A three-dimensional positioning method and system suitable for indoor substation

Abstract

The present application relates to the field of indoor positioning, and provides a three-dimensional positioning method and system suitable for indoor substation. The method comprises matching corresponding the distance measurement data between the acquired mobile target and the base station and the height data of the indoor substation according to the observation data time to obtain the matched label position data, thereby obtaining single-point multi-group label position data; based on the single-point multi-group label position data, a clustering algorithm is used to identify different target objects; based on the single-point multi-group label position data corresponding to the target objects, a trained double-channel parallel convolution LSTM neural network is used to obtain the three-dimensional position information of the mobile target.

Description

Technical Field

[0001] This invention relates to the field of indoor positioning, and more particularly to a three-dimensional positioning method and system suitable for use indoors in substations. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Indoor positioning technology refers to the accurate measurement and identification of information such as the location and movement trajectory of people and objects in an indoor environment through various technical means. In practical applications, indoor positioning technology can provide people with various convenient services, such as location-based services, indoor navigation, and indoor security.

[0004] Currently, mainstream indoor positioning technologies include wireless signal strength fingerprinting, inertial navigation, and visual recognition. However, wireless signals are easily affected by environmental factors such as walls, windows, floors, and interference, which can reduce signal quality and stability. Furthermore, wireless signals are easily intercepted or interfered with, potentially leading to security issues. Inertial navigation systems accumulate errors over time, which can reduce navigation accuracy. Visual recognition systems are also affected by lighting conditions; for example, in darkness or direct sunlight, the system's recognition accuracy may decrease. Therefore, existing technologies cannot fully meet the requirements of high accuracy, low cost, and high reliability for indoor positioning. Summary of the Invention

[0005] To address the technical problems mentioned above, this invention provides a three-dimensional positioning method and system suitable for indoor substations, which has the advantages of high precision, high reliability, low cost, and low power consumption.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] The first aspect of the present invention provides a three-dimensional positioning method suitable for use indoors in substations.

[0008] A three-dimensional positioning method suitable for indoor substations includes:

[0009] The distance measurement data between the acquired mobile target and the base station and the indoor height data are matched and matched according to the observation data time to obtain the matched tag location data, thereby obtaining multiple sets of tag location data for a single point;

[0010] Based on single-point multi-group label location data, a clustering algorithm is used to identify different target objects;

[0011] Based on the single-point multi-set label location data corresponding to the target object, a trained dual-channel parallel convolutional LSTM neural network is used to obtain the three-dimensional position information of the moving target.

[0012] The dual-channel parallel convolutional LSTM neural network is used to extract features from distance measurement data and height data using two channels respectively, and then fuses the features from distance measurement data and height data to obtain fused features. The fused features are then passed through two layers of LSTM neural network and through a fully connected layer to obtain the three-dimensional position information of the moving target.

[0013] Furthermore, the process of acquiring distance measurement data between the mobile target and the base station includes: setting up a UWB signal transmitter at each base station location in the indoor environment, receiving signals transmitted by multiple UWB signal transmitters, and calculating the distance between the mobile target and each base station based on the time delay and phase information of the signals.

[0014] Furthermore, the process of acquiring the indoor height data includes: installing barometers on both the base station and the mobile target to measure the indoor height.

[0015] Furthermore, the process of using the clustering algorithm includes: setting parameters for each data point in the single-point multi-group label location data that corresponds to an anchor point or the distance information between target objects, and using the DBSCAN algorithm to obtain clusters of different target objects.

[0016] Furthermore, the process of using the DBSCAN algorithm to obtain clusters of different target objects includes: starting from any data point, exploring other data points within its Eps neighborhood; if the number of data points within the Eps neighborhood is greater than or equal to MinPts, then the data point is marked as a core point, and all points within its Eps neighborhood are marked as density reachable; for non-core points within the Eps neighborhood of a core point, if a path is constructed through a series of core points to connect to a core point, then these non-core points will be marked as density reachable; through density reachability relationships, all core points and density reachable points belonging to the same cluster are connected to form a cluster.

