tag and the UWB positioning base station at different distances in both unobstructed and obstructed situations, i.e., the corresponding LOS signal and NLOS signal; then assign different labels to the two types of UWB positioning signal data to produce a UWB positioning data sample dataset; (2) Selecting signal identification features: In the NLOS propagation environment, the direct path between the UWB positioning base station and the UWB positioning tag is damaged due to reflection or obstacles. The signal will have different degrees of loss when passing through obstacles, resulting in a large difference in the channel impulse response waveform CIR between the NLOS and LOS propagation environments. Combined with the waveform characteristics of the NLOS propagation environment, the position difference dix, saturation S, and maximum amplitude r are extracted from the CIR waveform data. max , total energy ε4 key parameters are used as the identification and classification characteristic indicators of LOS and NLOS signals, and these characteristic indicators are normalized to have similar scales; (3) Training the NLOS signal recognition model based on SVM: The UWB positioning data sample dataset is divided into a training set and a test set according to a ratio of 5:

1. The Gaussian kernel function is selected as the kernel function. The kernel function parameters and the regularization parameter C are adjusted to optimize the objective function and find the optimal hyperplane to distinguish LOS and NLOS signals.

5. The UWB-based real-time positioning method for humans and machines in long-distance tunnels according to claim 1 is characterized by: The S4 is specifically: correcting the distance between the UWB positioning base station and the UWB positioning tag in the NLOS environment. The specific calculation is as follows: in: is the correction value of ranging in NLOS environment; d NLOS is the measured value of the distance between the positioning base station and the positioning tag in the NLOS environment; ε NLOS is the ranging error in NLOS environment; (1) Constructing an NLOS ranging error database: The ranging error at different distances in the NLOS environment is measured in a long-distance tunnel. The error measurement at each distance is performed multiple times to establish an NLOS ranging error database including actual distance, measured distance, ranging error, average error, maximum error, minimum error, and standard deviation. (2) Establish an error distribution fitting model: For the ranging error data of all ranging points, error analysis is performed to observe their distribution patterns. According to the central limit theorem, the ranging error of the i-th ranging point obeys Normal distribution. According to the characteristics of normal distribution, most of the ranging errors are concentrated in Based on the ranging error of the ranging point, the error distribution fitting model is established. According to the upper and lower limits of the error of each ranging point, the fitting curves of the upper and lower limits of the error are established. When the input measurement distance matches the measurement distance of the i-th ranging point in the error database, ε is set. NLOS exist When the input measurement distance does not match the measurement point in the error database, the model performs interpolation of the previous and next data and outputs ε NLOS .

6. The UWB-based real-time positioning method for humans and machines in long-distance tunnels according to claim 1 is characterized by: The S5 is specifically as follows: the pile number gradually increases from the tunnel entrance to the tunnel face, and the position pile number of the positioning tag is calculated by the following formula: in: is the measurement position stake number of the human-machine ID at time t; x Mac The installation stake number of the positioning base station is marked as Mac, which has been determined when the UWB positioning base station is installed.

7. The UWB-based real-time positioning method for humans and machines in long-distance tunnels according to claim 6 is characterized by: The S6 is specifically: using Kalman filter KF to measure the position stake number The Kalman filter updates the human-machine position estimate based on the prediction step and the update step. In the prediction step, the Kalman filter uses the dynamic model of the moving target in the tunnel to predict the position of the moving personnel and machinery at the next moment; in the update step, the Kalman filter combines the actual measurement data and corrects the predicted position estimate by weighted averaging, thereby obtaining a more accurate and stable position estimate.

8. The UWB-based real-time positioning method for humans and machines in long-distance tunnels according to claim 2 is characterized by: S7 specifically includes: matching the ID number of the positioning tag with the corresponding input process team and name identity information to obtain the position number x of the operator or mechanical equipment j in process i at time t. i,j (t), that is, x i,j (t) = x ID (t), where i = 1, 2, 3, 4, 5, respectively representing the five processes of drilling, charging and blasting, slag removal, erection of frames, and concrete spraying; j = 1, 2, ..., p i , p i is the number of operators and mechanical equipment in process i.

Images