[0017] Furthermore, the process of setting the parameters includes: randomly selecting any point and calculating the distance between that point and all other points; sorting the distances in ascending order to obtain an array; and calculating the maximum difference s based on the array. max And record its index as k; based on the recorded s max The value of k corresponds to the distance d. k Repeat the above process to calculate d for other points. kObtain the array and take the mode as Eps; set MinPts, and adjust MinPts and Eps during the clustering process until the optimal parameters are selected that produce the maximum number of clusters and the minimum number of noise points.

[0018] A second aspect of the present invention provides a three-dimensional positioning system suitable for use indoors in substations.

[0019] A three-dimensional positioning system suitable for indoor substations includes:

[0020] The data acquisition and fusion module is configured to match the acquired distance measurement data between the mobile target and the base station and the indoor height data according to the observation data time to obtain the matched tag location data, thereby obtaining multiple sets of tag location data for a single point;

[0021] The clustering module is configured to identify different target objects based on single-point, multi-group label location data using a clustering algorithm.

[0022] The localization module is configured to: obtain the three-dimensional position information of the moving target by using a trained dual-channel parallel convolutional LSTM neural network based on the single-point multi-set label position data corresponding to the target object;

[0023] The dual-channel parallel convolutional LSTM neural network is used to extract features from distance measurement data and height data using two channels respectively, and then fuses the features from distance measurement data and height data to obtain fused features. The fused features are then passed through two layers of LSTM neural network and through a fully connected layer to obtain the three-dimensional position information of the moving target.

[0024] A third aspect of the present invention provides a three-dimensional positioning system suitable for use indoors in substations.

[0025] A three-dimensional positioning system suitable for indoor substations includes: a UWB signal transmitter installed on each base station, a moving target, a barometer installed on the base station and the moving target, and a computer; the UWB signal transmitter, the moving target, and the barometer are all connected to the computer;

[0026] The UWB signal transmitter is used to transmit signals;

[0027] The moving target is used to receive signals transmitted by the UWB signal transmitter and upload them to a computer;

[0028] The barometer is used to measure the indoor height and upload the data to a computer.

[0029] The computer is used to match the received distance measurement data between the mobile target and the base station and the indoor height data according to the observation data time to obtain the matched tag location data, thereby obtaining single-point multi-group tag location data; based on the single-point multi-group tag location data, a clustering algorithm is used to identify different target objects; based on the single-point multi-group tag location data corresponding to the target objects, a trained dual-channel parallel convolutional LSTM neural network is used to obtain the three-dimensional position information of the mobile target; wherein, the dual-channel parallel convolutional LSTM neural network is used to extract the features of the distance measurement data and the features of the height data using two channels respectively, and fuse the features of the distance measurement data and the features of the height data to obtain the fused features, and pass the fused features through two layers of LSTM neural network, and after passing through a fully connected layer, obtain the three-dimensional position information of the mobile target.

[0030] A fourth aspect of the present invention provides a computer-readable storage medium.

[0031] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the three-dimensional positioning method applicable to a substation interior as described in the first aspect above.

[0032] A fifth aspect of the present invention provides a computer device.

[0033] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the steps of the three-dimensional positioning method for use indoors in a substation as described in the first aspect above.

[0034] Compared with the prior art, the beneficial effects of the present invention are:

[0035] Compared to traditional methods that rely solely on a combination of UWB and DBA, this invention employs a combined LSTM neural network algorithm, offering superior adaptability and flexibility. Specifically, the mobile device communicates with multiple reference nodes via UWB to acquire distance information, then uses DBA to obtain altitude information. Finally, the LSTM neural network learns from and predicts based on historical location data, improving the accuracy and stability of location estimation. An optimization algorithm is then used to adjust the weights of the UWB, DBA, and LSTM neural networks to further enhance positioning accuracy and robustness. Therefore, this invention's combined UWB, DBA, and LSTM neural network algorithm effectively improves positioning accuracy and reliability.

[0036] This invention employs a combined algorithm of UWB, DBA, and LSTM neural networks to effectively improve the accuracy and reliability of 3D positioning suitable for use indoors in substations. UWB technology solves the location acquisition problem, the DBA algorithm addresses multipath interference, and the LSTM neural network solves the location prediction problem. By combining these three technologies, their respective advantages can be fully utilized to improve the accuracy and robustness of 3D positioning suitable for use indoors in substations.

[0037] This invention designs a dual-channel parallel convolutional LSTM neural network, with one layer taking two-dimensional coordinate information as input and the other taking height coordinate information as input. Combining the two-dimensional and height coordinate information allows for the extraction of the target's position information in three-dimensional space. The dual-channel LSTM neural network design reduces computational load and improves computational efficiency. It lowers computational complexity, reduces the computational burden, and makes neural network training and inference faster and more efficient. Attached Figure Description

[0038] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0039] Figure 1 This is a flowchart illustrating a three-dimensional positioning method applicable to indoor substations, as shown in this invention.

[0040] Figure 2 This is a structural diagram of the dual-channel CNN-LSTM model shown in this invention;

[0041] Figure 3 This is a framework diagram of a three-dimensional positioning system suitable for use indoors in substations, as shown in this invention. Detailed Implementation

[0042] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0043] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0044] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0045] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and systems according to various embodiments of this disclosure. It should be noted that each block in a flowchart or block diagram may represent a module, segment, or portion of code, which may include one or more executable instructions for implementing the logical functions specified in the various embodiments. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutively represented blocks may actually be executed substantially in parallel, or they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, may be implemented using a dedicated hardware-based system that performs the specified functions or operations, or using a combination of dedicated hardware and computer instructions.

[0046] Example 1

[0047] Terminology Explanation:

[0048] Ultra-wideband (UWB) technology is a short-range wireless communication technology characterized by high precision, high speed, and low power consumption. UWB technology achieves high-precision distance measurement and location positioning by transmitting short pulse signals and modulating and demodulating the signals using various parameters such as time, frequency, and amplitude. In three-dimensional positioning applications suitable for substations, UWB technology can be used for precise location and tracking of objects or personnel.

[0049] Low-cost differential barometer altimeter (DBA) technology obtains relatively high elevation accuracy (better than 0.3m indoors and better than 1m outdoors), and is not limited by the environment, can be used both indoors and outdoors, and can effectively distinguish indoor floors and outdoor auxiliary constraints GNSS to improve positioning accuracy.

[0050] LSTM (Long Short-Term Memory) is a special type of recurrent neural network that features memory units and gating mechanisms. LSTM networks can efficiently process long sequences of data and are widely used in fields such as natural language processing, speech recognition, and time series prediction.

[0051] The EarlyStopping function is a callback function used to stop training early, typically during model training. Its main purpose is to prevent overfitting, achieving good performance on the training set while ensuring decent performance on the validation set.

[0052] This embodiment provides a three-dimensional positioning method suitable for indoor substations, which includes the following steps:

[0053] The distance measurement data between the acquired mobile target and the base station and the indoor height data are matched and matched according to the observation data time to obtain the matched tag location data, thereby obtaining multiple sets of tag location data for a single point;

[0054] Based on single-point multi-group label location data, a clustering algorithm is used to identify different target objects;

[0055] Based on the single-point multi-set label location data corresponding to the target object, a trained dual-channel parallel convolutional LSTM neural network is used to obtain the three-dimensional position information of the moving target.

[0056] The dual-channel parallel convolutional LSTM neural network is used to extract features from distance measurement data and height data using two channels respectively, and then fuses the features from distance measurement data and height data to obtain fused features. The fused features are then passed through two layers of LSTM neural network and through a fully connected layer to obtain the three-dimensional position information of the moving target.

[0057] In practical applications, multiple UWB signal transmitters can be set up in an indoor environment. By placing UWB signal transmitters in different locations, indoor coverage can be achieved. A moving target receives signals transmitted by multiple UWB signal transmitters and calculates the distance to each transmitter based on the signal's time delay and phase information. Then, multiple sets of distance data are input into a computer, and a combined algorithm is used to calculate the moving target's location information, which will be described below.

[0058] In the embodiments of this invention, UWB technology employs a multipath signal transmission model. By preprocessing and filtering the signal, multipath interference and noise can be removed, resulting in a clean UWB signal. Simultaneously, by utilizing the time delay and phase information of the UWB signal, the distance between the moving target and the UWB signal transmitter can be calculated.

[0059] Using low-cost differential barometer altimeter (DBA) technology, which is not limited by the environment and can be used both indoors and outdoors, the relative elevation obtained is highly accurate. Low-cost barometers are integrated on both the UWB main base station and the mobile tag. The elevation of the UWB main base station is used as the elevation benchmark of the DBA model. The three-dimensional positioning combining UWB and DBA algorithms is shown in the following formula, which eliminates multipath interference and signal noise and improves positioning accuracy.

[0060]

[0061]

[0062] In the formula, (X1, Y1, h1) represents the UWB main base station B. S1 Given the coordinates, from the base station plane coordinates B Si(Xi，Yi，hi) i = 2, ..., N, and the label position is T ag(x，y，h) R i,1 P1 is the TDOA measurement value, and B is the value of TDOA. S1 The reading from the pressure module, P is T. ag At the air pressure module reading, t m For B S1 and T ag The average Celsius temperature between, i.e., t m =(t1+t) R ) / 2, t1 is B S1 Temperature, T R For T ag Temperature at the location.

[0063] TDOA (Time of Arrival / Delay) measurement refers to the positional relationship between two receivers calculated by measuring the time difference of arrival of a signal. TDOA measurements are commonly used in wireless signal positioning technology to determine the position of a target object in a two-dimensional plane or three-dimensional space. In TDOA measurement, each receiver first measures the absolute time of signal arrival (TOA), and then calculates the time difference of arrival between the two receivers to obtain the TDOA value. Based on the non-linear relationship between the object's position and the receiver's position, the target object's position can be calculated using the TDOA measurement.

[0064] The tag location Tag(x, y, h) is calculated by jointly solving the TDOA measurement values ​​of the main base station and three slave base stations and the corresponding known coordinates of the base stations.

[0065] In an embodiment of the present invention, specifically, the dual-channel parallel convolutional LSTM neural network algorithm is trained by inputting historical time-series data of multiple positioning results and actual positioning target data to obtain a positioning information model with higher accuracy by fusing multiple positioning results, thereby achieving target location positioning. The main steps of the improved LSTM algorithm are as follows: Figure 1 As shown:

[0066] Step 1: Acquire distance measurement data between the moving target and the base station, altitude data measured by the DBA, and image data acquired by the camera;

[0067] Step 2: Preprocess the time series data. The data is fused according to the observation time to obtain fused label location data. The time series trajectory data of multiple measurements of the single label location Tag(x, y, h) is used as the model feature input, and the actual target location of the label is used as the target value. Python is used to read the data, process the data type, and perform data preprocessing, including data normalization and target value standardization.

[0068] Step 3: Use the Density-Based Spatial Scan (DBSCAN) algorithm to cluster the fused distance data. By finding high-density regions in the data space, the data within these regions are grouped into clusters. The specific steps are as follows:

[0069] Input data preparation:

[0070] The fused distance data is considered as a dataset, where each data point corresponds to an anchor point or distance information between target objects.

[0071] Parameter settings:

[0072] The two main parameters of the DBSCAN algorithm are the radius (Eps) and the minimum number of data points (MinPts). The selection of these parameters depends on the data distribution and the application scenario. In this invention, appropriate parameters were selected based on the performance of the UWB sensor and experimental results. The process is as follows: calculating the distance from each data point to all other data points:

[0073] Step 1: Randomly select any point and calculate the distance between that point and all other points.

[0074] Step 2: Sort the distances in ascending order to obtain array D = [d0, d1, d2, d3...dn] etc.

[0075] Step 3: Calculate the maximum difference smax and record its index as k, as follows:

[0076]

[0077] Step 4: Based on the recorded s max The value of k corresponds to the distance d. k After this value, the distance value increases slowly, which may indicate that it is the boundary between different clusters.

[0078] Step 5: Repeat the above process to calculate d for other points. k The array Dk = [dk1, dk2, dk3...dkn] is obtained, and the mode is Eps.

[0079] Step 6: Set MinPts to 3. During the clustering process, adjust MinPts and Eps until the optimal values ​​of minPts and Eps that produce the maximum number of clusters and the minimum number of noise points are selected.

[0080] Key points and density directly:

[0081] DBSCAN starts with any data point and explores other data points within its Eps neighborhood. If the number of data points within the Eps neighborhood is greater than or equal to MinPts, then the data point is marked as a core point, and all points within its Eps neighborhood are marked as density reachable.

[0082] Density can reach:

[0083] For non-core points within the Eps neighborhood of a core point, if a path can be constructed from a series of core points to connect to a core point, then these non-core points will be marked as density-reachable.

[0084] Cluster formation:

[0085] By using density reachability relationships, all core points and density reachable points belonging to the same cluster are connected to form a cluster.

[0086] Noise control:

[0087] Data points that do not belong to any cluster will be considered noise and excluded from the localization process.

[0088] Using the DBSCAN algorithm, we can identify different target objects from distance data and classify them into different clusters. These clustering results will be used as input to an improved LSTM to help the model establish temporal relationships, fuse 3D data, and integrate camera image data. By inputting data from multiple sensors into the LSTM, we can provide a more comprehensive and accurate position estimate.

[0089] Step 4: Set the learning rate to 0.001-0.002, the number of iterations to 1000-2000, and the number of samples per pass to 20-40. Construct a dual-channel CNN-LSTM model, as follows: Figure 2 As shown, two parallel one-dimensional CNNs are used, with two convolutional layers in each channel. Different sized convolutional kernels are selected for the two channels to extract features at different scales from the original 3D localization data and the image. Multiple 3D coordinates of a single point are used as input, and the actual label location is used as output. Dropout is set to 0.3-0.4 to prevent overfitting, the activation function is tanh, and the loss function is "mean_squared_error".

[0090] Step 5: Process the data dimensions. Set the data dimensions according to the step size sliding window value, and then divide the training set, test set and validation set according to 6:2:2. Set the EarlyStopping function and train the model.

[0091] Step 6: Use the Python plugin library plot to draw the fitted graph of the predicted and actual values ​​and the graph of the loss function stabilizing. Adjust the parameters and optimize the model. Train the hyperparameters of the model using preprocessed historical data in the validation set. Use the correlation coefficient R as the evaluation index to evaluate and tune the hyperparameters. The higher the R, the better the model performance.

[0092] The formula for calculating the correlation coefficient R is:

[0093]

[0094] Where: y t y' t Let represent the true and predicted values ​​of the location at time t, respectively; N represents the number of samples in the test set; and Cov is the covariance function.

[0095] Step 7: Use a model with good prediction performance to perform 3D positioning fusion prediction suitable for substation indoors. Input the preprocessed data into the LSTM neural network for data fusion and algorithm calculation.

[0096] Step 8: Output the three-dimensional position information of the moving target.

[0097] Table 1. Performance Comparison of Different Prediction Models

[0098]

[0099]

[0100] The results in Table 1 show that by combining the DBSCAN algorithm with LSTM and performing pre-clustering of the location data, the proposed prediction model outperforms the current prediction model in terms of the correlation coefficient R error index.

[0101] The core technology of this invention is the combined algorithm of UWB, DBA, and LSTM neural networks, used to process and compute the acquired data. UWB technology measures the distance between the device and a reference point, DBA technology acquires height data, and an RGB-D camera is used to acquire image and depth information of the moving target. The combination of Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) networks offers significant advantages in processing sequential data. It can capture local features and long-term dependencies, adapt to variable-length signal sequences, and gradually extract abstract feature representations through multi-layer convolution and pooling operations, from low-level signal features to high-level features, improving the model's ability to represent complex data and enhancing its generalization ability.

[0102] The algorithm-based data processing and computation module includes functions such as data preprocessing, feature extraction, data cleaning, location calculation, and location result output. Specifically, the data preprocessing module performs preliminary processing on the collected raw data; the feature extraction module extracts features from the processed data; the data cleaning module cleans and processes the extracted features; the location calculation module calculates the device's location information based on the features; and the location result output module outputs the location results to the user interface or other applications.

[0103] Compared to traditional algorithms based solely on a combination of UWB and DBA, the combined LSTM neural network algorithm offers better adaptability and flexibility. In this algorithm, the mobile device communicates with multiple reference nodes via UWB technology to obtain distance information, then uses DBA technology to acquire altitude information, and finally utilizes an LSTM neural network to learn from and predict historical location data, improving the accuracy and stability of location estimation. Finally, an optimization algorithm is used to adjust the weights of the UWB, DBA, and LSTM neural networks to further enhance positioning accuracy and robustness. Therefore, this invention employs a combined UWB, DBA, and LSTM neural network algorithm to improve positioning accuracy and reliability.

[0104] This invention provides a 3D positioning technology suitable for indoor substations based on a combined algorithm of UWB, DBA, and LSTM neural networks. This combined algorithm effectively improves the accuracy and reliability of 3D positioning in substations. UWB technology solves the location acquisition problem, the DBA algorithm solves the multipath interference problem, and the LSTM neural network solves the location prediction problem. By combining these three technologies, their respective advantages can be fully utilized to improve the accuracy and robustness of 3D positioning in substations.

[0105] This combined algorithm can be widely applied in fields such as smart manufacturing, intelligent transportation, smart homes, and the Internet of Things to achieve real-time positioning and tracking of objects or people, improving work efficiency and safety. Simultaneously, this technology can also be used for indoor navigation, venue management, and medical monitoring, bringing greater convenience and safety to people's lives and work.

[0106] In summary, the 3D positioning technology for substation indoors provided by this invention, based on a combined algorithm of UWB, DBA, and LSTM neural networks, possesses high technical content and practicality. It can meet users' requirements for positioning accuracy, real-time performance, and stability, while also offering good economic benefits and application prospects, with broad market potential and room for promotion. In future development, this technology can be further integrated with artificial intelligence, cloud computing, and other technologies to achieve more intelligent and efficient indoor positioning services.

[0107] Example 2

[0108] This embodiment provides a three-dimensional positioning system suitable for indoor substations.

[0109] A three-dimensional positioning system suitable for indoor substations includes:

[0110] The data acquisition and fusion module is configured to match the acquired distance measurement data between the mobile target and the base station and the indoor height data according to the observation data time to obtain the matched tag location data, thereby obtaining multiple sets of tag location data for a single point;

[0111] The clustering module is configured to identify different target objects based on single-point, multi-group label location data using a clustering algorithm.

[0112] The localization module is configured to: obtain the three-dimensional position information of the moving target by using a trained dual-channel parallel convolutional LSTM neural network based on the single-point multi-set label position data corresponding to the target object;

[0113] The dual-channel parallel convolutional LSTM neural network is used to extract features from distance measurement data and height data using two channels respectively, and then fuses the features from distance measurement data and height data to obtain fused features. The fused features are then passed through two layers of LSTM neural network and through a fully connected layer to obtain the three-dimensional position information of the moving target.

[0114] It should be noted that the data acquisition and fusion module, clustering module, and positioning module described above are the same examples and application scenarios implemented in the steps of Embodiment 1, but are not limited to the content disclosed in Embodiment 1. It should also be noted that the above modules, as part of the system, can be executed in a computer system such as a set of computer-executable instructions.

[0115] Example 3

[0116] like Figure 3 As shown, this embodiment provides a three-dimensional positioning system suitable for indoor substations.

[0117] A three-dimensional positioning system suitable for indoor substations includes: a UWB signal transmitter installed on each base station, a moving target, a barometer installed on the base station and the moving target, and a computer; the UWB signal transmitter, the moving target, and the barometer are all connected to the computer;

[0118] The UWB signal transmitter is used to transmit signals;

[0119] The moving target is used to receive signals transmitted by the UWB signal transmitter and upload them to a computer;

[0120] The barometer is used to measure the indoor height and upload the data to a computer.

[0121] The computer is used to match the received distance measurement data between the mobile target and the base station and the indoor height data according to the observation data time to obtain the matched tag location data, thereby obtaining single-point multi-group tag location data; based on the single-point multi-group tag location data, a clustering algorithm is used to identify different target objects; based on the single-point multi-group tag location data corresponding to the target objects, a trained dual-channel parallel convolutional LSTM neural network is used to obtain the three-dimensional position information of the mobile target; wherein, the dual-channel parallel convolutional LSTM neural network is used to extract the features of the distance measurement data and the features of the height data using two channels respectively, and fuse the features of the distance measurement data and the features of the height data to obtain the fused features, and pass the fused features through two layers of LSTM neural network, and after passing through a fully connected layer, obtain the three-dimensional position information of the mobile target.

[0122] Hardware deployment: Deploy multiple UWB receivers and transmitters in an indoor environment, along with multiple Bluetooth base stations and WiFi routers, to utilize these signals for location positioning and tracking.

[0123] Data Acquisition: Using UWB receivers, Bluetooth, WiFi and other devices, collect multiple UWB signal information such as signal strength and distance.

[0124] Processing UWB signals: By preprocessing, filtering, and screening the acquired UWB signals, noise and outliers are removed to obtain clean UWB signals.

[0125] DBA data acquisition: Utilizing low-cost barometer differential barometric altimetry technology, more accurate altitude information is obtained.

[0126] Positioning calculation: Using methods such as triangulation and weighted averaging, the position of an object or person is calculated, and the results are verified and corrected based on the positioning results of multiple algorithms.

[0127] Data fusion: Using the LSTM neural network method with multi-channel parallel convolution, multiple localization results are fused to obtain more accurate and reliable 3D localization results.

[0128] Iterative optimization of positioning accuracy: Based on the timestamps from the tag to each base station obtained through the solution process, and combined with pre-known relevant information, using different base stations as reference points, N sets of positioning coordinates (N being the number of base stations) can be obtained. By weighting the coordinates, positioning accuracy can be further improved.

[0129] Example 4

[0130] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the three-dimensional positioning method applicable to substation rooms as described in Embodiment 1 above.

[0131] Example 5

[0132] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the three-dimensional positioning method for substation indoors as described in Embodiment 1 above.

[0133] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A three-dimensional positioning method suitable for use in a substation room, characterized by, include: The distance measurement data between the acquired mobile target and the base station and the indoor height data are matched and matched according to the observation data time to obtain the matched tag location data, thereby obtaining multiple sets of tag location data for a single point; Based on single-point multi-group label location data, a clustering algorithm is used to identify different target objects; Based on the single-point multi-set label location data corresponding to the target object, a trained dual-channel parallel convolutional LSTM neural network is used to obtain the three-dimensional position information of the moving target. The dual-channel parallel convolutional LSTM neural network is used to extract features of distance measurement data and height data respectively using two channels, and then fuses the features of distance measurement data and height data to obtain fused features. The fused features are then passed through two layers of LSTM neural network and through a fully connected layer to obtain the three-dimensional position information of the moving target. The process of using the clustering algorithm includes: setting parameters for each data point in the single-point multi-group label location data that corresponds to an anchor point or the distance information between target objects, and using the DBSCAN algorithm to obtain clusters of different target objects; using two parallel one-dimensional CNNs, each channel using two convolutional layers, with different sizes of convolutional kernels for the two channels, to extract features of different scales from the original three-dimensional positioning data and images; the clustering results will be used as input to an improved LSTM to help the model establish temporal relationships.

2. The method for three-dimensional positioning suitable for use in a substation room according to claim 1, characterized in that, The process of acquiring distance measurement data between the moving target and the base station includes: setting up a UWB signal transmitter at each base station location in the indoor environment, receiving signals transmitted by multiple UWB signal transmitters, and calculating the distance between the moving target and each base station based on the time delay and phase information of the signals.

3. The three-dimensional positioning method applicable to indoor substations according to claim 1, characterized in that, The process of acquiring the indoor height data includes: setting up barometers on both the base station and the mobile target to measure the indoor height.

4. The three-dimensional positioning method applicable to indoor substations according to claim 1, characterized in that, The process of using the DBSCAN algorithm to obtain clusters of different target objects includes: starting from any data point, exploring other data points within its Eps neighborhood; if the number of data points within the Eps neighborhood is greater than or equal to MinPts, then the data point is marked as a core point, and all points within its Eps neighborhood are marked as density reachable; for non-core points within the Eps neighborhood of a core point, if a path is constructed through a series of core points to connect to a core point, then these non-core points will be marked as density reachable; through density reachability relationships, all core points and density reachable points belonging to the same cluster are connected to form a cluster.

5. The three-dimensional positioning method applicable to indoor substations according to claim 1, characterized in that, The process of setting parameters includes: randomly selecting any point and calculating the distance between that point and all other points; sorting the distances in ascending order to obtain an array; and calculating the maximum difference based on the array. And record its index as k; based on the record k, the corresponding distance Repeat the above process to calculate other points. Obtain the array and take the mode as Eps; set MinPts, and adjust MinPts and Eps during the clustering process until the optimal parameters are selected that produce the maximum number of clusters and the minimum number of noise points.

6. A three-dimensional positioning system suitable for indoor substations, employing the three-dimensional positioning method for indoor substations as described in any one of claims 1-5, characterized in that, include: The data acquisition and fusion module is configured to match the acquired distance measurement data between the mobile target and the base station and the indoor height data according to the observation data time to obtain the matched tag location data, thereby obtaining multiple sets of tag location data for a single point; The clustering module is configured to identify different target objects based on single-point, multi-group label location data using a clustering algorithm. The localization module is configured to: obtain the three-dimensional position information of the moving target by using a trained dual-channel parallel convolutional LSTM neural network based on the single-point multi-set label position data corresponding to the target object; The dual-channel parallel convolutional LSTM neural network is used to extract features from distance measurement data and height data using two channels respectively, and then fuses the features from distance measurement data and height data to obtain fused features. The fused features are then passed through two layers of LSTM neural network and through a fully connected layer to obtain the three-dimensional position information of the moving target.

7. A three-dimensional positioning system suitable for indoor substations, employing the three-dimensional positioning method for indoor substations as described in any one of claims 1-5, characterized in that, include: Each base station is equipped with a UWB signal transmitter, a moving target, a barometer installed on the base station and the moving target, and a computer; the UWB signal transmitter, the moving target, and the barometer are all connected to the computer. The UWB signal transmitter is used to transmit signals; The moving target is used to receive signals transmitted by the UWB signal transmitter and upload them to a computer; The barometer is used to measure the indoor height and upload the data to a computer. The computer is used to match the received distance measurement data between the mobile target and the base station and the indoor height data according to the observation data time to obtain the matched tag location data, thereby obtaining single-point multi-group tag location data; based on the single-point multi-group tag location data, a clustering algorithm is used to identify different target objects; Based on the single-point multi-set label location data corresponding to the target object, a trained dual-channel parallel convolutional LSTM neural network is used to obtain the three-dimensional location information of the moving target. The dual-channel parallel convolutional LSTM neural network is used to extract the features of the distance measurement data and the features of the height data through two channels respectively, and then fuses the features of the distance measurement data and the features of the height data to obtain the fused features. The fused features are then passed through two layers of LSTM neural network and then through a fully connected layer to obtain the three-dimensional location information of the moving target.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the three-dimensional positioning method applicable to the interior of a substation as described in any one of claims 1-5.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the three-dimensional positioning method applicable to substation rooms as described in any one of claims 1-5.

Images